Three-City Survey Constructed Variables Documentation

This document provides description of the constructed variables in the data file--"3city.dta".
-----------------------------------------------------------------------------------------------------
ID		UNIQUE RESPONDENT'S ID	

		Zhen Zeng created this variable.
		This variable was obtained by multiplying the assigned questionnaire serial number by 10 plus the number of the sample form.
		
			ID=SERIALNU*10+FORM

		No missing code.
										
FORM		SAMPLE FORM		

		Codes: 1, 2, 3, and 4.
		Assigned value for the sample form.
		1: Young primary sample (age 18 to 60)
		2: Elder primary sample (age 61+)
		3: Matched parent sample
		4: Matched child sample (age 18+)
					
		No missing code.

MATCH		MATCHED PAIR STATUS	

		Codes: 0, 1, 2, and 3.
		0: uncertain
		1: matched
		2: not matched
		3: single, i.e., no paired respondent

		Zhen Zeng created this variable to indicate whether a respondent had a matched parent or offspring in the data.  See 10pair.sps and 11addmatch.sps

		The information of the matching (i.e., having a same serial number) pair's name, gender, age, and whether living together with the offspring/parent from the selected parent/offspring was compared to the corresponding information from the primary respondent to determine whether two cases that had a same serial number were really matched.
		
		MATCH=1: the pair was matched
		
		Specifically, conditions for MATCH to be coded 1 (i.e.,having a match pair) were for the two cases that had a same serial number and had one of the following conditions:
		1) names and gender matched as well as either living together matched or age matched; or
		2) gender, living together, and age matched; or
		3) gender and name of the primary young respondent's selected parent were matched with gender and name of the elderly respondent for FORM=3 (select_parent_respondent); or
		4) primary elderly respondent had single child and gender of primary elder's selected child was matched with gender of the young respondent for FORM=4 (select_offspring_respondent). 
		
		Due to typo errors, MATCH was needed to be individually coded 1 for some cases.  Serial numbers for these cases were:
		1010006 1030055 1070174 1150375 1270679 1320819 1330852 1330858 1340869 1340873 1350902 1401022 1401030 1441127 1501277 3040871 3050488 3060585 3100017 3140753 3160410 3210381 3260215 3281155 3300241 3320543 3350972 3350988 3350989 3410092 3410093 3420068 3420070 3440830 3490458 3511337 3591118 4010363 4020465 4050521 4050547 4060504 4070423 4070427 4151239 4161166 4161178 4161198 4280065 4440811 4450892 4460919 4620590 4620591 4630265 4630268 4671339 4681365 1110261 1110282 1170439 1210537 1350900 1390992 1451160 3010160 3060577 3130737 3250202 3300252 3320528 3320546 3391358 3410076 3410078 3410097 3430799 3470836 3591116 4010374 4010377 4120982 4151258 4240091 4250120 4300179 4350198 4370879 4460937 4460945 4711232 1030062 1040080 1070177 1140345 1300769 1401025 1441139 3020169 3300261 4460946 1020044 1070178 1090209 1260663 1340859 3270362 3330270 3350963 4050539 4050546 4230024 4270104 and 4390854
		
		Given two cases that had a same serial number, the conditions for name matched were defined as:
		1) if the name from the cover page for FORM=3 (select_parent_respondent) was the same as the name of the primary young respondent's selected parent; or
		2) if the name from the cover page for FORM=4 (select_offspring_respondent) was the same as the name of the primary elderly respondent's selected child; or
		3) if the signature of the respondent for FORM=3 (select_parent_respondent) was the same as the name of the primary young respondent's selected parent; or
		4) if the signature of the respondent for FORM=4 (select_offspring_respondent) was the same as the name of the primary elderly respondent's selected child.

		Furthermore, 53 cases were individually coded as name matched.  Serial numbers for these cases were: 1150382 1491263 1360922 1360923 1010007 1040099 1050113 1150384 1180453 1330841 1340864 1340883 1350890 1370940 1411056 3030938 3030950 3180131 3230034 3270363 3340420 3420069 3470847 3551093 3581175 3581192 4040482 4070393 4070409 4100456 4121010 4121030 4161182 4250132 4340211 4350195 4350197 4630272 1020037 1050108 1050124 1110284 1330836 3180139 3300268 4050525 4050546 4110288 4110291 4340203 4430827 4470627 and 4651313
		
		Given two cases that had a same serial number, the conditions for gender matched were defined as:
		1) if gender of the elderly respondent for FORM=3 (select_parent_respondent) was the same as the gender of the primary young respondent's selected parent; or
		2) if gender of the young respondent for FORM=4 (select_offspring_respondent) was the same as the gender of the primary elderly respondent's selected child. 

		Given two cases that had a same serial number, the conditions for age matched were defined as:
		1) if the selected parent of the primary young respondent was father and father's age was the same as the age of the elderly respondent for FORM=3 (select_parent_respondent); or
		2) if the selected parent of the primary young respondent was mother and mother's age was the same as the age of the elderly respondent for FORM=3 (select_parent_respondent); or
		3) if the selected child's age of the primary elderly respondent was the same as the age of the young respondent for FORM=4 (select_offspring_respondent).
		
		Father's and mother's age were calculated by subtracting father's and mother's birth years, respectively, from 1999.

		Given two cases that had a same serial number, the conditions for living together matched were defined as:
		1) if the primary young respondent's report regarding whether lived together with parents was the same as the selected parent's report regarding whether lived together with the select child; or
		2) if the primary elderly respondent's report regarding whether lived together with the select child was the same as the selected child's report regarding whether lived together with parents.

		Furthermore, 3 cases were individually coded as living together matched.  Serial numbers for these cases were: 1050111 4370883 and 3531296.


		MATCH=2: not matched

		MATCH was coded 2 for the conditions:
		1) for the cases that had a same serial number, but respondent's spouse or sibling of the original sample substituted for the interview; or 
		2) the basic information was mismatched; or
		3) respondents from the matching sample whose serial numbers did not appear in the primary sample 
		(*** NOTE: the 3rd condition probably should be coded as single or a different code.  8 cases of them. id=30101523 40103703 40913253 40913293 41102863 43402173 43807463 and 44108403 ***)
		
		MATCH=0: uncertain
		
		MATCH was individually coded 0 (uncertain) for 6 serial numbers.  This situation was occurred mainly due to unable to locate the respective questionnaires.
		Serial numbers for these cases were: 3150389 1030071 3100021 3350990 3551097 4651316
		
		MATCH=3: single, i.e., no paired respondent
		
		MATCH was coded 3 (single) for primary respondents whose serial numbers did not appear in the matching sample.

		No missing code.

SUBCAT		COMBINATION CATEGORIES OF INTERVIEWING CITY AND FORM

		Zhen Zeng created this variable.
		The variable was obtained by multiplying the assigned city code by 10 plus the number of the sample form.
		
			SUBCAT=CITY*10+FORM
					
		No missing code.

MALE		RESPONDENT'S GENDER	

		0, 1 dummy variable.  Recoded from SEX.  
		
			MALE=1		if SEX=1
			    =0		if SEX=2
		
		No missing code

AGE		RESPONDENT'S AGE	

		Measured in years.  Range: 16-95
		Year and month of interview date and birth date were used to calculate respondent's age.  Birth date was subtracted from interview date, and the quantity of the difference in year was truncated at integer.  Mid-year (6) was assigned to birth month (BM) when birth month was missing.  The mode of interview month (6) was assigned to interview month (IM) when both interview month and completion month of interview were missing.
		
			AGE=TRUNC((99+IM/12)-(BYEAR+BM/12))
		
		There was no missing code on birth year.  Thus, there was no missing code for respondent's age.

		There were 9 cases of respondents whose age were younger than 18 at the survey time while these respondents' age from cover pages were 18 or 19 years old.
					
		No missing code.
						
DAD_AGE		FATHER'S AGE IN 1999	

		Measured in years.  Range: 38-119
		Available for the young sample.
		This measure was obtained by subtracting father's birth year from 1999.  No case was imputed.
		
			DAD_AGE=1999-FYBIRTH
		
		-999: FYBIRTH blank when response expected
		-998: DK
		-888: Elderly respondents (FORM=2 or 3)
		-777: Father's age when gave birth to respondent was 3 for id=11202891 and was 4 for id=12005151. So, recode father's birth year, age, and age giving birth to respondent into error codes for these two cases
					
MOM_AGE		MOTHER'S AGE IN 1999	

		Measured in years.  Range: 39-105
		Available for the young sample.
		This measure was obtained by subtracting mother's birth year from 1999.  No case was imputed.

			MOM_AGE=1999-MYBIRTH
					
		-999: MYBIRTH blank when response expected
		-998: DK
		-888: Elderly respondent (FORM=2 or 3)
		-777: Mother's age when gave birth to respondent was 0 for id=11202891 and was 4 for id=12005151. So, recode mother's birth year, age, and age giving birth to respondent into error codes for these two cases

DADAGE_R	FATHER'S AGE AT R'S BIRTH

		Measured in years.  Range: 10-65
		Available for the young sample.
		This measure was obtained by subtracting respondent's age from father's age.

			DADAGE_R=DAD_AGE-AGE

		-99: FYBIRTH blank when response expected
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-77: Father's age when gave birth to respondent was 3 for id=11202891 and was 4 for id=12005151. So, recode father's birth year, age, and age giving birth to respondent into error codes for these two cases

MOMAGE_R	MOTHER'S AGE AT R'S BIRTH

		Measured in years.  Range: 11-55
		Available for the young sample.
		This measure was obtained by subtracting respondent's age from mother's age.

			MOMAGE_R=MOM_AGE-AGE

		-99: MYBIRTH blank when response expected
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-77: Mother's age when gave birth to respondent was 0 for id=11202891 and was 4 for id=12005151. So, recode mother's birth year, age, and age giving birth to respondent into error codes for these two cases

WED_AGE 	R'S AGE AT MARRIAGE

		Measured in years.  Range: 8-73
		Available for respondents who were married at the survey time (i.e., marital status=1).
		This measure was obtained by subtracting respondent's birth year from year at marriage.
							
			WED_AGE=WEDYEAR-BYEAR

		-99: WEDYEAR blank when response expected
		-98: DK
		-87: Respondent not married (MARITAL=2, 3, or 4)

SWED_AGE	SPOUSE'S AGE AT MARRIAGE

		Measured in years.  Range: 7-69
		Available for respondents who were married at the survey time (i.e., marital status=1).
		This measure was obtained by subtracting birth year of respondent's spouse from year at marriage.
							
			SWED_AGE=WEDYEAR-SYBIRTH

		-99: WEDYEAR or SYBIRTH blank when response expected
		-98: DK
		-87: Respondent not married (MARITAL=2, 3, or 4)

MAR_DUR 	MARITAL DURATION IN YEARS

		Measured in years.  Range: 0-76
		Available for respondents who were married at the survey time.  (i.e., marital status=1).
		This measure was obtained by subtracting year at marriage from 1999, the year of the survey.

			MAR_DUR=99-WEDYEAR

		-99: WEDYEAR blank when response expected
		-98: DK
		-87: Respondent not married (MARITAL=2, 3, or 4)

YRCITY  	CALENDAR YEAR SINCE LIVED IN CITY

		Range: 1906-1999.
		Recode from year moved to city (YMOVE).  Recode rule is:

			YRCITY=1900+BYEAR	if YMOVE=1			
			YRCITY=1900+YMOVE	if YMVOE not missing	
			YRCITY=-9999		if YMOVE missing		

		-9999: Blank when response expected

CEDHIS		EDUC HISTORY CORRECTED

		0, 1 dummy variable.  Coded 1 if R's education history (in particular, year left school) was corrected, 0 otherwise.
		
		No missing code
		
JHJC		JUNIOR HIGH TO JUNIOR COLLEGE

		0, 1 dummy variable.  Coded 1 if R directly went from a junior high school graduate to a junior college attender without attending either senior high school or technical high school, 0 otherwise.

			JHJC=1	if (a102c=1 and a103a=2 and a104a=2 and a105a=1)
			JHJC=0	otherwise

		No missing code
		
JHU		JUNIOR HIGH TO UNIVERSITY

		0, 1 dummy variable.  Coded 1 if R directly went from a junior high school graduate to a university attender without attending either senior high school, technical high school or junior college, 0 otherwise.

			JHU=1	if (a102c=1 and a103a=2 and a104a=2 and a105a=2 and a106a=1)
			JHU=0	otherwise

		No missing code

AGEAS1		AGE ATTENDED PRIMARY SCHOOL

		Measured in years.  Range: 4-36.
		This measure was obtained by subtracting the year graduated from primary school from the birth year and # of years stayed in primary school.
		
			AGEAS1=A10_1D-BYEAR-A10_1B
		
		-99: Blank when values on the year graduated from primary school and # of years stayed in primary school were expected
		-98: DK
		-87: Not attend primary school
		-86: Attended primary school, but not graduate from it
		-77: Problematic education history

AGEAS2		AGE ATTENDED JUNIOR HIGH SCHOOL

		Measured in years.  Range: 9-35.
		This measure was obtained by subtracting the year graduated from junior high school from the birth year and # of years stayed in junior high school.
		
			AGEAS2=A10_2D-BYEAR-A10_2B
		
		-99: Blank when values on the year graduated from junior high school and # of years stayed in junior high school were expected
		-98: DK
		-87: Not attend junior high school
		-86: Attended junior high school, but not graduate from it
		-77: Problematic education history

AGEAS3		AGE ATTENDED SENIOR HIGH SCHOOL

		Measured in years.  Range: 12-39.
		This measure was obtained by subtracting the year graduated from senior high school from the birth year and # of years stayed in senior high school.
		
			AGEAS3=A10_3D-BYEAR-A10_3B
		
		-99: Blank when values on the year graduated from senior high school and # of years stayed in senior high school were expected
		-98: DK
		-87: Not attend senior high school
		-86: Attended senior high school, but not graduate from it
		-77: Problematic education history

AGEAS4		AGE ATTENDED TECH HIGH SCHOOL

		Measured in years.  Range: 12-43.
		This measure was obtained by subtracting the year graduated from technical high school from the birth year and # of years stayed in technical high school.
		
			AGEAS4=A10_4D-BYEAR-A10_4B
		
		-99: Blank when values on the year graduated from technical high school and # of years stayed in technical high school were expected
		-98: DK
		-87: Not attend technical high school
		-86: Attended technical high school, but not graduate from it
		-77: Problematic education history

AGEAS5		AGE ATTENDED JUNIOR COLLEGE

		Measured in years.  Range: 14-56.
		This measure was obtained by subtracting the year graduated from junior college from the birth year and # of years stayed in junior college.
		
			AGEAS5=A10_5D-BYEAR-A10_5B
		
		-99: Blank when values on the year graduated from junior college and # of years stayed in junior college were expected
		-87: Not attend junior college
		-86: Attended junior college, but not graduate from it
		-77: Problematic education history

AGEAS6		AGE ATTENDED UNIVERSITY

		Measured in years.  Range: 15-42.
		This measure was obtained by subtracting the year graduated from university from the birth year and # of years stayed in university.
		
			AGEAS6=A10_6D-BYEAR-A10_6B
		
		-99: Blank when values on the year graduated from university and # of years stayed in university were expected
		-87: Not attend university
		-86: Attended university, but not graduate from it
		-77: Problematic education history

AGEAS7		AGE ATTENDED GRADUATE SCHOOL

		Measured in years.  Range: 21-37.
		This measure was obtained by subtracting the year graduated from graduate school from the birth year and # of years stayed in graduate school.
		
			AGEAS7=A10_7D-BYEAR-A10_7B
		
		-99: Blank when values on the year graduated from graduate school and # of years stayed in graduate school were expected
		-87: Not attend graduate school
		-86: Attended graduate school, but not graduate from it

S_T_HS  	WHETHER ATTENDING SENIOR/TECH HIGH SCHOOL

		0, 1 dummy variable.  Coded 1 for those who attended either senior high school (a10_3a=1) or vocational or technical high school (a10_4a=1).  If information for both of a10_3a and a10_4a was missing, this variable was coded -99 (missing).  

		  0: Not attended senior high school or secondary technical school
		  1: attended either senior high school or secondary technical school
		-99: Blank when response expected for both a10_3a and a10_4a
					
HED_C7  	HIGHEST EDUCATION LEVEL ATTENDED

		A 6 educational levels coded from 0 to 5.
		Information on whether attending a particular educational level from the education history (Questionnair Section A Q10) was used to construct this variable.  The information from a higher level of education has a priority over a lower level.
					
			The 6 categories were defined as:
			HED_C7=5	if A10_7A=1 OR A10_6A=1		
			HED_C7=4	if A10_5A=1		
			HED_C7=3	if A10_4A=1 or A10_3A=1	
			HED_C7=2	if A10_2A=1		
			HED_C7=1	if A10_1A=1		
			HED_C7=0	if A10_1A=2		
		
		There was No missing code on a10_1a so that there was No missing code on hed_c7.
					
		No missing code.		

SCHYR_C7	YEARS OF EDUCATION

		Measure in years.  Range: 0-19.
		The information from the education history (Questionnairs Section A Q10) was used to construct this variable.
					
		The rules to construct this variable were as follow.  
		1) The information from a higher level of education has a priority over a lower level.  In other words, education history was checked from graduate school and downward for creating this variable.
					
		2) The norm of the schooling years to complete a particular educational level was used to assign for years of education for those who graduated the particular education level.  
		
		The norm of schooling years to complete each education level is:
					Years to Complete	Years Cumulated
		graduate school:		3		6+3+3+4+3=19 
		  university:			4		6+3+3+4=16
		  junior college:		3		6+3+3+3=15
		    5-years junior college:	5		6+3+5=14
		    secondary technical school:	3		6+3+3=12
		    senior high school:		3		6+3+3=12
		      junior high school:	3		6+3=9
			primary school:		6		6
		
		The 5-year junior college was defined as those who attended junior college without attending either senior high school or secondary technical school.
		
		3) For those who attended, but not graduate from a particular education level, the number of years staying on that level was added to the norm of schooling years for the previous level.  However, if the number of years stayed was equal to or longer than the norm of schooling years to complete for that level, the norm of schooling years minus 1 was used.  If the number of years stayed was missing, the mode of years stayed for that level for those who did not graduate was used.  Imputed status for years of education completed was coded 1 for the last situation.
					
		4) Those who attended a particular education level, but whose information on whether graduated from the level was missing, were assumed not graduating from that level. The rule #3 for computing years of education was applied.
					
		The calculation of years of education for junior college was different for those who attended either senior high school or secondary technical school and those who did not.  The qualification of attending junior college was senior high school or secondary technical school graduate and the norm to complete junior college was three years.  However there was a type of junior college that only required junior high school graduate to attend and the norm to complete this type of junior college was five years.  So, we distinguished these two types of junior college attenders by whether attended either senior high school or secondary technical school.  For those junior college attenders who did not attended senior high school or secondary technical school were classified as 5-year junior college attenders and the norm to complete this level of education was five years.  For the junior college attenders who attended either senior high school or secondary technical school, the norm to complete this level of education was three years.
					
		No missing code.

MSCHYR_7	IMPUTED YEARS OF EDUCATION

		0, 1 dummy variable.  See note #3 of years of education for conditions coded 1.

		0: Years of education was not imputed
		1: Years of education was imputed
		No missing code.

YRENDED 	CALENDAR YEAR LEFT SCHOOL

		Measure in calendar years.  Range: 1920-1999.

		The information from the education history was used to construct this variable.
					
		The rules to construct this variable were as follow.  
		1) The information from a higher level of education has a priority over a lower level.  In other words, education history was checked from graduate school and downward for creating this variable.

		2) When the calendar year graduated or left school for respondent's highest education level was available, this was calendar year left school.
					
		3) If the year graduated or left school for respondent's highest education level attended was missing, the calendar year left school was approximated by adding the number of years stayed at that education level to the calendar year left school from the previous level.  For those whose highest education level attended was primary school and missing on year left school, then the birth year plus 6 was used in the place for the year left school from the previous level.  Imputed status for this variable was coded 1 for conditions indicated here.
					
		4) If the year graduated or left school for respondent's highest education level attended and the number of years stayed were both missing, the calendar year left school were coded missing (-9999).  Or, if both information of the years graduated or left school for respondent's highest education level attended and for the previous level were missing, the calendar year left school were coded missing (-9999).
					
		5) For those who did not attend school, the calendar year left school was coded -8887.
		
		6) For those who were students at the survey time, the calendar year left school was coded -8886.
					
		7) There were two cases (id=10501092 and 14210802) who attended primary school and left school at age 6. These two respondents attended primary school and stayed for 0 year without information of years left school.  No further education.  So, the year left school was at age 6 by the approximate rule #3.  In sum, the data of YRENDED for these 2 cases were fine.
		
		-9999: Blank when response expected
		-8887: No schooling (A10_1A=2)
		-8886: Current students

MSENDEDY	APPROXIMATED YEAR LEFT SCHOOL

		0, 1 dummy variable.  Coded 1 for year left school was missing.  See note for year left school for conditions to be coded 1.

		0: Year left school was not missing
		1: Year left school was missing
		-99: Missing

HIED    	WHETHER ATTENDED HIGH EDU A10_5A TO A10_7A

		0, 1 dummy variable.  Coded 1 if attended junior college (a10_5a=1), university (a10_6a=1), or graduate school (a10_7a=1).

		0: No high education
		1: having high education
		No missing code.

A10_4E3 	MAJOR IN 3 CATEGORIES: TECH HIGH SCHOOL

		Categorical Variable coded 1, 2, or 3
		Recode from A10_4E.  Recode rule is:	
			
			A10_4E3=1	if A10_4E=1, 7, 8, or 9	
			A10_4E3=2	if A10_4E=2, 4, or 5	
			A10_4E3=3	if A10_4E=3 or 6	
			A10_4E3=-98	if A10_4E=10		

		-99: Missing
		-98: DK or other major
		-87: Not attend vocational/technical high school
		-77: Problematic education history

A10_5E3 	MAJOR IN 3 CATEGORIES: JUNIOR COLLEGE

		Categorical Variable coded 1, 2, or 3
		Recode from A10_5E.  Recode rule is:	

			A10_5E3=1	if A10_5E=1, 7, 8, or 9	
			A10_5E3=2	if A10_5E=2, 4, or 5	
			A10_5E3=3	if A10_5E=3 or 6	
			A10_5E3=-98	if A10_5E=10		

		-99: Missing
		-98: DK or other major
		-87: Not junior college
		-77: Problematic education history

A10_6E3 	MAJOR IN 3 CATEGORIES: 4-YEARS COLLEGE

		Categorical Variable coded 1, 2, or 3
		Recode from A10_6E.  Recode rule is:	

			A10_6E3=1	if A10_6E=1, 7, 8, or 9	
			A10_6E3=2	if A10_6E=2, 4, or 5	
			A10_6E3=3	if A10_6E=3 or 6	
			A10_6E3=-98	if A10_6E=10		

		-99: Missing
		-98: DK or other major
		-87: Not attend four years college

A10_7E3 	MAJOR IN 3 CATEGORIES: GRADUATE SCHOOL

		Categorical Variable coded 1, 2, or 3
		Recode from A10_7E.  Recode rule is:	

			A10_7E3=1	if A10_7E=1, 7, 8, or 9	
			A10_7E3=2	if A10_7E=2, 4, or 5	
			A10_7E3=3	if A10_7E=3 or 6	
			A10_7E3=-98	if A10_7E=10		

		-99: Missing
		-98: DK or other major
		-87: Not graduate school

EYYEAR  	LENGTH OF YEARS IN EDUCATED YOUTH

		Measure in years.  Range: 0-36
		The variable was obtained by subtracting the year beginning the youth education from the year ending the youth education.
		
			EYYEAR=EYEND-EYSTART
		
		-99: Missing
		-98: DK or other major
		-87: Not an educated youth

LENPARTY	YEARS SINCE JOINED THE PARTY TILL 1999

		Measure in years.  Range:0-61
		The variable was obtained by subtracting the year joining the part from Year 1999.

			LENPARTY=99-YPARTY

		-99: Missing
		-87: Not a party member

FEDLEVL8	FATHER'S EDUCATION LEVEL MISS=0

		Range: 0-8
		Recode missing codes of father's education level (FEDLEVL) such as blank, DK, and elderly respondents into 0 while the other codes were the same as of FEDLEVL.

		No missing code.

MSFED   	MISSING INDEX:FATHER'S EDUCATION LEVEL

		0, 1 dummy variable.  Coded 1 if information on father's education level (FEDLEVL) was missing including Blank when response expected, dk, and elderly respondents.

MEDLEVL8	MOTHER'S EDUCATION LEVEL MISS=0

		Range: 0-8
		Recode missing codes of mother's education level (MEDLEVL) such as blank, DK, and elderly respondents into 0 while the other codes were the same as of MEDLEVL.
		
		No missing code.
		
MSMED   	MISSING INDEX:MOTHER'S EDUCATION LEVEL

		0, 1 dummy variable.  Coded 1 if information on mother's education level (MEDLEVL) was missing including Blank when response expected, dk, and elderly respondents.

SPED    	SPOUSE'S YEARS OF EDUCATION

		Range: 3-19
		Recode spouse's education level (SEDLEVL) into years of education.
		Recode rule is:

			SPED=3		if SEDLEVL=1
			SPED=6		if SEDLEVL=2
			SPED=9		if SEDLEVL=3		
			SPED=12		if SEDLEVL=4 or 5	
			SPED=15		if SEDLEVL=6		
			SPED=16		if SEDLEVL=7		
			SPED=19		if SEDLEVL=8		

		-99: Missing
		-87: Not married (MARITAL=2, 3, or 4)

DADED   	FATHER'S YEARS OF EDUCATION

		Range: 3-19
		Recode father's education level (FEDLEVL) into years of education.
		Recode rule is:
			if FEDLEVL=1		DADED=3
			if FEDLEVL=2		DADED=6
			if FEDLEVL=3		DADED=9
			if FEDLEVL=4 or 5	DADED=12
			if FEDLEVL=6		DADED=15
			if FEDLEVL=7		DADED=16
			if FEDLEVL=8		DADED=19

		-99: Missing
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)

MOMED   	MOTHER'S YEARS OF EDUCATION

		Range: 3-16
		Recode mother's education level (MEDLEVL) into years of education.
		Recode rule is:
			if MEDLEVL=1		MOMED=3
			if MEDLEVL=2		MOMED=6
			if MEDLEVL=3		MOMED=9
			if MEDLEVL=4 or 5	MOMED=12
			if MEDLEVL=6		MOMED=15
			if MEDLEVL=7		MOMED=16
			if MEDLEVL=8		MOMED=19

		-99: Missing
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)

SIBED   	SIBLING'S YEARS OF EDUCATION

		Range: 3-19
		Recode sibling's education level (SIBEDLV) into years of education.
		Recode rule is:

			SIBED=3		if SIBEDLV=1
			SIBED=6		if SIBEDLV=2
			SIBED=9		if SIBEDLV=3
			SIBED=12	if SIBEDLV=4 or 5
			SIBED=15	if SIBEDLV=6
			SIBED=16	if SIBEDLV=7
			SIBED=19	if SIBEDLV=8

		-99: Missing
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-87: No sibling (HAVESIB=2)

K1ED    	1ST KID'S YEARS OF EDUCATION

		Range: 3-19
		Recode 1st child's education level (K1EDUC) into years of education.
		Recode rule is:
			
			K1ED=3		if K1EDUC=1		
			K1ED=6		if K1EDUC=2		
			K1ED=9		if K1EDUC=3		
			K1ED=12		if K1EDUC=4 or 5	
			K1ED=15		if K1EDUC=6		
			K1ED=16		if K1EDUC=7		
			K1ED=19		if K1EDUC=8		

		-99: Missing
		-98: DK
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: Childless (NUMKIDS=0)

K2ED    	2ND KID'S YEARS OF EDUCATION

		Range: 3-19
		Recode 2nd child's education level (K2EDUC) into years of education.
		Recode rule is:
			
			K2ED=3		if K2EDUC=1
			K2ED=6		if K2EDUC=2
			K2ED=9		if K2EDUC=3
			K2ED=12		if K2EDUC=4 or 5
			K2ED=15		if K2EDUC=6
			K2ED=16		if K2EDUC=7
			K2ED=19		if K2EDUC=8

		-99: Missing
		-98: DK
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS=0 or 1)

K3ED    	3RD KID'S YEARS OF EDUCATION

		Range: 3-19
		Recode 3rd child's education level (K3EDUC) into years of education.
		Recode rule is:

			K3ED=3		if K3EDUC=1
			K3ED=6		if K3EDUC=2
			K3ED=9		if K3EDUC=3
			K3ED=12		if K3EDUC=4 or 5
			K3ED=15		if K3EDUC=6
			K3ED=16		if K3EDUC=7
			K3ED=19		if K3EDUC=8

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS=0, 1 or 2)

K4ED    	4TH KID'S YEARS OF EDUCATION

		Range: 3-19
		Recode 4th child's education level (K4EDUC) into years of education.
		Recode rule is:

			K4ED=3		if K4EDUC=1		
			K4ED=6		if K4EDUC=2		
			K4ED=9		if K4EDUC=3		
			K4ED=12		if K4EDUC=4 or 5	
			K4ED=15		if K4EDUC=6		
			K4ED=16		if K4EDUC=7		
			K4ED=19		if K4EDUC=8		

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS=0, 1, 2 or 3)

K5ED    	5TH KID'S YEARS OF EDUCATION

		Range: 3-16
		Recode 5th child's education level (K5EDUC) into years of education.
		Recode rule is:

			K5ED=3		if K5EDUC=1		
			K5ED=6		if K5EDUC=2		
			K5ED=9		if K5EDUC=3		
			K5ED=12		if K5EDUC=4 or 5	
			K5ED=15		if K5EDUC=6		
			K5ED=16		if K5EDUC=7		
			K5ED=19		if K5EDUC=8		

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS le 4)

K6ED    	6TH KID'S YEARS OF EDUCATION

		Range: 3-19
		Recode 6th child's education level (K6EDUC) into years of education.
		Recode rule is:

			K6ED=3		if K6EDUC=1		
			K6ED=6		if K6EDUC=2		
			K6ED=9		if K6EDUC=3		
			K6ED=12		if K6EDUC=4 or 5	
			K6ED=15		if K6EDUC=6		
			K6ED=16		if K6EDUC=7		
			K6ED=19		if K6EDUC=8		

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS le 5)

K7ED    	7TH KID'S YEARS OF EDUCATION

		Range: 6-16
		Recode 7th child's education level (K7EDUC) into years of education.
		Recode rule is:

			K7ED=3		if K7EDUC=1		
			K7ED=6		if K7EDUC=2		
			K7ED=9		if K7EDUC=3		
			K7ED=12		if K7EDUC=4 or 5	
			K7ED=15		if K7EDUC=6		
			K7ED=16		if K7EDUC=7		
			K7ED=19		if K7EDUC=8		

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS le 6)

K8ED    	8TH KID'S YEARS OF EDUCATION

		Range: 6-15
		Recode 8th child's education level (K8EDUC) into years of education.
		Recode rule is:

			K8ED=3		if K8EDUC=1		
			K8ED=6		if K8EDUC=2		
			K8ED=9		if K8EDUC=3		
			K8ED=12		if K8EDUC=4 or 5	
			K8ED=15		if K8EDUC=6		
			K8ED=16		if K8EDUC=7		
			K8ED=19		if K8EDUC=8		

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS le 7)

K9ED    	9TH KID'S YEARS OF EDUCATION

		Range: 6-12
		Recode 9th child's education level (K9EDUC) into years of education.
		Recode rule is:

			K9ED=3		if K9EDUC=1		
			K9ED=6		if K9EDUC=2		
			K9ED=9		if K9EDUC=3		
			K9ED=12		if K9EDUC=4 or 5	
			K9ED=15		if K9EDUC=6		
			K9ED=16		if K9EDUC=7		
			K9ED=19		if K9EDUC=8		

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS le 8)

DK1MAR		1st kid's marital status_dummy

		Recode 1st child's marital status (K1MAR) into a dummy variable.
		Recode rule is:

			DK1MAR=0	if K1MAR=1
			DK1MAR=1	if K1MAR=2

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: Childless (NUMKIDS=0)

DK2MAR		2nd kid's marital status_dummy

		Recode 2nd child's marital status (K2MAR) into a dummy variable.
		Recode rule is:

			DK2MAR=0	if K2MAR=1
			DK2MAR=1	if K2MAR=2

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS=0 or 1)

DK3MAR		3rd kid's marital status_dummy

		Recode 3rd child's marital status (K3MAR) into a dummy variable.
		Recode rule is:

			DK3MAR=0	if K3MAR=1		
			DK3MAR=1	if K3MAR=2		

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS=0, 1 or 2)

DK4MAR		4th kid's marital status_dummy

		Recode 4th child's marital status (K4MAR) into a dummy variable.
		Recode rule is:

			DK4MAR=0	if K4MAR=1
			DK4MAR=1	if K4MAR=2

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS=0, 1, 2 or 3)

DK5MAR		5th kid's marital status_dummy

		Recode 5th child's marital status (K5MAR) into a dummy variable.
		Recode rule is:

			DK5MAR=0	if K5MAR=1
			DK5MAR=1	if K5MAR=2

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS le 4)

DK6MAR		6th kid's marital status_dummy

		Recode 6th child's marital status (K6MAR) into a dummy variable.
		Recode rule is:

			DK6MAR=0	if K6MAR=1		
			DK6MAR=1	if K6MAR=2		

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS le 5)

DK7MAR		7th kid's marital status_dummy

		Recode 7th child's marital status (K7MAR) into a dummy variable.
		Recode rule is:

			DK7MAR=0	if K7MAR=1		
			DK7MAR=1	if K7MAR=2		

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS le 6)

DK8MAR		8th kid's marital status_dummy

		Recode 8th child's marital status (K8MAR) into a dummy variable.
		Recode rule is:

			DK8MAR=0	if K8MAR=1
			DK8MAR=1	if K8MAR=2

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS le 7)

DK9MAR		9th kid's marital status_dummy

		Recode 9th child's marital status (K9MAR) into a dummy variable.
		Recode rule is:

			DK9MAR=0	if K9MAR=1		
			DK9MAR=1	if K9MAR=2		

		-99: Missing
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: No child this birth order (NUMKIDS le 8)

RETIRE1 	EVER RETIRED: 1 YES 0 OTHERWISE

		0, 1 dummy variable
		Coded 1 if RNOWORK eq 3 or EVRETIR eq 1 or RTYPE eq 1 or RTYPE eq 2; 
		coded 0 otherwise.
			
		No missing code.

JOB12   	WHETHER FIRST AND CURRENT JOBS ARE THE SAME

		0, 1 dummy variable
		Zhen Zeng created this variable.  See 7clean.sps
		Based on Job History on Section I Q22 of the questionnaire.  
		
		Coded 1 for those 
		1) who indicated the last/current jobs were the same as the first jobs on the last/current job responsibility description;
		2) who held the first job at the survey time (i.e., the ending year of the first job was "up to present");
		3) who started the first job and the last/current job on the same year, but the starting and ending years for the first job were not the same; or
		4) whose first job and longest job were the same while the last/current job and the longest job were also the same.
		
		Job12 was also coded 1 individually for id=12406072, 30408741, 31705101, 31705111, and 33610112 in 7clean.sps
			
		JOB12 was recoded to be 1 for id=11203031, 11704273, and 11804581 based on their job histories after the typo errors of ending years for the longest jobs were corrected.  The corrected job histories indicated that they had only one job for their entire career.  (Corrected by Meichu Chen)
		
		No missing code.

JOB13   	WHETHER FIRST AND LONGEST JOBS ARE THE SAME

		0, 1 dummy variable
		Zhen Zeng created this variable.  See 7clean.sps
		Based on Job History on Section I Q22 of the questionnaire.  
		
		Coded 1 for those 
		1) who indicated the longest jobs were the same as the first jobs on the longest job responsibility description;
		2) who started the first job and the longest job on the same year, but the starting and ending years for the first job were not the same; or
		3) whose first job and last/current job were the same while the last/current job and the longest job were also the same.
		
		Job13 was also coded 1 individually for id=14712081, 30408741, 30504801, 30504861, 30706011, 30706121, 32703593, 32703691, 33610112, 33913482, 35811881, 35911141, 46513112, and 41004411 in 7clean.sps
		
		JOB13 was recoded to be 1 for id=11203031, 11704273, and 11804581 based on their job histories after the typo errors of ending years for the longest jobs were corrected.  The corrected job histories indicated that they had only one job for their entire career.  (Corrected by Meichu Chen)

		JOB13 was changed to 1 for id=12706953 and to 0 for id=44207732 based on their job histories from the scanned questionnaires (12706953.pdf and 44207732.pdf).  (Corrected by Meichu Chen)

		No missing code.
		
JOB23   	WHETHER CURRENT AND LONGEST JOBS ARE THE SAME

		0, 1 dummy variable
		Zhen Zeng created this variable.  See 7clean.sps
		Based on Job History on Section I Q22 of the questionnaire.  
		
		Coded 1 for those 
		1) who indicated the longest job was the same as the last/current job on the last/current job responsibility description;
		2) who started the last/current job and the longest job on the same year, but the starting and ending years for the last/current job were not the same; or
		3) whose first job and last/current job were the same while the first job and the longest job were also the same.

		Job23 was also coded 1 individually for id=10501244, 30101602, 31000291, 31705101, and 31909311 in 7clean.sps
		
		JOB23 was recoded to be 1 for id=11203031, 11704273, and 11804581 based on their job histories after the typo errors of ending years for the longest jobs were corrected.  The corrected job histories indicated that they had only one job for their entire career.  (Corrected by Meichu Chen)

		JOB23 was changed to 1 for id=44207732 based on the job history from the scanned questionnaire (44207732.pdf).  (Corrected by Meichu Chen)

		No missing code.
		
JOB1EY8 	ENDING YEAR FIRST JOB-PRESENT=1999

		Range: 1928-1999
		Recode "up to present" for the ending year for the first job (JOB1EY=-8000) into 1999.

		-9999: Blank when response expected
		-9998: DK
		-8887: Never worked (EVWORK=2)
		-7777: Unreasonable year information
					
JOB2EY8 	ENDING YEAR LAST/CURRENT JOB-PRESENT=1999

		Range: 1931-1999
		Recode "up to present" for the ending year for the last/current job (JOB2EY=-8000) into 1999.

		-9999: Blank when response expected
		-9998: DK
		-8887: Never worked (EVWORK=2)

JOB3EY8 	ENDING YEAR LONGEST JOB-PRESENT=1999

		Range: 1931-1999
		Recode "up to present" for the ending year for the longest job (JOB3EY=-8000) into 1999.

		-9999: Blank when response expected
		-9998: DK
		-8887: Never worked (EVWORK=2)

JOB1EYT		year ending first job truncated in 1999

		Dummy indicator whether the ending year for the first job was truncated in 1999 
		coded 1 if respondent held the first job at the survey time (i.e. JOB1EY=-8000), and coded 0 if respondent ended the first job in 1999 or earlier.

		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)
		-77: Unreasonable year information
					
JOB2EYT		year ending last/current job truncated in 1999

		Dummy indicator whether the ending year for the last/current job was truncated in 1999 
		coded 1 if respondent held the last/current job at the survey time (i.e. JOB2EY=-8000), and 
		coded 0 if respondent ended the first job in 1999 or earlier.

		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)
					
JOB3EYT		year ending longest job truncated in 1999

		Dummy indicator whether the ending year for the longest job was truncated in 1999 
		coded 1 if respondent held the longest job at the survey time (i.e. JOB3EY=-8000), and coded 0 if respondent ended the longest job in 1999 or earlier.

		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)
					
LENGTH1 	LENGTH (IN YEARS) FOR FIRST JOB

		Range: 0-64.
		The variable was obtained by subtracting the year starting the first job from the year ending the first job when both staring and ending years data were valid.
		
			LENGTH1=JOB1EY8-JOB1SY

		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)
		-77: Unreasonable year information

LENGTH2 	LENGTH (IN YEARS) FOR CURRENT/LAST JOB

		Range: 0-64.
		The variable was obtained by subtracting the year starting the last/current job from the year ending the last/current job when both staring and ending years data were valid.
		
			LENGTH2=JOB2EY8-JOB2SY

		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)
		-77: Unreasonable year information

LENGTH3 	LENGTH (IN YEARS) FOR LONGEST JOB

		Range: 0-64.
		The variable was obtained by subtracting the year starting the longest job from the year ending the longest job when both staring and ending years data were valid.
		
			LENGTH3=JOB3EY8-JOB3SY

		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)
		-77: Unreasonable year information

EXP     	YEARS OF WORK EXPERIENCE

		Range: 0-64
		The variable was obtained by subtracting the year beginning the 1st job from the year ending the last/current job or the year retired when the year ending the last/current job was missing.  For those who never work, this variable was coded 0.

			EXP=JOB2EY8-JOB1SY		if JOB2EY8 GT 0 AND JOB1SY GT 0, or
			   =1900+YRETIRE-JOB1SY		if YRETIRE GT 0 AND JOB1SY GT 0, or
			   =0				if EVWORK=2

		-99: Blank when response expected
		-98: DK
		-77: Unreasonable year information

RUNIT1  	1ST JOB WORK UNIT: 4 GROUPS

		Recode from UNIT1.
		Recode rule is:	

			RUNIT1=1	if UNIT1=1, 2, 3, or 9
			RUNIT1=2	if UNIT1=4
			RUNIT1=3	if UNIT1=5, 6, 7, or 10
			RUNIT1=4	if UNIT1=8, 11, or 12
		
		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)

RUNIT2  	CURRENT/LAST JOB WORK UNIT: 4 GROUPS

		Recode from UNIT2.
		Recode rule is:	

			RUNIT2=1	if UNIT2=1, 2, 3, or 9	
			RUNIT2=2	if UNIT2=4		
			RUNIT2=3	if UNIT2=5, 6, 7, or 10	
			RUNIT2=4	if UNIT2=8, 11, or 12 	
		
		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)

RUNIT3  	LONGEST JOB WORK UNIT: 4 GROUPS

		Recode from UNIT3.
		Recode rule is:	

			RUNIT3=1	if UNIT3=1, 2, 3, or 9	
			RUNIT3=2	if UNIT3=4		
			RUNIT3=3	if UNIT3=5, 6, 7, or 10	
			RUNIT3=4	if UNIT3=8, 11, or 12 	
		
		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)

ROJOBUNI	OTHER JOB WORK UNIT: 4 GROUPS

		Recode from OJOBUNI.
		Recode rule is:	

			ROJOBUNI=1	if OJOBUNI=1, 2, 3, or 9
			ROJOBUNI=2	if OJOBUNI=4
			ROJOBUNI=3	if OJOBUNI=5, 6, 7, or 10
			ROJOBUNI=4	if OJOBUNI=8, 11, or 12
		
		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)
		-85: No second job (OTHJOB=2)

RSPUNIT 	SPOUSE'S WORK UNIT: 4 GROUPS

		Recode from SPUNIT.
		Recode rule is:	

			RSPUNIT=1	if SPUNIT=1, 2, 3, or 9
			RSPUNIT=2	if SPUNIT=4
			RSPUNIT=3	if SPUNIT=5, 6, 7, or 10
			RSPUNIT=4	if SPUNIT=8, 11, or 12
		
		-99: Blank when response expected
		-98: DK
		-87: Not married (MARITAL=2, 3, or 4)
		-86: Spouse never worked (SPUNIT=-86, i.e., SPUNIT=-6 as instructed on the questionnaire)

RFUNIT16	FATHER'S WORK UNIT WHEN R AGE 16:4 GROUPS

		Recode from FUNIT16.
		Recode rule is:	

			RFUNIT16=1	if FUNIT16=1, 2, 3, or 9
			RFUNIT16=2	if FUNIT16=4
			RFUNIT16=3	if FUNIT16=5, 6, 7, or 10
			RFUNIT16=4	if FUNIT16=8, 11, or 12
		
		-99: Blank when response expected
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-87: Father was dead when respondent was at age 16 (FUNIT16=-87, i.e., FUNIT16=-8 as instructed on the questionnaire)
		-86: Father had no job when respondent was at age 16 (FUNIT16=-86, i.e., FUNIT16=-6 as instructed on the questionnaire)

RMUNIT16	MOTHER'S WORK UNIT WHEN R AGE 16:4 GROUPS

		Recode from MUNIT16.
		Recode rule is:	

			RMUNIT16=1	if MUNIT16=1, 2, 3, or 9
			RMUNIT16=2	if MUNIT16=4
			RMUNIT16=3	if MUNIT16=5, 6, 7, or 10
			RMUNIT16=4	if MUNIT16=8, 11, or 12
		
		-99: Blank when response expected
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-87: Mother was dead when respondent was at age 16 (MUNIT16=-87, i.e., MUNIT16=-8 as instructed on the questionnaire)
		-86: Mother had no job when respondent was at age 16 (MUNIT16=-86, i.e., MUNIT16=-6 as instructed on the questionnaire)
		-77: Possibly coded as no job erroneously.  One case (ID=14511451) MUNIT16=-6 while MSECT16=3, & MOCC16=738; the other case (ID=42200031) MUNIT16=-6 & MSECT16=-6 while MOCC16=756.

RSIBUNIT	SIBLING'S WORK UNIT: 4 GROUPS

		Recode from SIBUNIT.
		Recode rule is:	

			RSIBUNIT=1	if SIBUNIT=1, 2, 3, or 9
			RSIBUNIT=2	if SIBUNIT=4
			RSIBUNIT=3	if SIBUNIT=5, 6, 7, or 10
			RSIBUNIT=4	if SIBUNIT=8, 11, or 12
		
		-99: Blank when response expected
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-87: No sibling (HAVESIB=2)
		-86: Sibling no job at survey time (SIBUNIT=-86, i.e., SIBUNIT=-6 as instructed on the questionnaire)

STATE1  	1ST JOB UNIT:STATE OWNED

		0, 1 dummy variable.
		Dichotomize RUNIT1.

			STATE1=1	if RUNIT1=1
			STATE1=0	otherwise
			
		No missing code.

COLLC1  	1ST JOB UNIT:COLLECTIVE

		0, 1 dummy variable.
		Dichotomize RUNIT1.

			COLLC1=1	if RUNIT1=2
			COLLC1=0	otherwise
			
		No missing code.

PRIVT1  	1ST JOB UNIT:PRIVATE

		0, 1 dummy variable.
		Dichotomize RUNIT1.

			PRIVT1=1	if RUNIT1=3		
			PRIVT1=0	otherwise
			
		No missing code.

PEASN1  	1ST JOB UNIT:PEASANT; NO UNIT

		0, 1 dummy variable.
		Dichotomize RUNIT1.

			PEASN1=1	if RUNIT1=4
			PEASN1=0	otherwise
			
		No missing code.

MSUNIT1 	MISSING INDICATOR:1ST JOB WORK UNIT

		0, 1 dummy variable.
		Dichotomize RUNIT1.

			MSUNIT1=1	if RUNIT1 lt 0
			MSUNIT1=0	otherwise
			
		No missing code.

STATE2  	CT/LAST JOB UNIT:STATE OWNED

		0, 1 dummy variable.
		Dichotomize RUNIT2.

			STATE2=1	if RUNIT2=1
			STATE2=0	otherwise
			
		No missing code.

COLLC2  	CT/LAST JOB UNIT:COLLECTIVE

		0, 1 dummy variable.
		Dichotomize RUNIT2.

			COLLC2=1	if RUNIT2=2
			COLLC2=0	otherwise
			
		No missing code.

PRIVT2  	CT/LAST JOB UNIT:PRIVATE

		0, 1 dummy variable.
		Dichotomize RUNIT2.

			PRIVT2=1	if RUNIT2=3
			PRIVT2=0	otherwise
			
		No missing code.

PEASN2  	CT/LAST JOB UNIT:PEASANT; NO UNIT

		0, 1 dummy variable.
		Dichotomize RUNIT2.

			PEASN2=1	if RUNIT2=4
			PEASN2=0	otherwise
			
		No missing code.

MSUNIT2 	MISSING INDICATOR:CT/LAST JOB WORK UNIT

		0, 1 dummy variable.
		Dichotomize RUNIT2.

			MSUNIT2=1	if RUNIT2 lt 0
			MSUNIT2=0	otherwise
			
		No missing code.

STATE3  	LONGEST JOB UNIT:STATE OWNED

		0, 1 dummy variable.
		Dichotomize RUNIT3.

			STATE3=1	if RUNIT3=1
			STATE3=0	otherwise
			
		No missing code.

COLLC3  	LONGEST JOB UNIT:COLLECTIVE

		0, 1 dummy variable.
		Dichotomize RUNIT3.

			COLLC3=1	if RUNIT3=2
			COLLC3=0	otherwise
			
		No missing code.

PRIVT3  	LONGEST JOB UNIT:PRIVATE

		0, 1 dummy variable.
		Dichotomize RUNIT3.

			PRIVT3=1	if RUNIT3=3
			PRIVT3=0	otherwise
			
		No missing code.

PEASN3  	LONGEST JOB UNIT:PEASANT; NO UNIT

		0, 1 dummy variable.
		Dichotomize RUNIT3.

			PEASN3=1	if RUNIT3=4
			PEASN3=0	otherwise
			
		No missing code.

MSUNIT3 	MISSING INDICATOR:LONGEST JOB WORK UNIT

		0, 1 dummy variable.
		Dichotomize RUNIT3.

			MSUNIT3=1	if RUNIT3 lt 0
			MSUNIT3=0	otherwise
			
		No missing code.

STATEOJ 	OTHER JOB UNIT:STATE OWNED

		0, 1 dummy variable.
		Dichotomize ROJOBUNI.

			STATEOJ=1	if ROJOBUNI=1
			STATEOJ=0	otherwise
			
		No missing code.

COLLCOJ 	OTHER JOB UNIT:COLLECTIVE

		0, 1 dummy variable.
		Dichotomize ROJOBUNI.

			COLLCOJ=1	if ROJOBUNI=2
			COLLCOJ=0	otherwise
			
		No missing code.

PRIVTOJ 	OTHER JOB UNIT:PRIVATE

		0, 1 dummy variable.
		Dichotomize ROJOBUNI.

			PRIVTOJ=1	if ROJOBUNI=3
			PRIVTOJ=0	otherwise
			
		No missing code.

PEASNOJ 	OTHER JOB UNIT:PEASANT; NO UNIT

		0, 1 dummy variable.
		Dichotomize ROJOBUNI.

			PEASNOJ=1	if ROJOBUNI=4
			PEASNOJ=0	otherwise
			
		No missing code.

MSUNITOJ	MISSING INDICATOR:OTHER JOB WORK UNIT

		0, 1 dummy variable.
		Dichotomize ROJOBUNI.

			MSUNITOJ=1	if ROJOBUNI lt 0
			MSUNITOJ=0	otherwise
			
		No missing code.

STATESP 	SPOUSE WORK UNIT:STATE OWNED

		0, 1 dummy variable.
		Dichotomize RSPUNIT.

			STATESP=1	if RSPUNIT=1
			STATESP=0	otherwise
			
		No missing code.

COLLCSP 	SPOUSE WORK UNIT:COLLECTIVE

		0, 1 dummy variable.
		Dichotomize RSPUNIT.

			COLLCSP=1	if RSPUNIT=2
			COLLCSP=0	otherwise
			
		No missing code.

PRIVTSP 	SPOUSE WORK UNIT:PRIVATE

		0, 1 dummy variable.
		Dichotomize RSPUNIT.

			PRIVTSP=1	if RSPUNIT=3
			PRIVTSP=0	otherwise
			
		No missing code.

PEASNSP 	SPOUSE WORK UNIT:PEASANT; NO UNIT

		0, 1 dummy variable.
		Dichotomize RSPUNIT.

			PEASNSP=1	if RSPUNIT=4
			PEASNSP=0	otherwise
			
		No missing code.

MSUNITSP	MISSING INDICATOR:SPOUSE'S WORK UNIT

		0, 1 dummy variable.
		Dichotomize RSPUNIT.

			MSUNITSP=1	if RSPUNIT lt 0
			MSUNITSP=0	otherwise
			
		No missing code.

STATEDAD	FATHER'S WORK UNIT:STATE OWNED

		0, 1 dummy variable.
		Dichotomize RFUNIT16.
			STATEDAD=1	if RFUNIT16=1
			STATEDAD=0	otherwise
			
		No missing code.

COLLCDAD	FATHER'S WORK UNIT:COLLECTIVE

		0, 1 dummy variable.
		Dichotomize RFUNIT16.
			COLLCDAD=1	if RFUNIT16=2
			COLLCDAD=0	otherwise
			
		No missing code.

PRIVTDAD	FATHER'S WORK UNIT:PRIVATE

		0, 1 dummy variable.
		Dichotomize RFUNIT16.
			PRIVTDAD=1	if RFUNIT16=3
			PRIVTDAD=0	otherwise
			
		No missing code.

PEASNDAD	FATHER'S WORK UNIT:PEASANT; NO UNIT

		0, 1 dummy variable.
		Dichotomize RFUNIT16.
			PEASNDAD=1	if RFUNIT16=4
			PEASNDAD=0	otherwise
			
		No missing code.

MSUNITDA	MISSING INDICATOR:FATHER'S WORK UNIT

		0, 1 dummy variable.
		Dichotomize RFUNIT16.
			MSUNITDA=1	if RFUNIT16 lt 0
			MSUNITDA=0	otherwise
			
		No missing code.

STATEMOM	MOTHER'S WORK UNIT:STATE OWNED

		0, 1 dummy variable.
		Dichotomize RMUNIT16.
			STATEMOM=1	if RMUNIT16=1
			STATEMOM=0	otherwise
			
		No missing code.

COLLCMOM	MOTHER'S WORK UNIT:COLLECTIVE

		0, 1 dummy variable.
		Dichotomize RMUNIT16.
			COLLCMOM=1	if RMUNIT16=2
			COLLCMOM=0	otherwise
			
		No missing code.

PRIVTMOM	MOTHER'S WORK UNIT:PRIVATE

		0, 1 dummy variable.
		Dichotomize RMUNIT16.
			PRIVTMOM=1	if RMUNIT16=3
			PRIVTMOM=0	otherwise
			
		No missing code.

PEASNMOM	MOTHER'S WORK UNIT:PEASANT; NO UNIT

		0, 1 dummy variable.
		Dichotomize RMUNIT16.
			PEASNMOM=1	if RMUNIT16=4
			PEASNMOM=0	otherwise
			
		No missing code.

MSUNITMO	MISSING INDICATOR:MOTHER'S WORK UNIT

		0, 1 dummy variable.
		Dichotomize RMUNIT16.
			MSUNITMO=1	if RMUNIT16 lt 0
			MSUNITMO=0	otherwise
			
		No missing code.

STATESIB	SIBLING'S WORK UNIT:STATE OWNED

		0, 1 dummy variable.
		Dichotomize RSIBUNIT.
			STATESIB=1	if RSIBUNIT=1
			STATESIB=0	otherwise
			
		No missing code.

COLLCSIB	SIBLING'S WORK UNIT:COLLECTIVE

		0, 1 dummy variable.
		Dichotomize RSIBUNIT.
			COLLCSIB=1	if RSIBUNIT=2
			COLLCSIB=0	otherwise
			
		No missing code.

PRIVTSIB	SIBLING'S WORK UNIT:PRIVATE

		0, 1 dummy variable.
		Dichotomize RSIBUNIT.
			PRIVTSIB=1	if RSIBUNIT=3
			PRIVTSIB=0	otherwise
			
		No missing code.

PEASNSIB	SIBLING'S WORK UNIT:PEASANT; NO UNIT

		0, 1 dummy variable.
		Dichotomize RSIBUNIT.
			PEASNSIB=1	if RSIBUNIT=4
			PEASNSIB=0	otherwise
			
		No missing code.

MSUNITSI	MISSING INDICATOR:SIBLING'S WORK UNIT

		0, 1 dummy variable.
		Dichotomize RSIBUNIT.
			MSUNITSI=1	if RSIBUNIT lt 0
			MSUNITSI=0	otherwise
			
		No missing code.

ISCO1   	1ST JOB:ISCO 1968 ENHANCED EDITION

		Range: 130-9995
		Converted from the first job CSCO95 occupational codes (OCC1) based on the mapping list from Treiman, filename CSCO_to_ISCO68.rtf.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-9999: Blank when response expected
		-9998: DK
		-8887: Never worked (EVWORK=2)

ISCO2   	LAST/CURRENT JOB:ISCO 1968 ENHANCED EDITION

		Range: 130-9995
		Converted from the last/current job CSCO95 occupational codes (OCC2) based on the mapping list from Treiman, filename CSCO_to_ISCO68.rtf.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).

		-9999: Blank when response expected
		-9998: DK
		-8887: Never worked (EVWORK=2)
					
ISCO3   	LONGEST JOB:ISCO 1968 ENHANCED EDITION

		Range: 130-9995
		Converted from the longest job CSCO95 occupational codes (OCC3) based on the mapping list from Treiman, filename CSCO_to_ISCO68.rtf.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).

		-9999: Blank when response expected
		-9998: DK
		-8887: Never worked (EVWORK=2)
										
ISCOOJ  	OTHER JOB:ISCO 1968 ENHANCED EDITION

		Range: 290-9900
		Converted from respondent's supplemental job CSCO95 occupational codes (OJOBOCC) based on the mapping list from Treiman, filename CSCO_to_ISCO68.rtf.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
					
		-9999: Blank when response expected
		-8887: Never worked (EVWORK=2)
		-8885: No supplemental job (OTHJOB=2)
										
ISCOSP  	SPOUSE'S JOB:ISCO 1968 ENHANCED EDITION

		Range: 130-9995
		Converted from spouse's job CSCO95 occupational codes (SPOCC) based on the mapping list from Treiman, filename CSCO_to_ISCO68.rtf.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
					
		-9999: Blank when response expected
		-9998: DK
		-8887: Not married (MARITAL=2, 3, or 4)
		-8886: Spouse never worked (SPUNIT=-86, i.e., SPUNIT=-6 as instructed on the questionnaire)
		-7777: Not valid CSCO95 occupational codes
										
ISCODAD 	FATHER'S JOB:ISCO 1968 ENHANCED EDITION

		Range: 140-9995
		Converted from father's job at respondent age 16 CSCO95 occupational codes (FOCC16) based on the mapping list from Treiman, filename CSCO_to_ISCO68.rtf.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).

		-9999: Blank when response expected
		-9998: DK
		-8888: Elderly respondent (FORM=2 or 3)
		-8887: Father dead at respondent age 16 (FUNIT16=-87, i.e., FUNIT16=-8 as instructed on the questionnaire)
		-8886: Father no job at respondent age 16 (FUNIT16=-86, i.e., FUNIT16=-6 as instructed on the questionnaire)

ISCOMOM 	MOTHER'S JOB:ISCO 1968 ENHANCED EDITION

		Range: 140-9995
		Converted from mother's job at respondent age 16 job CSCO95 occupational codes (MOCC16) based on the mapping list from Treiman, filename CSCO_to_ISCO68.rtf.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).

		-9999: Blank when response expected
		-9998: DK
		-8888: Elderly respondent (FORM=2 or 3)
		-8887: Mother dead at respondent age 16 (MUNIT16=-87, i.e., MUNIT16=-8 as instructed on the questionnaire)
		-8886: Mother no job at respondent age 16 (MUNIT16=-86, i.e., MUNIT16=-6 as instructed on the questionnaire)
		-7777: Possibly coded as no job erroneously.  MOCC16 was coded -6 while MSECT16 was coded 2 and MUNIT16 was coded either 1 or 3.

ISCOSIB 	SIBLING'S JOB:ISCO 1968 ENHANCED EDITION

		Range: 130-9995
		Converted from sibling's job CSCO95 occupational codes (SIBOCC) based on the mapping list from Treiman, filename CSCO_to_ISCO68.rtf.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).

		-9999: Blank when response expected
		-9998: DK
		-8888: Elderly respondent (FORM=2 or 3)
		-8887: No sibling (HAVESIB=2)
		-8886: Sibling no job at the survey time (SIBUNIT=-86, i.e., SIBUNIT=-6 as instructed on the questionnaire)
		-7777: Possibly coded as no job erroneously.  SIBOCC was coded -6 while SIBUNIT and SIBSECT had valid codes

SIOPS1  	1ST JOB:TREIMAN'S SIOPS PRESTIGE SCORE

		Range: 13-78
		Converted from the first job ISCO68 codes (ISCO1) based on the mapping list from Treiman, filename iscopres.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)

SIOPS2  	LAST/CURRENT JOB:TREIMAN'S SIOPS PRESTIGE SCORE

		Range: 13-78
		Converted from the last/current job ISCO68 codes (ISCO2) based on the mapping list from Treiman, filename iscopres.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)

SIOPS3  	LONGEST JOB:TREIMAN'S SIOPS PRESTIGE SCORE

		Range: 13-78
		Converted from the longest job ISCO68 codes (ISCO3) based on the mapping list from Treiman, filename iscopres.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)

SIOPSOJ 	OTHER JOB:TREIMAN'S SIOPS RESTIGE SCORE

		Range: 15-78
		Converted from the supplemental job ISCO68 codes (ISCOOJ) based on the mapping list from Treiman, filename iscopres.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-87: Never worked (EVWORK=2)
		-85: No supplemental job (OTHJOB=2)

SIOPSSP 	SPOUSE'S JOB:TREIMAN'S SIOPS PRESTIGE SCORE

		Range: 13-78
		Converted from spouse's job ISCO68 codes (ISCOSP) based on the mapping list from Treiman, filename iscopres.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-87: Not married (MARITAL=2, 3, or 4)
		-86: Spouse never worked (SPUNIT=-86, i.e., SPUNIT=-6 as instructed on the questionnaire)
		-77: Spouse's occupational code was not a valid code

SIOPSDAD	FATHER'S JOB:TREIMAN'S SIOPS PRESTIGE SCORE

		Range: 13-78
		Converted from father's job ISCO68 codes (ISCODAD) based on the mapping list from Treiman, filename iscopres.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-87: Father dead at respondent age 16 (FUNIT16=-87, i.e., FUNIT16=-8 as instructed on the questionnaire)
		-86: Father had no job at respondent age 16 (FUNIT16=-86, i.e., FUNIT16=-6 as instructed on the questionnaire)

SIOPSMOM	MOTHER'S JOB:TREIMAN'S SIOPS PRESTIGE SCORE

		Range: 13-78
		Converted from mother's job ISCO68 codes (ISCOMOM) based on the mapping list from Treiman, filename iscopres.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-87: Mother dead at respondent age 16 (MUNIT16=-87, i.e., MUNIT16=-8 as instructed on the questionnaire)
		-86: Mother had no job at respondent age 16 (MUNIT16=-86, i.e., MUNIT16=-6 as instructed on the questionnaire)
		-77: Possibly coded as no job erroneously.  MOCC16 was coded -6 while MSECT16 was coded 2 and MUNIT16 was coded either 1 or 3.

SIOPSSIB	SIBLING'S JOB:TREIMAN'S SIOPS PRESTIGE SCORE

		Range: 13-78
		Converted from sibling's job ISCO68 codes (ISCOSIB) based on the mapping list from Treiman, filename iscopres.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-87: No sibling (HAVESIB=2)
		-86: Sibling no job at the survey time (SIBUNIT=-86, i.e., SIBUNIT=-6 as instructed on the questionnaire)
		-77: Possibly coded as no job erroneously.  SIBOCC was coded -6 while SIBUNIT and SIBSECT had valid codes.

ISEI1   	1ST JOB:ISEI SCORE

		Range: 10-90
		Converted from the first job ISCO68 codes (ISCO1) based on the mapping list from Treiman, filename iscoisei.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)

ISEI2   	LAST/CURRENT JOB:ISEI SCORE

		Range: 10-88
		Converted from the last/current job ISCO68 codes (ISCO2) based on the mapping list from Treiman, filename iscoisei.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)

ISEI3   	LONGEST JOB:ISEI SCORE

		Range: 10-88
		Converted from the longest job ISCO68 codes (ISCO3) based on the mapping list from Treiman, filename iscoisei.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-87: Never worked (EVWORK=2)

ISEIOJ  	OTHER JOB:ISEI SCORE

		Range: 10-85
		Converted from the supplemental job ISCO68 codes (ISCOOJ) based on the mapping list from Treiman, filename iscoisei.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-87: Never worked (EVWORK=2)
		-85: No supplemental job (OTHJOB=2)

ISEISP  	SPUSE'S JOB:ISEI SCORE

		Range: 10-90
		Converted from spouse's job ISCO68 codes (ISCOSP) based on the mapping list from Treiman, filename iscoisei.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-87: Not married (MARITAL=2, 3, or 4)
		-86: Spouse never worked (SPUNIT=-86, i.e., SPUNIT=-6 as instructed on the questionnaire)
		-77: Spouse's occupational code was not a valid code

ISEIDAD 	FATHER'S JOB:ISEI SCORE

		Range: 10-88
		Converted from father's job ISCO68 codes (ISCODAD) based on the mapping list from Treiman, filename iscoisei.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-87: Father dead at respondent age 16 (FUNIT16=-87, i.e., FUNIT16=-8 as instructed on the questionnaire)
		-86: Father had no job at respondent age 16 (FUNIT16=-86, i.e., FUNIT16=-6 as instructed on the questionnaire)

ISEIMOM 	MOTHER'S JOB:ISEI SCORE

		Range: 10-88
		Converted from mother's job ISCO68 codes (ISCOMOM) based on the mapping list from Treiman, filename iscoisei.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-87: Mother dead at respondent age 16 (MUNIT16=-87, i.e., MUNIT16=-8 as instructed on the questionnaire)
		-86: Mother had no job at respondent age 16 (MUNIT16=-86, i.e., MUNIT16=-6 as instructed on the questionnaire)
		-77: Possibly coded as no job erroneously.  MOCC16 was coded -6 while MSECT16 was coded 2 and MUNIT16 was coded either 1 or 3.

ISEISIB 	SIBLING'S JOB:ISEI SCORE

		Range: 10-88
		Converted from sibling's job ISCO68 codes (ISCOSIB) based on the mapping list from Treiman, filename iscoisei.do.  See the document of "Converting Occupational Codes" for details (filename: occ_convert_doc.txt).
		
		-99: Blank when response expected
		-98: DK
		-88: Elderly respondent (FORM=2 or 3)
		-87: No sibling (HAVESIB=2)
		-86: Sibling no job at the survey time (SIBUNIT=-86, i.e., SIBUNIT=-6 as instructed on the questionnaire)
		-77: Possibly coded as no job erroneously.  SIBOCC was coded -6 while SIBUNIT and SIBSECT had valid codes.

SALARY8 	1998 ANNUAL SALARY -8=0

		Range: 0 to 120000
		This measure was obtained by multiplying 1998 monthly salary (SALARYM) by 12.  For those coded -888887 (never worked) on SALARYM, 1998 annual salary was 0.  

			SALARY8=SALARYM*12	if SALARYM>=0
			       =0		if SALARYM=-888887		

		For those 18 cases that SALARYM=-999999 or -999998, SALARY8 were imputed by regression method.  Respondent's gender, age, years of schooling, whether attended higher education, whether ever retired, and the series of dummy variables to indicate the last/current job work unit (in 4 groups) were used to predicted 1998 annual salary.  No missing data on any of the predictors. When the predicted annual salary (SALARY8) was negative (1 case), the value 0 was assigned.
		
		The predicted values of SALARY8 for the missing cases were calculated as:

			SALARY8=ROUND(1188.715645*MALE-6.539786*AGE+ 141.502420*SCHYR_C7+
			  	1437.067831*HIED-3484.273433*RETIRE1+1432.847581*STATE2+
			  	1372.624660*COLLC2+4968.330062*PRIVT2-2104.654840*MSUNIT2+
			  	1617.850574)

		No missing code.

MSSALA  	MISSING INDEX:1998 ANNUAL SALARY

		0, 1 dummy variable.  
		Coded 1 if SALARY8 was imputed.

RETIRE8 	1998 ANNUAL RETIRED PENSION -8=0

		Range from 0 to 24,000.
		This measure was obtained by multiplying 1998 monthly retirement pension (RETIREM) by 12.  For those coded -888887 (never worked) or -888884 (never retired), the 1998 annual retired pension was 0.

			RETIRE8=RETIREM*12	if RETIREM>=0
			       =0		if RETIREM=-888887 or -888884

		-999999: Blank when response expected
		-999998: DK

BONUS8  	1998 ANNUAL BONUS FROM MONTHLY -8=0

		Range from 0 to 36000.
		This measure was obtained by multiplying 1998 monthly bonus (BONUSM) by 12.  For those coded -888887 (never worked) on BONUSM, the 1998 annual bonus from monthly bonus was 0.

			BONUS8=BONUSM*12	if BONUSM>=0
			       =0		if BONUSM=-888887

		-999999: Blank when response expected
		-999998: DK

OTHJOB8 	1998 ANNUAL SALARY FROM OTHER JOB -8=0

		Range: 0 to 36000.
		This measure was obtained by multiplying monthly pay from other job (OTHJOBM) by 12.  For those coded  -888887 (never worked) or -888885 (no other job) on OTHJOBM, 1998 annul pay from other job was 0.


			OTHJOB8=OTHJOBM*12	if OTHJOBM>=0
			       =0		if OTHJOBM=-888887 or -888885

		-999999: Blank when response expected

BONUSY8 	1998 END OF YEAR BONUS -8=0

		Range from 0 to 30000.
		Recode -888887 (never work) on 1998 year end bonus (BONUSY) to 0.

			BONUSY8=BONUSY		if BONUSY>=0
			       =0		if BONUSY=-888887

		-999999: Blank when response expected
		-999998: DK

INVESTY8	1998 ANNUAL INVESTMENT INCOME -8=0

		Range: -20000 to 65000.
		Recode -888887 (never work) on 1998 annual investment income (INVESTY) to 0.

			INVESTY8=INVESTY	if INVESTY>=-20000
			        =0		if INVESTY=-888887

		-999999: Blank when response expected
		-999998: DK

OTHINCY8	1998 OTHER INCOME -8=0

		Range: 0 to 104000.
		Recode -888887 (never work) on 1998 income from all other sources (OTHINCY) to 0.

			OTHINCY8=OTHINCY	if OTHINCY>=0
			        =0		if OTHINCY=-888887

		-999999: Blank when response expected
		-999998: DK

INC98P  	R'S 1998 PERSONAL ANNAUAL INCOME

		Range: -11568 to 185000.
		Sum over 1998 personal income from all sources (i.e., SALARY8, RETIRE8, BONUS8, OTHJOB8, BONUSY8, INVESTY8, OTHINCY8). Except for SALARY8 where was no missing data, the missing data for each individual income source was implicitly assumed to be 0 for the sum.

			INC98P=SALARY8+RETIRE8+BONUS8+OTHJOB8+BONUSY8+INVESTY8+OTHINCY8

		No missing code.

HHINC0  	1998 MONTHLY HOUSEHOLD INCOME: NO MISSING

		Range: 0 to 59650.
		Impute values for 53 cases that 1998 monthly household income (HHINC) was missing.
		
			HHINC0=HHINC		if HHINC>= 0
		
		Among the 53 missing cases, 14 of them lived together with their match pairs whose HHINC values were not missing.  The HHINC values from their match pairs were assigned to HHINC0 for these 14 cases.

		For the other missing cases, regression method was used to impute the missing data.  Personal income in 1998 (INC98P), household size (HHSIZE), number of household members having income (HHEARN), gender (MALE), and age (AGE) were used to predict household monthly income. One case (id=35312972) has missing data on HHSIZE and HHEARN, and another case (id=35811753) has missing data on HHEARN.  The mode substitute was used to compute predicted values when the value of a predictor was missing.  The modes for HHSIZE and HHEARN were 3 and 2, respectively.
		
		The predicted values of HHINC0 for the missing cases were calculated as:

			HHINC0=ROUND(.080093*INC98P+93.831855*HHSIZE+366.398479*HHEARN-
    				197.025923*MALE-.831647*AGE-135.843758)			

		No negative predicted values

		No missing code.

MSHHINC 	MISSING INDICATOR: HH INCOME

		0, 1 dummy variable.
		Coded 1 if 1998 household income (hhinc0) was imputed, 0 otherwise.
		
AEXPA   	ANNUAL HH EXPENSE ON FOODS-1999

		Range: 0 to 120000.
		This measure was obtained by multiplying the monthly household expense on foods (EXPNA) by 12.

			AEXPA=EXPNA*12		if EXPNA>=0
		
		When a respondent's EXPNA was missing (coded -999999 or -999998), the respondent's AEXPA value was obtained from the AEXPA value of his/her living together match pair.  There were 80 cases of them.  When the AEXPA value for the living together match pair was not available, regression method was used to impute the missing value of AEXPA.

		Age (AGE), years of schooling (SCHYR_C7), household size (HHSIZE), number of household members having income (HHEARN), household income (HHINC0) and personal income in 1998 (INC98P) were used to predict monthly household expense on foods.  The mode substitute was used to compute predicted values when the value of a predictor was missing.  The modes for HHSIZE and HHEARN were 3 and 2, respectively.  The value of 0 was assigned when the predicted value was negative.
		
		The predicted values of AEXPA for the missing cases were calculated as:

			AEXPA=ROUND((.549545*AGE+6.046025*SCHYR_C7+41.950906*HHSIZE+
  				70.516639*HHEARN+.059153*HHINC0+.006634*INC98P+107.425085)*12)
  				
		The regression coefficients used for the imputation here are from the estimates of an early run of the model and are slightly different from the corresponding estimated regression coefficents in the last run of the model in 3city_c10a.lis.  However, the differences are minor.  The information of the differeces is avilable from Porfessor Yu Xie.

		No missing code.

AEXPB   	ANNUAL HH EXPENSE ON HOUSING-1999

		Range: 0 to 720000.
		This measure was obtained by multiplying the monthly household expense on housing (EXPNB) by 12.

			AEXPB=EXPNB*12		if EXPNB>=0
		
		When a respondent's EXPNB was missing (coded -999999 or -999998), the respondent's AEXPB value was obtained from the AEXPB value of his/her living together match pair.  There were 45 cases of them.  When the AEXPB value for the living together match pair was not available, regression method was used to impute the missing value of AEXPB.

		Age (AGE), years of schooling (SCHYR_C7), household size (HHSIZE), number of household members having income (HHEARN), household income (HHINC0) and personal income in 1998 (INC98P) were used to predict monthly household expense on housing.  The mode substitute was used to compute predicted values when the value of a predictor was missing.  The modes for HHSIZE and HHEARN were 3 and 2, respectively.  The value of 0 was assigned when the predicted value was negative.
		
		The predicted values of AEXPB for the missing cases were calculated as:

			AEXPB=ROUND((-3.049198*AGE-8.009523*SCHYR_C7-12.756417*HHSIZE-
    				13.759330*HHEARN-8.53993E-05*HHINC0+.007109*INC98P+
    				319.486724)*12)

		Note: The estimated regression coefficients for SCHYR_C7, HHSIZE, HHEARN, and HHINC0 were not statistically significant.

		The regression coefficients used for the imputation here are from the estimates of an early run of the model and are slightly different from the corresponding estimated regression coefficents in the last run of the model in 3city_c10a.lis.  However, the differences are minor.  The information of the differeces is avilable from Porfessor Yu Xie.

		No missing code.

AEXPC   	ANNUAL HH EXPENSE ON UTILITIES-1999

		Range: 0 to 14400.
		This measure was obtained by multiplying the monthly household expense on utilities (EXPNC) by 12.

			AEXPC=EXPNC*12		if EXPNC>=0
		
		When a respondent's EXPNC was missing (coded -999999 or -999998), the respondent's AEXPC value was obtained from the AEXPC value of his/her living together match pair.  There were 69 cases of them.  When the AEXPC value for the living together match pair was not available, regression method was used to impute the missing value of AEXPC.

		Age (AGE), years of schooling (SCHYR_C7), household size (HHSIZE), number of household members having income (HHEARN), household income (HHINC0) and personal income in 1998 (INC98P) were used to predict monthly household expense on utilities.  The mode substitute was used to compute predicted values when the value of a predictor was missing.  The modes for HHSIZE and HHEARN were 3 and 2, respectively.  The value of 0 was assigned when the predicted value was negative.
		
		The predicted values of AEXPC for the missing cases were calculated as:

			AEXPC=ROUND((.151852*AGE+.488880*SCHYR_C7+5.473810*HHSIZE+
      				7.926670*HHEARN+.011101*HHINC0+8.06571E-04*INC98P+
      				36.483257)*12)

			The regression coefficients used for the imputation here are from the estimates of an early run of the model and are slightly different from the corresponding estimated regression coefficents in the last run of the model in 3city_c10a.lis.  However, the differences are minor.  The information of the differeces is avilable from Porfessor Yu Xie.

		No missing code.

AEXPD   	ANNUAL HH EXPENSE ON TEL/COMMUNICATION-1999

		Range: 0 to 110880.
		This measure was obtained by multiplying the monthly household expense on telephone and communication (EXPND) by 12.

			AEXPD=EXPND*12		if EXPND>=0
		
		When a respondent's EXPND was missing (coded -999999 or -999998), the respondent's AEXPD value was obtained from the AEXPD value of his/her living together match pair.  There were 58 cases of them.  When the AEXPD value for the living together match pair was not available, regression method was used to impute the missing value of AEXPD.

		Age (AGE), years of schooling (SCHYR_C7), household size (HHSIZE), number of household members having income (HHEARN), household income (HHINC0) and personal income in 1998 (INC98P) were used to predict monthly household expense on telephone and communication.  The mode substitute was used to compute predicted values when the value of a predictor was missing.  The modes for HHSIZE and HHEARN were 3 and 2, respectively.  The value of 0 was assigned when the predicted value was negative.
		
		The predicted values of AEXPD for the missing cases were calculated as:

			AEXPD=ROUND((.197266*AGE+1.676756*SCHYR_C7+2.210798*HHSIZE+
   				8.617693*HHEARN+.004532*HHINC0+.002179*INC98P-18.516245)*12)

		The regression coefficients used for the imputation here are from the estimates of an early run of the model and are slightly different from the corresponding estimated regression coefficents in the last run of the model in 3city_c10a.lis.  However, the differences are minor.  The information of the differeces is avilable from Porfessor Yu Xie.

		No missing code.

AEXPE   	ANNUAL HH EXPENSE ON TRANPORTATION-1999

		Range: 0 to 18000.
		This measure was obtained by multiplying the monthly household expense on transportation (EXPNE) by 12.

			AEXPE=EXPNE*12		if EXPNE>=0
		
		When a respondent's EXPNE was missing (coded -999999 or -999998), the respondent's AEXPE value was obtained from the AEXPE value of his/her living together match pair.  There were 57 cases of them.  When the AEXPE value for the living together match pair was not available, regression method was used to impute the missing value of AEXPE.

		Age (AGE), years of schooling (SCHYR_C7), household size (HHSIZE), number of household members having income (HHEARN), household income (HHINC0) and personal income in 1998 (INC98P) were used to predict monthly household expense on transportation.  The mode substitute was used to compute predicted values when the value of a predictor was missing.  The modes for HHSIZE and HHEARN were 3 and 2, respectively.  The value of 0 was assigned when the predicted value was negative.
		
		The predicted values of AEXPE for the missing cases were calculated as:

			AEXPE=ROUND((-.259661*AGE+.989033*SCHYR_C7+.606179*HHSIZE+7.176112*HHEARN+
    				.007945*HHINC0+.001891*INC98P+1.545631)*12)

		The regression coefficients used for the imputation here are from the estimates of an early run of the model and are slightly different from the corresponding estimated regression coefficents in the last run of the model in 3city_c10a.lis.  However, the differences are minor.  The information of the differeces is avilable from Porfessor Yu Xie.

		No missing code.

AEXPF   	ANNUAL HH EXPENSE ON EDUCATION-1999

		Range: 0 to 36000.
		This measure was obtained by multiplying the monthly household expense on education (EXPNF) by 12.

			AEXPF=EXPNF*12		if EXPNF>=0
		
		When a respondent's EXPNF was missing (coded -999999 or -999998), the respondent's AEXPF value was obtained from the AEXPF value of his/her living together match pair.  There were 42 cases of them.  When the AEXPF value for the living together match pair was not available, regression method was used to impute the missing value of AEXPF.

		Age (AGE), years of schooling (SCHYR_C7), household size (HHSIZE), number of household members having income (HHEARN), household income (HHINC0) and personal income in 1998 (INC98P) were used to predict monthly household expense on education.  The mode substitute was used to compute predicted values when the value of a predictor was missing.  The modes for HHSIZE and HHEARN were 3 and 2, respectively.  The value of 0 was assigned when the predicted value was negative.
		
		The predicted values of AEXPF for the missing cases were calculated as:

			AEXPF=ROUND((-.588411*AGE+3.661574*SCHYR_C7+29.635062*HHSIZE-
   				24.523440*HHEARN+.015632*HHINC0+.001006*INC98P+36.622139)*12)

		The regression coefficients used for the imputation here are from the estimates of an early run of the model and are slightly different from the corresponding estimated regression coefficents in the last run of the model in 3city_c10a.lis.  However, the differences are minor.  The information of the differeces is avilable from Porfessor Yu Xie.

		No missing code.

AEXPG   	ANNUAL HH EXPENSE ON MEDICATON-1999

		Range: 0 to 150000.
		This measure was obtained by multiplying the monthly household expense on medication (EXPNG) by 12.

			AEXPG=EXPNG*12		if EXPNG>=0
		
		When a respondent's EXPNG was missing (coded -999999 or -999998), the respondent's AEXPG value was obtained from the AEXPG value of his/her living together match pair.  There were 49 cases of them.  When the AEXPG value for the living together match pair was not available, regression method was used to impute the missing value of AEXPG.

		Age (AGE), years of schooling (SCHYR_C7), household size (HHSIZE), number of household members having income (HHEARN), household income (HHINC0) and personal income in 1998 (INC98P) were used to predict monthly household expense on medication.  The mode substitute was used to compute predicted values when the value of a predictor was missing.  The modes for HHSIZE and HHEARN were 3 and 2, respectively.  The value of 0 was assigned when the predicted value was negative.
		
		The predicted values of AEXPG for the missing cases were calculated as:

			AEXPG=ROUND((.935971*AGE+.380251*SCHYR_C7+10.506319*HHSIZE+
    				10.381703*HHEARN+.004419*HHINC0+3.29783E-04*INC98P-
    				23.846325)*12)

		The regression coefficients used for the imputation here are from the estimates of an early run of the model and are slightly different from the corresponding estimated regression coefficents in the last run of the model in 3city_c10a.lis.  However, the differences are minor.  The information of the differeces is avilable from Porfessor Yu Xie.

		No missing code.

AEXPH   	ANNUAL HH EXPENSE ON ENTERTAINMENT-1999

		Range: 0 to 36000.
		This measure was obtained by multiplying the monthly household expense on entertainment (EXPNH) by 12.

			AEXPH=EXPNH*12		if EXPNH>=0
		
		When a respondent's EXPNH was missing (coded -999999 or -999998), the respondent's AEXPH value was obtained from the AEXPH value of his/her living together match pair.  There were 50 cases of them.  When the AEXPH value for the living together match pair was not available, regression method was used to impute the missing value of AEXPH.

		Age (AGE), years of schooling (SCHYR_C7), household size (HHSIZE), number of household members having income (HHEARN), household income (HHINC0) and personal income in 1998 (INC98P) were used to predict monthly household expense on entertainment.  The mode substitute was used to compute predicted values when the value of a predictor was missing.  The modes for HHSIZE and HHEARN were 3 and 2, respectively.  The value of 0 was assigned when the predicted value was negative.
		
		The predicted values of AEXPH for the missing cases were calculated as:

			AEXPH=ROUND((-.440478*AGE+1.299695*SCHYR_C7-4.218061*HHSIZE+
   				8.630224*HHEARN+.005739*HHINC0+.001330*INC98P+23.348276)*12)

		The regression coefficients used for the imputation here are from the estimates of an early run of the model and are slightly different from the corresponding estimated regression coefficents in the last run of the model in 3city_c10a.lis.  However, the differences are minor.  The information of the differeces is avilable from Porfessor Yu Xie.

		No missing code.

AEXPI   	ANNUAL HH EXPENSE ON OTHER-1999

		Range: 0 to 45600.
		This measure was obtained by multiplying the monthly household expense on other (EXPNI) by 12.

			AEXPI=EXPNI*12		if EXPNI>=0

		There were two missing cases on EXPNI (coded -999998).  Both of them MATCH=3 (i.e., single case without a matching pair).  The value 0 which is the mode of EXPNI was assigned to them.

		No missing code.

MEXPA   	MISSING INDICATOR:HH EXPENSE ON FOODS

		0, 1 dummy.
		Coded 1 if the data on monthly household expense on foods are missing, that is, if the data on annual household expense on foods were imputed; 0 otherwise.

MEXPB   	MISSING INDICATOR:HH EXPENSE ON HOUSING

		0, 1 dummy.
		Coded 1 if the data on monthly household expense on housing are missing, that is, if the data on annual household expense on housing were imputed; 0 otherwise.

MEXPC   	MISSING INDICATOR:HH EXPENSE ON UTILITIES

		0, 1 dummy.
		Coded 1 if the data on monthly household expense on utilities are missing, that is, if the data on annual household expense on utilities were imputed; 0 otherwise.

MEXPD   	MISSING INDICATOR:HH EXPENSE ON TEL/COMMUNICATION

		0, 1 dummy.
		Coded 1 if the data on monthly household expense on telephone and communication are missing, that is, if the data on annual household expense on telephone and communication were imputed; 0 otherwise.

MEXPE   	MISSING INDICATOR:HH EXPENSE ON TRANPORTATION

		0, 1 dummy.
		Coded 1 if the data on monthly household expense on transportation are missing, that is, if the data on annual household expense on transportation were imputed; 0 otherwise.

MEXPF   	MISSING INDICATOR:HH EXPENSE ON EDUCATION

		0, 1 dummy.
		Coded 1 if the data on monthly household expense on education are missing, that is, if the data on annual household expense on education were imputed; 0 otherwise.

MEXPG   	MISSING INDICATOR:HH EXPENSE ON MEDICATON

		0, 1 dummy.
		Coded 1 if the data on monthly household expense on medication are missing, that is, if the data on annual household expense on medication were imputed; 0 otherwise.

MEXPH   	MISSING INDICATOR:HH EXPENSE ON ENTERTAINMENT

		0, 1 dummy.
		Coded 1 if the data on monthly household expense on entertainment are missing, that is, if the data on annual household expense on entertainment were imputed; 0 otherwise.

MEXPI   	MISSING INDICATOR:HH EXPENSE ON OTHER

		0, 1 dummy.
		Coded 1 if the data on monthly household expense on other are missing, that is, if the data on annual household expense on other were imputed; 0 otherwise.

MA      	DATA ON FOOD EXPENSE FROM LIVED TOGETHER PAIR

		0, 1 dummy.
		Coded 1 if the data on annual household expense on foods was obtained from the living together match pair's data; 0 otherwise.

MB      	DATA ON HOUSING EXPENSE FROM LIVED TOGETHER PAIR

		0, 1 dummy.
		Coded 1 if the data on annual household expense on housing was obtained from the living together match pair's data; 0 otherwise.

MC      	DATA ON UTILITIES EXPENSE FROM LIVED TOGETHER PAIR

		0, 1 dummy.
		Coded 1 if the data on annual household expense on utilities was obtained from the living together match pair's data; 0 otherwise.

MD      	DATA ON TEL/COMMUNICATION EXPENSE FROM LIVED TOGETHER PAIR

		0, 1 dummy.
		Coded 1 if the data on annual household expense on telephone and communication was obtained from the living together match pair's data; 0 otherwise.

ME      	DATA ON TRANSPORTATION EXPENSE FROM LIVED TOGETHER PAIR

		0, 1 dummy.
		Coded 1 if the data on annual household expense on transportation was obtained from the living together match pair's data; 0 otherwise.

MF      	DATA ON EDUCATION EXPENSE FROM LIVED TOGETHER PAIR

		0, 1 dummy.
		Coded 1 if the data on annual household expense on education was obtained from the living together match pair's data; 0 otherwise.

MG      	DATA ON MEDICATION EXPENSE FROM LIVED TOGETHER PAIR

		0, 1 dummy.
		Coded 1 if the data on annual household expense on medication was obtained from the living together match pair's data; 0 otherwise.

MH      	DATA ON ENTERTAINMENT EXPENSE FROM LIVED TOGETHER PAIR

		0, 1 dummy.
		Coded 1 if the data on annual household expense on entertainment was obtained from the living together match pair's data; 0 otherwise.

ALLEXP99	1999 ANNUAL TOTAL HH EXPENSES

		Range: 1200 to 739020.
		This measure was obtained by summing the nine annual household expenses items.
		
			ALLEXP99=AEXPA+AEXPB+AEXPC+AEXPD+AEXPE+AEXPF+AEXPG+AEXPH+AEXPI
		
		No missing code.

MSEXP99 	MISSING INDEX:ANY HH EXPENSE ITEM MISSING

		0, 1 dummy.
		Coded 1 if data on any monthly household expense were missing; 0 otherwise.

NVEXP   	# OF VALID HH EXPENSE ITEMS

		Range: 1 to 9.
		Count the numbers of expense items the data of which were not missing.
		
		No missing code.

LTOG    	INTERGENERATION LIVE TOGETHER:PARTOG=1, K1TOG-K9TOG=1

		0, 1 dummy.
		Coded 1 if young respondents indicated living together with parents or elder respondents indicated living together with any kids; 0 otherwise.
 
RAGEK1  	ELDER: R'S AGE AT 1ST KID'S BIRTH

		Range: 11 to 60.
		This measure was obtained by subtracting the 1st child's age from respondent's age.
		
			RAGEK1=AGE-K1AGE

		-99: First child's age missing (K1AGE=-99)
		-98: First child's age DK (K1AGE=-98)
		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-86: Childless (NUMKIDS=0)
		-77: Error on first child's age

SUMKHP  	AMOUNT ALL ADULT KIDS HELP R (PARENT)

		Range: 0 to 35000.
		This measure was obtained by summing amounts of money that elderly respondent's all adult children (defined by age 18 or older) gave to the respondent.  Missing data on the amount of money giving were assumed to be 0.  When a offspring's age was missing, the child was not counted as an adult (i.e., implicitly the offspring was assumed to be a minor).

		-888888: Young respondent (FORM=1 or 4)
		-888887: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-888886: Childless (NUMKIDS=0) or no adult child (ASON=0 and ADAUG=0)

SUMPHK  	AMOUNT R (PARENT) HELP ALL ADULT KIDS

		Range: 0 to 21000.
		This measure was obtained by summing amounts of money that elderly respondent gave to all his/her adult children (defined by age 18 or older).  Missing data on the amount of money giving were assumed to be 0.  When a offspring's age was missing, the child was not counted as an adult (i.e., implicitly the offspring was assumed to be a minor).

		-888888: Young respondent (FORM=1 or 4)
		-888887: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-888886: Childless (NUMKIDS=0) or no adult child (ASON=0 and ADAUG=0)

SUMSHP  	AMOUNT ALL ADULT SONS HELP R (PARENT)

		Range: 0 to 20000.
		This measure was obtained by summing amounts of money that elderly respondent's all adult sons (defined by age 18 or older) gave to the respondent.  Missing data on the amount of money giving were assumed to be 0.  When a offspring's age or gender was missing, the offspring was not counted as an adult son.

		-888888: Young respondent (FORM=1 or 4)
		-888887: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-888886: Childless (NUMKIDS=0) or no adult son (ASON=0)

SUMDHP  	AMOUNT ALL ADULT DAUGHTERS HELP R (PARENT)

		Range: 0 to 35000.
		This measure was obtained by summing amounts of money that elderly respondent's all adult daughters (defined by age 18 or older) gave to the respondent.  Missing data on the amount of money giving were assumed to be 0.  When a offspring's age or gender was missing, the offspring was not counted as an adult daughter.

		-888888: Young respondent (FORM=1 or 4)
		-888887: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-888886: Childless (NUMKIDS=0) or no adult daughter (ADAUG=0)

SUMPHS  	AMOUNT R (PARENT) HELP ALL ADULT SONS

		Range: 0 to 21000.
		This measure was obtained by summing amounts of money that elderly respondent gave to all his/her adult sons (defined by age 18 or older).  Missing data on the amount of money giving were assumed to be 0.  When a offspring's age or gender was missing, the offspring was not counted as an adult son.

		-888888: Young respondent (FORM=1 or 4)
		-888887: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-888886: Childless (NUMKIDS=0) or no adult son (ASON=0)

SUMPHD  	AMOUNT R (PARENT) HELP ALL ADULT DAUGHTERS

		Range: 0 to 15000.
		This measure was obtained by summing amounts of money that elderly respondent gave to all his/her adult daughters (defined by age 18 or older).  Missing data on the amount of money giving were assumed to be 0.  When a offspring's age or gender was missing, the offspring was not counted as an adult daughter.

		-888888: Young respondent (FORM=1 or 4)
		-888887: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-888886: Childless (NUMKIDS=0) or no adult daughter (ADAUG=0)

SUMMSHP 	AMOUNT ALL MARRIED SONS HELP R (PARENT)

		Range: 0 to 20000.
		This measure was obtained by summing amounts of money that elderly respondent's all married sons gave to the respondent.  Missing data on the amount of money giving were assumed to be 0.  When a offspring's marital status or gender was missing, the offspring was not counted as a married son.

		-888888: Young respondent (FORM=1 or 4)
		-888887: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-888886: Childless (NUMKIDS=0) or no married son (MSON=0)

SUMMDHP 	AMOUNT ALL MARRIED DAUGHTERS HELP R (PARENT)

		Range: 0 to 35000.
		This measure was obtained by summing amounts of money that elderly respondent's all married daughters gave to the respondent.  Missing data on the amount of money giving were assumed to be 0.  When a offspring's marital status or gender was missing, the offspring was not counted as a married daughter.

		-888888: Young respondent (FORM=1 or 4)
		-888887: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-888886: Childless (NUMKIDS=0) or no married daughter (MDAUG=0)

SUMPHMS 	AMOUNT R (PARENT) HELP ALL MARRIED SONS

		Range: 0 to 10000.
		This measure was obtained by summing amounts of money that elderly respondent gave to all his/her married sons.  Missing data on the amount of money giving were assumed to be 0.  When a offspring's marital status or gender was missing,  the offspring was not counted as a married son.

		-888888: Young respondent (FORM=1 or 4)
		-888887: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-888886: Childless (NUMKIDS=0) or no married son (MSON=0)

SUMPHMD 	AMOUNT R (PARENT) HELP ALL MARRIED DAUGHTERS

		Range: 0 to 15000.
		This measure was obtained by summing amounts of money that an elderly respondent gave to all his/her married daughters.  Missing data on the amount of money giving were assumed to be 0.  When a offspring's marital status or gender was missing, the offspring was not counted as a married daughter.

		-888888: Young respondent (FORM=1 or 4)
		-888887: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)
		-888886: Childless (NUMKIDS=0) or no married daughter (MDAUG=0)

ASON    	# OF ADULT SONS:A+ OR B

		Range: 0 to 5.
		This measure was obtained by counting the number of sons age 18 or older from the elderly respondent's reports on Section B Q. 8.  The nth child was assumed not an adult son when the nth child's age or gender was missing.

		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)		

ADAUG   	# OF ADULT DAUGHTERS:A+ OR B

		Range: 0 to 7.
		This measure was obtained by counting the number of daughters age 18 or older from the elderly respondent's reports on Section B Q. 8.  The nth child was assumed not an adult daughter when the nth child's age or gender was missing.

		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)		

MSON    	# OF MARRIED SONS:A+ OR B

		Range: 0 to 5.
		This measure was obtained by counting the number of married sons from the elderly respondent's reports on Section B Q. 8.  The nth child was assumed not a married son when the nth child's marital status or gender was missing.

		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)		

MDAUG   	# OF MARRIED DAUGHTERS:A+ OR B

		Range: 0 to 7.
		This measure was obtained by counting the number of married daughters from the elderly respondent's reports on Section B Q. 8.  The nth child was assumed not a married daughter when the nth child's marital status or gender was missing.

		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)		

MINOR   	HAVING MINOR KIDS:A+ OR B

		0, 1 dummy
		Coded 1 if the elderly respondent reported any offspring's age younger than 18.

		-88: Young respondent (FORM=1 or 4)
		-87: Elderly respondent never married ((FORM=2 or 3) and MARITAL=4)		

WKTIMA  	WEEKDAY:# HRS WORK/DAY 1 IMP DECIMAL PT

		Range: 0 to 240
		This measure (numbers of hours worked per day on regular job during weekday with one implicit decimal point) was converted from information of hours and minutes worked per day on regular job during weekday (WKTIME1 and WKTIMM1, respectively).
		
			WKTIMA=ROUND(10*(WKTIME1+WKTIMM1/60))

		When minutes worked on regular job during weekday (WKTIMM1) was missing, WKTIMM1 was assumed to be 0 for the calculation.  When hours worked on regular job during weekday (WKTIME1) was missing, WKTIMA was assigned a missing code -997.
		
		-997: Missing data on hours worked on regular job during weekday (WKTIME1=-999)

WKTIMB  	WEEKEND:# HRS WORK/DAY 1 IMP DECIMAL PT

		Range: 0 to 240
		This measure (numbers of hours worked per day on regular job during weekend with one implicit decimal point) was converted from information of hours and minutes worked per day on regular job during weekend (WKTIME2 and WKTIMM2, respectively).
		
			WKTIMB=ROUND(10*(WKTIME2+WKTIMM2/60))

		When minutes worked on regular job during weekend (WKTIMM2) was missing, WKTIMM2 was assumed to be 0 for the calculation.  When hours worked on regular job during weekend (WKTIME2) was missing, WKTIMB was assigned a missing code -997.
		
		-997: Missing data on hours worked on regular job during weekend (WKTIME2=-999)

WKSTIMA 	WEEKDAY:# HRS WORK/DAY 2ND JOB 1 IMP DECIMAL PT

		Range: 0 to 200
		This measure (numbers of hours worked per day on supplemental job during weekday with one implicit decimal point) was converted from information of hours and minutes worked per day on supplemental job during weekday (WKSTIM1 and WKSTMM1, respectively).
		
			WKSTIMA=ROUND(10*(WKSTIM1+WKSTMM1/60))

		When minutes worked on supplemental job during weekday (WKSTMM1) was missing, WKSTMM1 was assumed to be 0 for the calculation.  When hours worked on supplemental job during weekday (WKSTIM1) was missing, WKSTIMA was assigned a missing code -997.
		
		-997: Missing data on hours worked on supplemental job during weekday (WKSTIM1=-999)

WKSTIMB 	WEEKEND:# HRS WORK/DAY 2ND JOB 1 IMP DECIMAL PT

		Range: 0 to 180
		This measure (numbers of hours worked per day on supplemental job during weekend with one implicit decimal point) was converted from information of hours and minutes worked per day on supplemental job during weekend (WKSTIM2 and WKSTMM2, respectively).
		
			WKSTIMB=ROUND(10*(WKSTIM2+WKSTMM2/60))

		When minutes worked on supplemental job during weekend (WKSTMM2) was missing, WKSTMM2 was assumed to be 0 for the calculation.  When hours worked on supplemental job during weekend (WKSTIM2) was missing, WKSTIMB was assigned a missing code -997.
		
		-997: Missing data on hours worked on supplemental job during weekend (WKSTIM2=-999)

HKTIMA  	WEEKDAY:# HRS HOUSEKEEPING/DAY 1 IMP DECIMAL PT

		Range: 0 to 240
		This measure (numbers of hours spent on housekeeping per day during weekday with one implicit decimal point) was converted from information of hours and minutes housekeeping per day during weekday (HKTIME1 and HKTIMM1, respectively).
		
			HKTIMA=ROUND(10*(HKTIME1+HKTIMM1/60))

		When minutes spent on housekeeping during weekday (HKTIMM1) was missing, HKTIMM1 was assumed to be 0 for the calculation.  When hours spent on housekeeping during weekday (HKTIME1) was missing, HKTIMA was assigned a missing code -997.
		
		-997: Missing data on hours spent on housekeeping during weekday (HKTIME1=-999)

HKTIMB  	WEEKEND:# HRS HOUSEKEEPING/DAY 1 IMP DECIMAL PT

		Range: 0 to 160
		This measure (numbers of hours spent on housekeeping per day during weekend with one implicit decimal point) was converted from information of hours and minutes housekeeping per day during weekday (HKTIME2 and HKTIMM2, respectively).
		
			HKTIMB=ROUND(10*(HKTIME2+HKTIMM2/60))

		When minutes spent on housekeeping during weekend (HKTIMM2) was missing, HKTIMM2 was assumed to be 0 for the calculation.  When hours spent on housekeeping during weekend (HKTIME2) was missing, HKTIMB was assigned a missing code -997.
		
		-997: Missing data on hours spent on housekeeping during weekend (HKTIME2=-999)

LSTIMA  	WEEKDAY:# HRS LEISURE/DAY 1 IMP DECIMAL PT

		Range: 0 to 200
		This measure (numbers of hours spent on entertainment per day during weekday with one implicit decimal point) was converted from information of total hours and minutes spent on entertainment during weekday (LSTIME1 and LSTIMM1, respectively).
		
			LSTIMA=ROUND(10*((LSTIME1+LSTIMM1/60)/5))

		When minutes spent on entertainment during weekday (LSTIMM1) was missing, LSTIMM1 was assumed to be 0 for the calculation.  When hours spent on entertainment during weekday (LSTIME1) was missing, LSTIMA was assigned a missing code -997.
		
		-997: Missing data on hours spent on entertainment during weekday (LSTIME1=-999)

LSTIMB  	WEEKEND:# HRS LEISURE/DAY 1 IMP DECIMAL PT

		Range: 0 to 225
		This measure (numbers of hours spent on entertainment per day during weekend with one implicit decimal point) was converted from information of total hours and minutes spent on entertainment during weekend (LSTIME2 and LSTIMM2, respectively).
		
			LSTIMB=ROUND(10*((LSTIME2+LSTIMM2/60)/2))

		When minutes spent on entertainment during weekend (LSTIMM2) was missing, LSTIMM2 was assumed to be 0 for the calculation.  When hours spent on entertainment during weekend (LSTIME2) was missing, LSTIMB was assigned a missing code -997.
		
		-997: Missing data on hours spent on entertainment during weekend (LSTIME2=-999)

MSRADIO 	MISSING INDEX:HOURS LISTEN RADIO=30 OR 40

		Range: 0, 1, 2
		There were 137 cases who reported listening to radio (rddays ne 0) for 30 or 40 hours per day (RDHOUR=30 or 40 and RDMIN=blank) on the day that they listened to radio. For these cases, the reported information on hours was assumed to be for minutes, and hour was to be 0.  MARADIO was coded 1 for those cases that RDHOUR was originally coded 30, coded 2 for those cases that RDHOUR was originally coded 40, coded 0 for those cases that either never listening to radio or with valid values of RDHOUR.

		-98: DK on hours listening to radio (RDHOUR=-998)
		-97: Missing on hours listening to radio (RDHOUR=-999)

SUMQUIZ 	SUM OF 14 QUIZ SCORES

		Range: 0 to 31.
		This measure was obtained by summing the 14 quiz scores.  For each quiz, the score was assumed to be 0 if the response of the quiz was missing including blank (-99), no answer (-86), or the quiz not asked (-87).
		
		No missing code.

MSQUIZ  	MISSING ANY QUIZ: 1=YES

		0, 1 dummy.
		Coded 1 if any response of the 14 quizzes was blank (-99), no answer (-86), or the quiz not asked (-87); 0 otherwise.
		
NPOSTAL 	POSTAL (ZIP) CODES:MISSING ASSIGNED

		Zip codes for Shanghai are in the series of 200000.
		Zip codes for Wuhan are in the series of 430000.
		Zip codes for Xian are in the series of 710000.
		http://www.travelchinaguide.com/essential/area_zip

		Range: 200001 to 710086.
		Assigned missing zip codes on POSTAL when the zip codes of the respondents' living together pairs were available.  There were 134 cases.  Among these 134 cases, 127 cases were matched with their pairs, 6 cases were not matched, and 1 case whose pair matching status was uncertain.
		
		-999999: Blank when response expected

		Note: There were many missing cases on POSTAL (1575 cases).  It seems unreasonable, but there is no document indicating why.  There were about this many cases with missing data on address or telephone in clean10.sav too.  Address and telephone were confidential string data and were deleted from the current data file.

MSPOSTAL	NPOSTAL DATA SOURCES INDICATOR

		Status for correction of out of range zip codes or assignment of missing zip codes for NPOSTAL.
		0: zip codes on NPOSTAL were from original POSTAL
		1: out of range zip codes on POSTAL were corrected
		2: zip codes were assigned from matched lived together pair's zip code
		3: zip codes were assigned from not matched lived together pair's zip code
		
		Out of range zip codes (18 cases) were assigned new values based on the zip codes of their pairs and living together status if possible and logical reasoning (see 3city_c10a.sps).  They were:
		
		id=13508892, POSTAL change from 300333 to 200333
		id=30201661, POSTAL change from 410015 to 430010
		id=30201741, POSTAL change from 400010 to 430010
		id=32907831, POSTAL change from 432051 to 430051
		id=34708573, POSTAL change from 400043 to 430043
		id=40505273, POSTAL change from 7100015 to 710015
		id=40604971, POSTAL change from 700058  to 710058
		id=40704141, POSTAL change from 7100322 to 710032
		id=41210051, POSTAL change from 610054  to 710054
		id=41810471, POSTAL change from 410043  to 710043
		id=43501953, POSTAL change from 730041  to 710041
		id=43607771, POSTAL change from 731003  to 710003
		id=43607802, POSTAL change from 731003  to 710003
		id=43607812, POSTAL change from 731003  to 710003
		id=44108344, POSTAL change from 7100160 to 710016
		id=44907154, POSTAL change from 720077  to 710077
		id=45006362, POSTAL change from 770077  to 710077
		id=46205921, POSTAL change from 740054  to 710054
		id=46205923, POSTAL change from -999999 to 710054
		
ITIME   	LENGTH OF INTERVIEW IN MINUTES

		Range: 8 to 380
		This measure was obtained by subtracting the starting interview time from the ending interview time when the ending interview time was later than the starting interview time.
		
			ITIME=(IEND1*60+IEND2)-(ISTART1*60+ISTART2)
		
		If minute of starting or ending interview time was missing, the interview time was implicitly assumed to be at the sharp of the hour (i.e., minutes were assumed to be 0).  If either one of the hour of the interview time was missing, ITIME was coded as missing (-997).  If the interview starting time was the same as or later than the ending time, the error code -777 was assigned to ITIME as well as the interview time ISTART1, ISTART2, IEND1, and IEND2.  If the interview starting and ending hours were the same and one of the interview minutes was missing, ITIME and the interview time were assigned the error code (-777).
		
		There were 49 cases that interview staring hour were later than the ending hour (i.e., START1 > END1).  Among them there were 13 cases that the ending hour was greater than 12 and 2 cases that the ending hour was 11.  No correction was made for these 15 cases, and the error code (-777) were assigned as indicated above.  
		
		However, there were 34 cases that the ending interview hour was 10 or less (i.e., END1 <= 10).  Thirty two out of these 34 cases, the length of interview time (ITIME) would be reasonable (range from 45 minutes to 153 minutes) when the reported ending hour was treated as PM and was converted to the 24 hour system (i.e., IEND1=IEND1+12).  For these 32 cases, ITIME was calculated as:
		
			IEND1=IEND1+12
			ITIME=(IEND1*60+IEND2)-(ISTART1*60+ISTART2)
		
		The other two cases of the 34 cases (ID=43001821 and ID=11704193), this fixing (i.e., treating the reported IEND1 as PM) did not work.  For ID=43001821, ISTART1=21 and IEND1=2.  The starting time was still later than the ending time when the ending hour was treated as PM.  For ID=11704193, ISTART1=7, ISTART2=0, IEND1=1, and IEND2=10.  If the ending hour was treated as PM and converted to 24 hour system, IEND1=13 and ITIME would be 370 minute, which was unreasonable long.  For these two cases, ISTART1, ISTART2, IEND1, IEND2, and ITIME were all coded errors (-777).

		Additionally, interview time were corrected for two cases that the interview length was extreme long.  These two cases were: ID=10902331 with ITIME=790 and ID=41102981 with ITIME=1170.  For ID=10902331, the interview time was ISTART1=7, ISTART2=5, IEND1=20, and IEND2=15.  The 7 hour of istart1=7 was treated as 7 P.M. and was converted to the 19 hour in 24 hour system.  As a result, the corrected ITIME=70. For ID=41102981, the interview time was ISTART1=2, ISTART2=10, IEND1=21, and IEND2=40.  The 2 hour of istart1=2 might be a typo error of the 20 hour.  When the ISTART1 was changed to 20, the corrected ITIME=90.  There were cases that interview length seemed to be too short (ex., 8 or 9 minutes) or too long (ex., 380 minutes), but no good logics to correct them.  For those cases, they were kept as they were.
					
		Furthermore, there were 104 cases that ISTART1 was between 1 and 6 with IEND1 between 1 and 8 and either equal to or greater than ISTART1.  It is not reasonable to conduct interviews between 1am and 6am so that the interview time (both ISTATR1 and IEND1) was treated as PM and converted to 24 hour system (i.e., ISTART1=ISTART1+12; IEND1=IEND1+12). This change would not affect ITIME at all.

		-997: Missing data on interview starting hour or ending hour (START1=-999 or END1=-999).
		-777: Interview starting time being the same as or later than the ending time.

MSITIME 	IMPUTED STATUS:LENGTH OF INTERVIEW TIME

		Range: 0-7
		Coded 1 when reported interview ending hour was treated as PM and was converted to 24 hour system (i.e., IEND1=IEND1+12);
		coded 2 when reported interview starting hour was treated as PM and was converted to 24 hour system (i.e., START1=START1+12);
		coded 3 when reported interview starting and ending hours were treated as PM and were converted to 24 hour system (i.e., START1=START1+12 and IEND1=IEND1+12);
		coded 4 when interview staring and ending minutes were both missing and were assumed to be 0 (i.e., change START2=-999 to START2=0 and END2=-999 to END2=0);
		coded 5 when only interview staring minutes was missing and was assumed to be 0 (i.e., change START2=-999 to START2=0);
		coded 6 when only interview ending minutes was missing and was assumed to be 0 (i.e., END2=-999 to END2=0);
		coded 7 when interview starting hour (ISTART1) was changed from 2 to 20 for ID=41102981 due to unreasonable length of interview time when ISTART1=2;
		coded 0 when data on reported interview time were available and were not changed.

		-97: Missing data on interview starting hour or ending hour (START1=-999 or END1=-999).
		-77: Interview starting time being the same as or later than the ending time.

-----------------------------------------------------------------------------------------------------
Variables YFILE to NASIS_Y were created by eldsup2.sps.  The output data file was eldsup2.sav, in which each serial number had one record of the data. The data were matched to each respondent's record by serial number in 3city_c10b.sps.  Thus, respondents with a same serial number had the same values for these variables.
-----------------------------------------------------------------------------------------------------
YFILE   	HAVING YOUNG RESPODENT DATA

		0, 1 dummy variable.  Coded 1 if there was a young respondent record for the serial number, 0 otherwise.
		
		This measure was obtained from the index of matching the young respondent records with the elderly respondent records by serial numbers
		
		No missing code		
		
YSKHP   	YOUNG:SELECT KID HELP PARENT MONEY

		The amount of the money the select child and the child's nuclear family helped parents in 1998: report from the offspring.

		Measured in yuans.  Range: 0-12000.
		Available for young respondents with match parents.
		This measure was obtained from recoding the variable YSUPAMT.
		
			YSKHP=0			if YSUPAMT=-888886 (no monetary support to parents)
			YSKHP=YSUPAMT		otherwise
		
		where YSUPAMT was SUPAMT from the respondents with FORM=1 or 4.

		The elderly respondent with the same serial number as the young respondent had the same values for this variable.

		-999999: Blank when response expected
		-999998: DK
		-888887: Both parents passed away
		-888886: No match parent for the serial number     
		-666666: No young respondent for the serial number

YSKHP0  	YOUNG:SK HELP PARENT MONEY 0-199=0

		The amount of the money the select child and the child's nuclear family helped parents in 1998 with the amount less than 200 to be coded as 0: report from the offspring.

		Measured in yuans.  Range: 0-12000.
		Available for young respondents with match parents.
		This measure was obtained from recoding YSUPAMT.  
		
			YSKHP0=0		if YSUPAMT=-888886 or 0-199
			YSKHP0=YSUPAMT		otherwise
			
		where YSUPAMT was SUPAMT from the respondents with FORM=1 or 4.

		The elderly respondent with the same serial number as the young respondent had the same values for this variable.

		-999999: Blank when response expected
		-999998: DK
		-888887: Both parents passed away
		-888886: No match parent for the serial number     
		-666666: No young respondent for the serial number

YPHSK   	YOUNG:PARENT HELP SELECT KID MONEY

		The amount of the money parents helped the select child and the child's nuclear family in 1998: report from the offspring.

		Measured in yuans.  Range: 0-30000.
		Available for young respondents with match parents.
		This measure was obtained from recoding the variable YPSUPAMT.  
				
			YPHSK=0			if YPSUPAMT=-888886
			YPHSK=YPSUPAMT		otherwise

		where YPSUPAMT was PSUPAMT from the respondents with FORM=1 or 4.						

		The elderly respondent with the same serial number as the young respondent had the same values for this variable.

		-999999: Blank when response expected
		-999998: DK
		-888887: Both parents passed away
		-888886: No matched parent for the serial number     
		-666666: No young respondent for the serial number

YPHSK0  	YOUNG:PARENT HELP SK MONEY 0-199=0

		The amount of the money parents helped the select child and the child's nuclear family in 1998 with the amount less than 200 to be coded as 0: report from the offspring.

		Measured in yuans.  Range: 0-30000.
		Available for young respondents with match parents.
		This measure was obtained from recoding the variable YPSUPAMT.  
						
			YPHSK0=0		if YPSUPAMT=-888886 or 0-199
			YPHSK0=YPSUPAMT		otherwise
								
		where YPSUPAMT was PSUPAMT from the respondents with FORM=1 or 4.

		The elderly respondent with the same serial number as the young respondent had the same values for this variable.

		-999999: Blank when response expected
		-999998: DK
		-888887: Both parents passed away
		-888886: No matched parent for the serial number     
		-666666: No young respondent for the serial number

YSKTOP  	YOUNG:NET AMOUNT MONEY SK HELP PARENT

		The net amount of the money the select child and the child's nuclear family helped parents in 1998 with the amount less than 200 to be coded as 0: report from the offspring.

		Measured in yuans.  Range: -30000-12000.
		Available for young respondents with match parents.
		This measure was obtained by subtracting the amount of the money the parents helped the select child and the child's nuclear family from the amount of the money the select child and the child's nuclear family helped the parents using the measures with 0-199 being coded as 0.
		
			YSKTOP=YSKHP0-YPHSK0

		The elderly respondent with the same serial number as the young respondent had the same values for this variable.

		-999997: Missing either YSKHP0 or YPHSK0
		-888887: Both parents passed away
		-888886: No match parent for the serial number     
		-666666: No young respondent for the serial number

YSKTOP3 	YOUNG:3 CATEGORIES OF NET TRANSFER

		Codes: 1, 2, 3.
		Available for the young respondents with the match parents.
		This measure was obtained from categorizing the net amount of the money that the select child and the child's nuclear family helped the parents in 1998 (YSKTOP): report from the offspring.
		
			YSKTOP3=1	if YSKHP0-YPHSK0 < 0
			YSKTOP3=2	if YSKHP0-YPHSK0 = 0
			YSKTOP3=3	if YSKHP0-YPHSK0 > 0

		The elderly respondent with the same serial number as the young respondent had the same values for this variable.

		-97: Missing either YSKHP0 or YPHSK0
		-87: Both parents passed away
		-86: No match parent for the serial number     
		-66: No young respondent for the serial number

YPTOG1  	YOUNG:LIVED WITH PARENT 1=YES 0=ELSE

		0, 1 dummy variable.  Coded 1 if the young respondent lived with a surviving parent in 1999 from the offspring's report, 0 otherwise.
		Available for young respondents.
		Recoded from YPARTOG.
		
			YPTOG1=1	if YPARTOG=1
			YPTOG1=0	if YPARTOG=2
		
		where YPARTOG was PARTOG from the respondents with FORM=1 or 4.

		The elderly respondent with the same serial number as the young respondent had the same values for this variable.

		-99: Blank when response expected
		-87: Both parents passed away
		-66: No young respondent for the serial number

YPCOOK1 	YOUNG:PARENT HELP COOK 1=YES 0=ELSE

		0, 1 dummy variable.  Coded 1 if parent helped the young respondent on cooking in 1999 from the offspring's report, 0 otherwise.
		Available for young respondents.		
		Recoded from YPARCOOK.
		
			YPCOOK1=1	if YPARCOOK=1 or 2
			YPCOOK1=0	if YPARCOOK=3
		
		where YPARCOOK was PARCOOK from the respondents with FORM=1 or 4.

		The elderly respondent with the same serial number as the young respondent had the same values for this variable.

		-99: Blank when response expected
		-87: Both parents passed away
		-66: No young respondent for the serial number

YPSHOP1 	YOUNG:PARENT HELP SHOP 1=YES 0=ELSE

		0, 1 dummy variable.  Coded 1 if parent helped the young respondent on shopping in 1999 from the offspring's report, 0 otherwise.
		Available for young respondents.		
		Recoded from YPARSHOP.
		
			YPSHOP1=1	if YPARSHOP=1 or 2
			YPSHOP1=0	if YPARSHOP=3
		
		where YPARSHOP was PARSHOP from the respondents with FORM=1 or 4.

		The elderly respondent with the same serial number as the young respondent had the same values for this variable.

		-99: Blank when response expected
		-87: Both parents passed away
		-66: No young respondent for the serial number

YPBSIT1 	YOUNG:PARENT HELP BABY SIT 1=YES 0=ELSE

		0, 1 dummy variable.  Coded 1 if parent helped on taking care of the young respondent's children in 1999 from the offspring's report, 0 otherwise.
		Available for young respondents.
		Recoded from YPARBSIT.

			YPBSIT1=1	if YPARBSIT=1 or 2
			YPBSIT1=0	if YPARBSIT=3
		
		where YPARBSIT was PARBSIT from the respondents with FORM=1 or 4.

		The elderly respondent with the same serial number as the young respondent had the same values for this variable.

		-99: Blank when response expected
		-87: Both parents passed away
		-85: Young respondent's had no children
		-66: No young respondent for the serial number

SK		ELDER:SELECT KID'S BIRTH ORDER (IMPUTED)

		Range: 1-8
		Available for the elderly respondents with any offspring.
		This measure was obtained from the birth order of the select offspring from the parent's report (SKIDORD).  For the missing in SKIDORD, it was obtained from the birth order of the offspring whose age was matched with the young respondent.
		
			SK=ESKIDORD	if ESKIDORD ge 1
			SK=the order of the elderly's child who had the same age with the young respondent	if ESKIDORD was blank or DK
		
		where ESKIDORD was SKIDORD from the respondents with FORM=2 or 3

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.
		
		-99: Blank when response expected
		-86: No kid for elderly primary respondents
		-66: No elderly respondent

EFILE   	HAVING ELDERLY RESPONDENT DATA

		0, 1 dummy variable.  Coded 1 if there was an elderly respondent record for the serial number, 0 otherwise.

		This measure was obtained from the index of matching the young respondent records with the elderly respondent records by serial numbers
				
		No missing code		
	
ESKHP   	ELDER:SELECT KID HELP PARENT MONEY

		The amount of the money that the select child and the child's nuclear family helped the parents in 1998: report from the parent.

		Measured in yuans.  Range: 0-10000.
		Available for the elderly respondents with match offspring.
		This measure was obtained from the SK_th child's data from the elderly respondent's report of the amount of the money that his/her each child and the child's nuclear family gave to the elderly respondent' and his/her spouse (i.e., the elderly respondent's reports on Section B Q. 8. for the column of financial aids from children-K1HLPR, K2HLPR etc.).  The SK refers to the select child's birth order.

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.

		-999999: Blank when response expected
		-999998: DK
		-888887: No children or never married
		-888886: No match child for the serial number     
		-666666: No elderly respondent for the serial number

ESKHP0  	ELDER:SK HELP PARENT MONEY 0-199=0

		The amount of the money that the select child and the child's nuclear family helped the parents in 1998 with 0-199 coded as 0: report from the parent.

		Measured in yuans.  Range: 0-10000, 0=0-199.
		Available for the elderly respondents with match offspring.
		This measure was obtained from recoding ESKHP.
		
			ESKHP0=0		if ESKHP=0-199
			ESKHP0=ESKHP		otherwise

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.
		
		-999999: Blank when response expected
		-999998: DK
		-888887: No children or never married
		-888886: No match child for the serial number     
		-666666: No elderly respondent for the serial number

EPHSK   	ELDER:PARENT HELP SELECT KID MONEY

		The amount of the money that parents helped the select child and the child's nuclear family in 1998: report from the parent.

		Measured in yuans.  Range: 0-20000.
		Available for the elderly respondents with match offspring.
		This measure was obtained from the SK_th child's data from the elderly respondent's report of the amount of the money that the elderly respondent and his/her spouse gave to their each adult child and the child's nuclear family (i.e., the elderly respondent's reports on Section B Q. 8. for the column of financial aids to children-K1RHLPK, K2RHLPK, etc.).  The SK refers to the select child's birth order.

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.

		-999999: Blank when response expected
		-999998: DK
		-888887: No children or never married
		-888886: No match child for the serial number     
		-666666: No elderly respondent for the serial number

EPHSK0  	ELDER:PARENT HELP SK MONEY 0-199=0

		The amount of the money that parents helped the select child and the child's nuclear family in 1998 with 0-199 coded as 0: report from the parent.

		Measured in yuans.  Range: 0-20000, 0=0-199.
		Available for the elderly respondents with match offspring.
		This measure was obtained from recoding ESKHP.
		
			EPHSK0=0		if EPHSK=0-199
			EPHSK0=EPHSK		otherwise

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.
		
		-999999: Blank when response expected
		-999998: DK
		-888887: No children or never married
		-888886: No match child for the serial number     
		-666666: No elderly respondent for the serial number

ESKTOP  	ELDER:NET AMOUNT MONEY SK HELP PARENT

		The net amount of the money the select child and the child's nuclear family helped parents in 1998 with the amount less than 200 to be coded as 0: report from the parent.

		Measured in yuans.  Range: -20000-10000.
		Available for elderly respondents with match offspring.
		This measure was obtained by subtracting the amount of the money the parents helped the select child and the child's nuclear family from the amount of the money the select child and the child's nuclear family helped the parents using the measures with 0-199 being coded as 0.
		
			ESKTOP=ESKHP0-EPHSK0

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.
		
		-999997: Missing either ESKHP0 or EPHSK0
		-888887: No children or never married
		-888886: No match child for the serial number     
		-666666: No elderly respondent for the serial number

ESKTOP3 	ELDER:3 CATEGORIES OF NET TRANSFER

		Codes: 1, 2, 3.
		Available for the elderly respondents with the match offspring.
		This measure was obtained from categorizing the net amount of the money that the select child and the child's nuclear family helped the parents in 1998 (ESKTOP): report from the parent.
		
			ESKTOP3=1	if ESKHP0-EPHSK0 < 0
			ESKTOP3=2	if ESKHP0-EPHSK0 = 0
			ESKTOP3=3	if ESKHP0-EPHSK0 > 0

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.
		
		-97: Missing either ESKHP0 or EPHSK0
		-87: No children or never married
		-86: No match child for the serial number     
		-66: No elderly respondent for the serial number

ESKTOG1 	ELDER:LIVED WITH SELECT CHILD 1=YES 0=ELSE

		0, 1 dummy variable.  Coded 1 if the elderly respondent lived with the select child in 1999 from the parent's report, 0 otherwise.
		Available for the elderly respondents.
		Recoded from ESKIDTOG.
		
			ESKTOG1=1	if ESKIDTOG=1
			ESKTOG1=0	if ESKIDTOG=2
		
		where ESKIDTOG was SKIDTOG from the respondents with FORM=1 or 4.

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.

		-99: Blank when response expected		
		-86: No child for the primary elderly respondent
		-85: Select child age younger than 18
		-66: No elderly respondent for the serial number

ESKCOOK1	ELDER:HELP SELECT CHILD COOK 1=YES 0=ELSE

		0, 1 dummy variable.  Coded 1 if the elderly respondent helped the select child in cooking in 1999 from the parent's report, 0 otherwise.
		Available for the elderly respondents.
		Recoded from ESKCOOK.
		
			ESKCOOK1=1	if ESKCOOK=1 or 2
			ESKCOOK1=0	if ESKCOOK=3
		
		where ESKCOOK was SKCOOK from the respondents with FORM=1 or 4.

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.

		-99: Blank when response expected		
		-86: No child for the primary elderly respondent
		-85: Select child age younger than 18
		-66: No elderly respondent for the serial number

ESKSHOP1	ELDER:HELP SELECT CHILD SHOP 1=YES 0=ELSE

		0, 1 dummy variable.  Coded 1 if the elderly respondent helped the select child in shopping in 1999 from the parent's report, 0 otherwise.
		Available for the elderly respondents.
		Recoded from ESKSHOP.
		
			ESKSHOP1=1	if ESKSHOP=1 or 2
			ESKSHOP1=0	if ESKSHOP=3
		
		where ESKSHOP was SKSHOP from the respondents with FORM=1 or 4.

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.

		-99: Blank when response expected		
		-86: No child for the primary elderly respondent
		-85: Select child age younger than 18
		-66: No elderly respondent for the serial number

ESKBSIT1	ELDER:HELP SELECT CHILD BABY SIT 1=YES 0=ELSE

		0, 1 dummy variable.  Coded 1 if the elderly respondent helped the select child in baby sitting in 1999 from the parent's report, 0 otherwise.
		Available for the elderly respondents.
		Recoded from ESKBSIT.
		
			ESKBSIT1=1	if ESKBBSIT=1 or 2
			ESKBSIT1=0	if ESKBBSIT=3
		
		where ESKBBSIT was SKBBSIT from the respondents with FORM=1 or 4.

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.

		-99: Blank when response expected		
		-86: No child for the primary elderly respondent
		-85: Select child age younger than 18
		-83: Select child had no children
		-66: No elderly respondent for the serial number

ESKED   	ELDER:SELECT KID EDUCATION YEARS

		Measure in years.  Range: 3-19
		
		Available for the elderly respondents with match offspring.
		This measure was obtained from the SK_th child's data from the elderly respondent's report of the education level that his/her each child graduated (i.e., the elderly respondent's reports on Section B Q. 8. for the column of education level).  The SK refers to the select child's birth order.  The education level was converted to years of education (see K1ED to K9ED).

		The young respondent with the same serial number as the elderly respondent had the same values for this variable.
		
		-99: Blank when response expected
		-86: Elderly respondent had never married
		-84: Elderly respondent had no match offspring
		-83: Elderly respondent had no pair; reason unknown
		-77: Problematic birth order for the select child
		-66: No elderly respondent for the serial number

ENSIB   	ELDER:# OF SK'S SIBLING

		Range: 0-8
		Available for the elderly respondents with match offspring.
		This measure was obtained from the elderly respondent's report of the number of his/her children alive subtracting 1.
		
			ENSIB=ENUMKIDS-1
			
		where ENUMKIDS was NUMKIDS from the respondents with FORM=1 or 4.	

		-86: Elderly respondent had never married
		-84: Elderly respondent had no match offspring
		-83: Elderly respondent had no pair; reason unknown
		-77: Problematic birth order for the select child
		-66: No elderly respondent for the serial number
		
ESIBMED 	ELDER:MEAN EDUCATION OF SK'S SIB 2 IMP PT

		Measure in hundredth years.  Range: 300-1900
		
		Available for the elderly respondents with match offspring.
		This measure was obtained from the elderly respondent's report of the education level that his/her each child graduated (i.e., the elderly respondent's reports on Section B Q. 8. for the column of education level). The children's education levels were converted to years of education (see K1ED to K9ED).  Each child's years of education other than the select child's were summed.  The sum was divided by the select child's number of sibling, and then multiplying by 100.  The values were rounded to the nearest integer.  A sibling whose years of education was missing would be excluded from the average.

			ESIBMED=round((SUMKED/SIB)*100)
			
		where SUMKED was the sum of the valid years of education for the select child's all sibling and SIB was the numbers of sibling who had data on years of education.
		
		The young respondent with the same serial number as the elderly respondent had the same values for this variable.
		
		-9997: education for the select child's sibling were all missing
		-8886: Elderly respondent had never married
		-8885: Elderly respondent had only 1 child
		-8884: Elderly respondent had no match offspring
		-8883: Elderly respondent had no pair; reason unknown
		-7777: Problematic birth order for the select child
		-6666: No elderly respondent for the serial number

NABRO_Y 	ELDER:# OF ADULT BROTHERS OF SELECT CHILD

		Range: 0-5
		Available for the elderly respondents with match offspring.
		This measure was obtained by subtracting 1 from EASON when the match young respondent was male or equating to EASON when the match young respondent was female, where EASON was the count of the number of sons age 18 or older from the elderly respondent's reports on Section B Q. 8.  The nth child was assumed not an adult son when the nth child's age or gender was missing (see ASON).
		
			NABRO_Y=EASON-1		if YMALE=1
			NABRO_Y=EASON		if YMALE=0		
			
		where EASON was ASON from the respondents with FORM=1 or 4.

		-87: Young respondent's both parents passed away
		-85: Elderly respondent had never married
		-84: Young respondent had no match parents
		-83: Young respondent had no pair; reason unknown
		-77: Inconsistent data: Young respondent's both parents passed away, but there was the elderly respondent data or the young respondent was male and the elderly respondent had no son
		-66: No young respondent for the serial number
	
NASIS_Y 	ELDER:# OF ADULT SISTERS OF SELECT CHILD

		Range: 0-6
		Available for the elderly respondents with match offspring.
		This measure was obtained by subtracting 1 from EADAUG when the match young respondent was female or equating to EADAUG when the match young respondent was male, where EADAUG was the count of the number of daughters age 18 or older from the elderly respondent's reports on Section B Q. 8.  The nth child was assumed not an adult daughter when the nth child's age or gender was missing (see ADAUG).
		
			NASIS_Y=EADAUG-1	if YMALE=0
			NASIS_Y=EADAUG		if YMALE=1		
			
		where EADAUG was ASON from the respondents with FORM=1 or 4.

		-87: Young respondent's both parents passed away
		-85: Elderly respondent had never married
		-84: Young respondent had no match parents
		-83: Young respondent had no pair; reason unknown
		-77: Inconsistent data: Young respondent's both parents passed away, but there was the elderly respondent data or the young respondent was female and the elderly respondent had no daughter
		-66: No young respondent for the serial number
	
-----------------------------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------------------
Variables A10D1 to AEDE50 were created by agesed.sps.  The output data file was agesed.sav, and the data were matched to each respondent's record by ID in 3city_c10b.sps.
-----------------------------------------------------------------------------------------------------

A10D1		CALENDAR YEAR LEFT PRIMARY SCHOOL-IMPUTED

		Range: 15-96
		
		This variable was obtained from A10_1D (Questionnaire Section A Q.10 for year graduated from primary school), and the year was imputed when the data were missing while the response was expected or when R attended primary school, but did not graduate.  Birth year (BYEAR), number of years stayed in primary school (A10_1B), and whether graduated from primary school (A10_1C) were used for the imputation.  Respondents were assumed to attend primary school at age 6.

			A10D1=BYEAR+6+A10_1B	if A10_1B ge 0 (i.e., not missing)
			A10D1=BYEAR+6+4		if A10_1B lt 0 (i.e., missing) and 
						   A10_1C ne 1 (i.e., not graduate)
			A10D1=BYEAR+6+6		if A10_1B lt 0 (i.e., missing) and 
						   A10_1C eq 1 (i.e., graduate)

		where BYEAR was the calendar year respondent was born, A10_1B was the number of years stayed at primary school, and A10_1C was whether graduated from primary school.

		-77: Problematic education history
		-87: Not attend primary school
		
A10D2		CALENDAR YEAR LEFT JUNIOR HIGH SCHOOL-IMPUTED

		Range: 18-99
		
		This variable was obtained from A10_2D (Questionnaire Section A Q.10 for year graduated from junior high school), and the year was imputed when the data were missing while the response was expected or when R attended junior high school, but did not graduate.  Calendar year left primary school (A10D1), number of years stayed in junior high school (A10_2B), and whether graduated from junior high school (A10_2C) were used for the imputation.

			A10D2=A10D1+A10_2B	if A10_2B ge 0 (i.e., not missing)
			A10D2=A10D1+2		if A10_2B lt 0 (i.e., missing) and 
						   A10_2C ne 1 (i.e., not graduate)
			A10D2=A10D1+3		if A10_2B lt 0 (i.e., missing) and 
						   A10_2C eq 1 (i.e., graduate)

		where A10D1 was the calendar year respondent left primary school with the imputation being described above, A10_2B was the number of years stayed at junior high school, and A10_2C was whether graduated from junior high school.

		There was one case (ID=30201752) in which respondent attended junior high school without attending primary school and did not graduate from junior high school.  As a result, A10D2 needs to be imputed while A10D1 was coded -87.  This respondent graduated from secondary technical high school in 1951 at age 18.  This respondent's years stayed at junior high school was missing too.  Thus, her A10D2 was imputed as BYEAR+12+2.
		
		-77: Problematic education history
		-87: Not attend junior high school

A10D3		CALENDAR YEAR LEFT SENIOR HIGH SCHOOL-IMPUTED

		Range: 29-99
		
		This variable was obtained from A10_3D (Questionnaire Section A Q.10 for year graduated from senior high school), and the year was imputed when the data were missing while the response was expected or when R attended senior high school, but did not graduate.  Calendar year left junior high school (A10D2), number of years stayed in senior high school (A10_3B), and whether graduated from senior high school (A10_3C) were used for the imputation.

			A10D3=A10D2+A10_3B	if A10_3B ge 0 (i.e., not missing)
			A10D3=A10D2+2		if A10_3B lt 0 (i.e., missing) and 
						   A10_3C ne 1 (i.e., not graduate)
			A10D3=A10D2+3		if A10_3B lt 0 (i.e., missing) and 
						   A10_3C eq 1 (i.e., graduate)

		where A10D2 was the calendar year respondent left junior high school with the imputation being described above, A10_3B was the number of years stayed at senior high school, and A10_3C was whether graduated from senior high school.

		-77: Problematic education history
		-87: Not attend senior high school

A10D4		CALENDAR YEAR LEFT SECONDARY TECH SCHOOL-IMPUTED

		Range: 21-99
		
		This variable was obtained from A10_4D (Questionnaire Section A Q.10 for year graduated from secondary technical school), and the year was imputed when the data were missing while the response was expected or when R attended secondary technical school, but did not graduate.  Calendar year left junior high school (A10D2) or senior high school (A10D3) depending on whether attending senior high school (A10_3A), number of years stayed in secondary technical school (A10_4B), and whether graduated from secondary technical school (A10_4C) were used for the imputation.

			A10D4=A10D2+A10_4B	if A10_3A=2 & A10_4B ge 0 (i.e., not missing)
			A10D4=A10D3+A10_4B	if A10_3A=1 & A10_4B ge 0 (i.e., not missing)
			A10D4=A10D2+2		if A10_3A=2 & A10_4B lt 0 (i.e., missing) & 
						   A10_4C ne 1 (i.e., not graduate)
			A10D4=A10D3+2		if A10_3A=1 & A10_4B lt 0 (i.e., missing) & 
						   A10_4C ne 1 (i.e., not graduate)
			A10D4=A10D2+3		if A10_3A=2 & A10_4B lt 0 (i.e., missing) & 
						   A10_4C eq 1 (i.e., graduate)
			A10D4=A10D3+3		if A10_3A=1 & A10_4B lt 0 (i.e., missing) & 
						   A10_4C eq 1 (i.e., graduate)

		where A10D2/A10D3 was the calendar year respondent left junior/senior high school with the imputation being described above, A10_3A was whether respondent attended senior high school, A10_4B was the number of years stayed at secondary technical school, and A10_4C was whether graduated from secondary technical school.

		-77: Problematic education history
		-87: Not attend secondary technical school

A10D5		CALENDAR YEAR LEFT JUNIOR COLLEGE-IMPUTED

		Range: 42-99
		
		This variable was obtained from A10_5D (Questionnaire Section A Q.10 for year graduated from junior college), and the year was imputed when the data were missing while the response was expected or when R attended junior college, but did not graduate.  The imputation procedure to impute missing calendar years left junior college was complicate because students could enter junior college via various channels. It depended on whether respondents directly went to junior college from junior high school or whether attended senior high school or secondary technical school before entering junior college.  The information whether respondent was a student at the survey time (i.e., RNOWORK=4) was also used for the imputation.

		If respondent attended junior college without attending senior high school or secondary technical school (i.e., JHJC=1),
		
			A10D5=A10D2+A10_5B.
			
		There were 6 cases in this situation. A10_5B were all available for these six cases that A10D5 needs to be imputed with JHJC=1.

		If respondent attended senior high school or secondary technical school before attended junior college (i.e., A10_5A=1 & JHJC=0),
		
			A10D5=A10D3+A10_5B	if A10D3 gt 0 & A10_5B ge 0 (i.e., not missing)
			A10D5=A10D4+A10_5B	if A10D4 gt 0 & A10_5B ge 0 (i.e., not missing)
			A10D5=A10D3+1		if A10D3 gt 0 & A10_5B lt 0 (i.e., missing) & 
						   A10_5C ne 1 (i.e., not graduate)
			A10D5=A10D4+1		if A10D4 gt 0 & A10_5B lt 0 (i.e., missing) & 
						   A10_5C ne 1 (i.e., not graduate)
			A10D5=A10D3+3		if A10D3 gt 0 & A10_5B lt 0 (i.e., missing) & 
						   A10_5C eq 1 (i.e., graduate)
			A10D5=A10D4+3		if A10D4 gt 0 & A10_5B lt 0 (i.e., missing) & 
						   A10_5C eq 1 (i.e., graduate)

		where A10D3/A10D4 was the calendar year respondent left senior high school/secondary technical school with the imputation being described above, A10_5B was the number of years stayed at junior college, and A10_5C was whether graduated from junior college.

		For respondents who were students at the survey time (i.e., RNOWORK=4), did not attend university (A10_6A=2), and attended junior college without graduating (A10_5A=1 and A10_5C=2), A10D5=99.  For these respondents, A10D5=99 was not a guess so that I10D5 (imputing status) was equal to 0.
		
		-77: Problematic education history
		-87: Not attend junior college

A10D6		CALENDAR YEAR LEFT UNIVERISYT-IMPUTED

		Range: 36-99

		This variable was obtained from A10_6D (Questionnaire Section A Q.10 for year graduated from university), and the year was imputed when the data were missing while the response was expected or when R attended university, but did not graduate.  The imputation procedure to impute missing calendar years left university was complicate  because students could enter university via various channels. It depended on whether respondents directly went to university from junior high school or what types school university attendees attended before entering university.  The information whether respondent was a student at the survey time (i.e., RNOWORK=4) was also used for the imputation.

		If respondents did not directly attend university from junior high school graduates (i.e., JHU=0)
		
			A10D6=A10D5+A10_6B	if A10D5 gt 0 & A10_6B ge 0 (i.e., not missing)
			A10D6=A10D4+A10_6B	if A10D4 gt 0 & A10_6B ge 0 (i.e., not missing)
			A10D6=A10D3+A10_6B	if A10D3 gt 0 & A10_6B ge 0 (i.e., not missing)
			A10D6=A10D5+1		if A10D5 gt 0 & A10_6B lt 0 (i.e., missing) & 
						   A10_6C ne 1 (i.e., not graduate)
			A10D6=A10D4+2		if A10D4 gt 0 & A10_6B lt 0 (i.e., missing) & 
						   A10_6C ne 1 (i.e., not graduate)
			A10D6=A10D3+2		if A10D3 gt 0 & A10_6B lt 0 (i.e., missing) & 
						   A10_6C ne 1 (i.e., not graduate)
			A10D6=A10D5+2		if A10D5 gt 0 & A10_6B lt 0 (i.e., missing) & 
						   A10_6C eq 1 (i.e., graduate)
			A10D6=A10D4+4		if A10D4 gt 0 & A10_6B lt 0 (i.e., missing) & 
						   A10_6C eq 1 (i.e., graduate)
			A10D6=A10D3+4		if A10D3 gt 0 & A10_6B lt 0 (i.e., missing) & 
						   A10_6C eq 1 (i.e., graduate)

		If respondents directly attended university from junior high school graduates (i.e., JHU=1), the imputation of A10D6 was:
		
			A10D6=A10D2+A10_6B

		There was 1 case in this situation, and A10_6B was not missing.
		
		For respondents who were students at the survey time (i.e., RNOWORK=4), did not attend graduate school (A10_7A=2), and attended university without graduating (A10_6A=1 and A10_6C=2), A10D6=99.  For these respondents, A10D6=99 was not a guess so that I10D6 (imputing status) was equal to 0.
		
		-77: Problematic education history
		-87: Not attend university

A10D7		CALENDAR YEAR LEFT GRADUATE SCHOOL-IMPUTED

		Range: 53-99

		This variable was obtained from A10_7D (Questionnaire Section A Q.10 for year graduated from graduate school), and the year was imputed when the data were missing while the response was expected or when R attended graduate school, but did not graduate.  The information on the calendar year left university (A10D6), the number of years stayed at graduate school (A10_7B), and whether graduated from graduate school (A10_7C) was used for the imputation.  The information whether respondent was a student at the survey time (i.e., RNOWORK=4) was also used.

			A10D7=A10D6+A10_7B	if A10D6 gt 0 & A10_7B ge 0 (i.e., not missing)
			A10D7=A10D6+2		if A10D6 gt 0 & A10_7B lt 0 (i.e., missing) & 
						   A10_7C ne 1 (i.e., not graduate)
			A10D7=A10D6+3		if A10D6 gt 0 & A10_7B lt 0 (i.e., missing) & 
						   A10_7C eq 1 (i.e., graduate)

		For respondents who were students at the survey time (i.e., RNOWORK=4) and attended graduate school without graduating (A10_7A=1 and A10_7C=2), A10D7=99.  For these respondents, A10D7=99 was not a guess so that I10D7 (imputing status) was equal to 0.
		
		-77: Problematic education history
		-87: Not attend university

I10D1		IMPUTE YEAR LEFT PRIMARY SCHOOL

		0, 1 dummy variable.  Coded 1 if the calendar year left primary school that was used to compute age specific education (i.e., A10D1) was imputed, 0 otherwise.

		-87: Not attend primary school

I10D2		IMPUTE YEAR LEFT JUNIOR HIGH SCHOOL

		0, 1 dummy variable.  Coded 1 if the calendar year left junior high school that was used to compute age specific education (i.e., A10D2) was imputed, 0 otherwise.

		-87: Not attend junior high school

I10D3		IMPUTE YEAR LEFT SENIOR HIGH SCHOOL

		0, 1 dummy variable.  Coded 1 if the calendar year left senior high school that was used to compute age specific education (i.e., A10D3) was imputed, 0 otherwise.

		-87: Not attend senior high school

I10D4		IMPUTE YEAR LEFT TECH HIGH SCHOOL

		0, 1 dummy variable.  Coded 1 if the calendar year left technical high school (i.e., secondary technical school) that was used to compute age specific education (i.e., A10D4) was imputed, 0 otherwise.

		-87: Not attend technical high school (i.e., secondary technical school)

I10D5		IMPUTE YEAR LEFT JUNIOR COLLEGE

		0, 1 dummy variable.  Coded 1 if the calendar year left junior college that was used to compute age specific education (i.e., A10D5) was imputed, 0 otherwise.

		-87: Not attend junior college

I10D6		IMPUTE YEAR LEFT UNIVERSITY

		0, 1 dummy variable.  Coded 1 if the calendar year left university that was used to compute age specific education (i.e., A10D6) was imputed, 0 otherwise.

		-87: Not attend university

I10D7		IMPUTE YEAR LEFT GRADUATE SCHOOL

		0, 1 dummy variable.  Coded 1 if the calendar year left graduate school that was used to compute age specific education (i.e., A10D7) was imputed, 0 otherwise.

		-87: Not attend graduate school

*****************************************************************************************************
AEDYS14 to AEDYS50:
		The information from the education history (Questionnair Section A Q10) was used to construct age specific number of years stayed in school for age 14 upto age 50.  The procedure to construct these variables (AEDYS14 to AEDYS50) was (see agesed.sps):

		Initialize the variables to -96 so that the value for the ages earlier than the age attended primary school would be -96.  Then, at each education level starting from the lowest level primary school to the highest level graduate school,
		
		1) Sum up the years stayed in school from primary school upto the given education level.
		
		2) Compute the age left the given education level by subtracting the calendar year 
		left the given education level (A10D1 to A10D7) from the birth year (BYEAR).  For those who attended a given education and not graduate, the calendar year left school was estimated from the calendar year left the previous education level and the number of years stayed at the given education level (see A10D1 to A10D7).

		3) Assign the total years in school from the step 1 to the age left the given education level and the ages onward up to age 50.

		4) Assign 1 year less in school for each age back from the age left the given education level for as many years back as the number of the year stayed in the given education level minus 1.
		
		For example, at the primary school level if a respondent stayed at primary school for 6 years (A10_1B) and left the school at age 12 (A10D1-BYEAR), then this respondent would have 6 years in school for ages 12 and onward and would have 5 years in school at age 11, 4 years in school at age 10, ..., and 1 year in school at age 7 (12-5=7).  Furthermore, at junior high school level this respondent stayed at junior high school for 3 years (A10_2B) and left junior high school at age 17 (A10D2-BYEAR), this respondent would have 9 (6+3) years in school for ages 17 and onward and would have 8 years in school at age 16 and 7 years in school at age 15.  If no further education for this reppondent, this respondent's age-specific years in school would be 6 years at age 14, 7 years at age 15, 8 years at age 16, and 9 years at age 17, age 18, up to age 50.

		For those who did not attend a given education level, 0 years in school was assigned to the number of years stayed at that education level.  For those who attended a given education level, but the number of the years stayed at that education level was missing, the norm of years to graduate the education level was assigned if graduated and the mode for years stayed at that education level among non-graduates was assigned if not graduate.

****************************************************************************************************
AEDYS14 	YEARS IN SCHOOL:AGE 14

		Measure in years.  Range: 0-10.

		-97: All calendar years left school missing
		-96: Attended primary school after age 14
		-77: Problematic education history
		
AEDYS15 	YEARS IN SCHOOL:AGE 15

		Measure in years.  Range: 0-11.

		-97: All calendar years left school missing
		-96: Attended primary school after age 15
		-77: Problematic education history
	
AEDYS16 	YEARS IN SCHOOL:AGE 16

		Measure in years.  Range: 0-12.

		-97: All calendar years left school missing
		-96: Attended primary school after age 16
		-77: Problematic education history
	
AEDYS17 	YEARS IN SCHOOL:AGE 17

		Measure in years.  Range: 0-13.

		-97: All calendar years left school missing
		-96: Attended primary school after age 17
		-77: Problematic education history
	
AEDYS18 	YEARS IN SCHOOL:AGE 18

		Measure in years.  Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 18
		-77: Problematic education history
	
AEDYS19 	YEARS IN SCHOOL:AGE 19

		Measure in years.  Range: 0-15.

		-97: All calendar years left school missing
		-96: Attended primary school after age 19
		-77: Problematic education history
	
AEDYS20 	YEARS IN SCHOOL:AGE 20

		Measure in years.  Range: 0-16.

		-97: All calendar years left school missing
		-96: Attended primary school after age 20
		-77: Problematic education history
	
AEDYS21 	YEARS IN SCHOOL:AGE 21

		Measure in years.  Range: 0-17.

		-97: All calendar years left school missing
		-96: Attended primary school after age 21
		-77: Problematic education history
	
AEDYS22 	YEARS IN SCHOOL:AGE 22

		Measure in years.  Range: 0-17.

		-97: All calendar years left school missing
		-96: Attended primary school after age 22
		-77: Problematic education history
	
AEDYS23 	YEARS IN SCHOOL:AGE 23

		Measure in years.  Range: 0-17.

		-97: All calendar years left school missing
		-96: Attended primary school after age 23
		-77: Problematic education history
	
AEDYS24 	YEARS IN SCHOOL:AGE 24

		Measure in years.  Range: 0-18.

		-97: All calendar years left school missing
		-96: Attended primary school after age 24
		-77: Problematic education history
	
AEDYS25 	YEARS IN SCHOOL:AGE 25

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 25
		-77: Problematic education history
	
AEDYS26 	YEARS IN SCHOOL:AGE 26

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-96: Attended primary school after age 26
		-77: Problematic education history
	
AEDYS27 	YEARS IN SCHOOL:AGE 27

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-96: Attended primary school after age 27
		-77: Problematic education history
	
AEDYS28 	YEARS IN SCHOOL:AGE 28

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-96: Attended primary school after age 28
		-77: Problematic education history
	
AEDYS29 	YEARS IN SCHOOL:AGE 29

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-96: Attended primary school after age 29
		-77: Problematic education history
	
AEDYS30 	YEARS IN SCHOOL:AGE 30

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-96: Attended primary school after age 30
		-77: Problematic education history
	
AEDYS31 	YEARS IN SCHOOL:AGE 31

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-96: Attended primary school after age 31
		-77: Problematic education history
	
AEDYS32 	YEARS IN SCHOOL:AGE 32

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-96: Attended primary school after age 32
		-77: Problematic education history
	
AEDYS33 	YEARS IN SCHOOL:AGE 33

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-96: Attended primary school after age 33
		-77: Problematic education history
	
AEDYS34 	YEARS IN SCHOOL:AGE 34

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-96: Attended primary school after age 34
		-77: Problematic education history
	
AEDYS35 	YEARS IN SCHOOL:AGE 35

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-96: Attended primary school after age 35
		-77: Problematic education history
	
AEDYS36 	YEARS IN SCHOOL:AGE 36

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-96: Attended primary school after age 36
		-77: Problematic education history
	
AEDYS37 	YEARS IN SCHOOL:AGE 37

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS38 	YEARS IN SCHOOL:AGE 38

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS39 	YEARS IN SCHOOL:AGE 39

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS40 	YEARS IN SCHOOL:AGE 40

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS41 	YEARS IN SCHOOL:AGE 41

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS42 	YEARS IN SCHOOL:AGE 42

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS43 	YEARS IN SCHOOL:AGE 43

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS44 	YEARS OF SCHOOL:AGE 44

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS45 	YEARS OF SCHOOL:AGE 45

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS46 	YEARS OF SCHOOL:AGE 46

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS47 	YEARS OF SCHOOL:AGE 47

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS48 	YEARS OF SCHOOL:AGE 48

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS49 	YEARS OF SCHOOL:AGE 49

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDYS50 	YEARS OF SCHOOL:AGE 50

		Measure in years.  Range: 0-20.

		-97: All calendar years left school missing
		-77: Problematic education history
****************************************************************************************************
AEDY14 to AEDY50:
		The information from the education history (Questionnair Section A Q10) was used to construct age specific years of education for age 14 upto age 50.  The procedure to construct these variables (AEDY14 to AEDY50) was (see agesed.sps):

		Initialize the variables to -96 so that the value for the ages earlier than the age attended primary school would be -96.  Then, at each education level starting from the lowest level primary school to the highest level graduate school,

		1) Assign the equivalent year of education for each education level distinguishing between those who graduated and not graduate from the given level.  Specifically, the assigned the number of years of education for each level was:

				level			not graduate	graduated
			no schooling at all		0 year
			attended primary school		4 years		6 years
			attended junior high school	8 years		9 years
			attended senior high school	11 years	12 years
			attended secondary tech school	11 years	12 years
			attended junior college		13 years	15 years
			attended university		14 years	16 years
			attended graduate school	18 years	19 years
		
		2) Compute the age left the given education level by subtracting the calendar year left the given education level (A10D1 to A10D7) from the birth year (BYEAR).  For those who attended a given education and not graduate, the calendar year left school was estimated from the calendar year left the previous education level and the number of years stayed at the given education level (see A10D1 to A10D7).

		3) Assign the number of years of education for the given education level shown on the step 1 to the age left the given education level and the ages onward up to age 50.

		4) Assign the year of education from the column of not graduate on the step 1 to each age back from the age left the given education level for as many years back as the number of the year stayed in the given education level minus 1. 
		
		For example, at the primary school level if a respondent stayed at school for 6 years and graduated at age 12 (A10D1-BYEAR), then this respondent would have 6 years of education for age 12 and onward and would have 4 years of education for ages between 11 and  7 (12-(6-1)=7).  Furthermore, at junior high school level this respondent stayed at junior high school for 3 years (A10_2B) and graduated at age 17 (A10D2-BYEAR), this respondent would have 9 years of education for age 17 and onward and would have 8 years of education at ages 16 and 15.  If no further education for this reppondent, this respondent's age-specific years of education would be 6 years of education at age 14, 8 years at ages 15 and 16, and 9 years of education at age 17, age 18, up to age 50.

		Note: There were 9 cases that the respondent graduated from senior high school, which would have 12 years of education, and attended secondary technical school without graduate, which would have 11 years of education. For these 9 cases, the age-specific years of education were kept at 12 years for the ages at secondary technical school and onward until the age attened the next education level to ensure the monotonous increase for age-specific years of education.

		Similarily, there were 2 cases that the respondent graduated from junior college, which would have 15 years of education, and attended university without graduate, which would have 14 years of education.  For these 2 cases, the age-specific years of education were kept at 15 for the ages at university and onward to ensure the monotonous increase for age-specific years of education.

*****************************************************************************************************
	
AEDY14  	YEARS OF EDUCATION:AGE 14

		Measure in years.  Range: 0-13.

		-97: All calendar years left school missing
		-96: Attended primary school after age 14
		-77: Problematic education history
	
AEDY15  	YEARS OF EDUCATION:AGE 15

		Measure in years.  Range: 0-13.

		-97: All calendar years left school missing
		-96: Attended primary school after age 15
		-77: Problematic education history
	
AEDY16  	YEARS OF EDUCATION:AGE 16

		Measure in years.  Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 16
		-77: Problematic education history
	
AEDY17  	YEARS OF EDUCATION:AGE 17

		Measure in years.  Range: 0-15.

		-97: All calendar years left school missing
		-96: Attended primary school after age 17
		-77: Problematic education history
	
AEDY18  	YEARS OF EDUCATION:AGE 18

		Measure in years.  Range: 0-15.

		-97: All calendar years left school missing
		-96: Attended primary school after age 18
		-77: Problematic education history
	
AEDY19  	YEARS OF EDUCATION:AGE 19

		Measure in years.  Range: 0-16.

		-97: All calendar years left school missing
		-96: Attended primary school after age 19
		-77: Problematic education history
	
AEDY20  	YEARS OF EDUCATION:AGE 20

		Measure in years.  Range: 0-16.

		-97: All calendar years left school missing
		-96: Attended primary school after age 20
		-77: Problematic education history
	
AEDY21  	YEARS OF EDUCATION:AGE 21

		Measure in years.  Range: 0-16.

		-97: All calendar years left school missing
		-96: Attended primary school after age 21
		-77: Problematic education history
	
AEDY22  	YEARS OF EDUCATION:AGE 22

		Measure in years.  Range: 0-18.

		-97: All calendar years left school missing
		-96: Attended primary school after age 22
		-77: Problematic education history
	
AEDY23  	YEARS OF EDUCATION:AGE 23

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 23
		-77: Problematic education history
	
AEDY24  	YEARS OF EDUCATION:AGE 24

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 24
		-77: Problematic education history
	
AEDY25  	YEARS OF EDUCATION:AGE 25

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 25
		-77: Problematic education history
	
AEDY26  	YEARS OF EDUCATION:AGE 26

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 26
		-77: Problematic education history
	
AEDY27  	YEARS OF EDUCATION:AGE 27

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 27
		-77: Problematic education history
	
AEDY28  	YEARS OF EDUCATION:AGE 28

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 28
		-77: Problematic education history
	
AEDY29  	YEARS OF EDUCATION:AGE 29

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 29
		-77: Problematic education history
	
AEDY30  	YEARS OF EDUCATION:AGE 30

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 30
		-77: Problematic education history
	
AEDY31  	YEARS OF EDUCATION:AGE 31

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 31
		-77: Problematic education history
	
AEDY32  	YEARS OF EDUCATION:AGE 32

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 32
		-77: Problematic education history
	
AEDY33  	YEARS OF EDUCATION:AGE 33

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 33
		-77: Problematic education history
	
AEDY34  	YEARS OF EDUCATION:AGE 34

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 34
		-77: Problematic education history
	
AEDY35  	YEARS OF EDUCATION:AGE 35

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 35
		-77: Problematic education history
	
AEDY36  	YEARS OF EDUCATION:AGE 36

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-96: Attended primary school after age 36
		-77: Problematic education history
	
AEDY37  	YEARS OF EDUCATION:AGE 37

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY38  	YEARS OF EDUCATION:AGE 38

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY39  	YEARS OF EDUCATION:AGE 39

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY40  	YEARS OF EDUCATION:AGE 40

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY41  	YEARS OF EDUCATION:AGE 41

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY42  	YEARS OF EDUCATION:AGE 42

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY43  	YEARS OF EDUCATION:AGE 43

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY44  	YEARS OF EDUCATION:AGE 44

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY45  	YEARS OF EDUCATION:AGE 45

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY46  	YEARS OF EDUCATION:AGE 46

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY47  	YEARS OF EDUCATION:AGE 47

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY48  	YEARS OF EDUCATION:AGE 48

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY49  	YEARS OF EDUCATION:AGE 49

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDY50  	YEARS OF EDUCATION:AGE 50

		Measure in years.  Range: 0-19.

		-97: All calendar years left school missing
		-77: Problematic education history
	
****************************************************************************************************
AEDL14 to AEDL50:
		The information from the education history (Questionnair Section A Q10) was used to construct age specific education level for age 14 upto age 50.  The procedure to construct these variables (AEDL14 to AEDL50) was (see agesed.sps):

		Initialize the variables to -96 so that the value for the ages earlier than the age attended primary school would be -96.  Then, at each education level starting from the lowest level primary school to the highest level graduate school,

		1) Assign a hierarchical level for each education level distinguishing between those who graduated and not graduated from the given level.  Specifically, the assigned hierarchical level for each education level was:

				level			not graduate	graduated
			no schooling at all		0
			attended primary school		1		2
			attended junior high school	3		4
			attended senior high school	5		6
			attended secondary tech school	7		8
			attended junior college		9		10
			attended university		11		12
			attended graduate school	13		14
		
		2) Compute the age left the given education level by subtracting the calendar year left the given education level (A10D1 to A10D7) from the birth year (BYEAR).  For those who attended a given education and not graduate, the calendar year left school was estimated from the calendar year left the previous education level and the number of years stayed at the given education level (see A10D1 to A10D7).

		3) Assign the hierarchical level for the given education level shown on the step 1 to the age left the given education level and the ages onward upto 50.

		4) Assign the level of not graduate to each age back from the age left the given education level for as many years back as the number of the year stayed in the given education level minus 1. 
		
		For example, at the primary school level if a respondent stayed at primary school for 6 years and graduated at age 12 (A10D1-BYEAR), then this respondent would have 2 for the education level at ages 12 and onward and would have 1 for ages between 11 and  7 (12-(6-1)=7).  Furthermore, at junior high school level if this respondent stayed at junior high school for 3 years (A10_2B) and graduated at age 17 (A10D2-BYEAR), this respondent would have 4 for the education level for ages 17 and onward and would have 3 for ages 16 and 15.  If no further education for this reppondent, this respondent's age-specific education level would be 2 at age 14, 3 at ages 15 and 16, and 4 at age 17, age 18, up to age 50.

*****************************************************************************************************

AEDL14  	EDUCATION LEVEL:AGE 14

		Range: 0-9.

		-97: All calendar years left school missing
		-96: Attended primary school after age 14
		-77: Problematic education history
	
AEDL15  	EDUCATION LEVEL:AGE 15

		Range: 0-9.

		-97: All calendar years left school missing
		-96: Attended primary school after age 15
		-77: Problematic education history
	
AEDL16  	EDUCATION LEVEL:AGE 16

		Range: 0-11.

		-97: All calendar years left school missing
		-96: Attended primary school after age 16
		-77: Problematic education history
	
AEDL17  	EDUCATION LEVEL:AGE 17

		Range: 0-11.

		-97: All calendar years left school missing
		-96: Attended primary school after age 17
		-77: Problematic education history
	
AEDL18  	EDUCATION LEVEL:AGE 18

		Range: 0-11.

		-97: All calendar years left school missing
		-96: Attended primary school after age 18
		-77: Problematic education history
	
AEDL19  	EDUCATION LEVEL:AGE 19

		Range: 0-12.

		-97: All calendar years left school missing
		-96: Attended primary school after age 19
		-77: Problematic education history
	
AEDL20  	EDUCATION LEVEL:AGE 20

		Range: 0-12.

		-97: All calendar years left school missing
		-96: Attended primary school after age 20
		-77: Problematic education history
	
AEDL21  	EDUCATION LEVEL:AGE 21

		Range: 0-12.

		-97: All calendar years left school missing
		-96: Attended primary school after age 21
		-77: Problematic education history
	
AEDL22  	EDUCATION LEVEL:AGE 22

		Range: 0-13.

		-97: All calendar years left school missing
		-96: Attended primary school after age 22
		-77: Problematic education history
	
AEDL23  	EDUCATION LEVEL:AGE 23

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 23
		-77: Problematic education history
	
AEDL24  	EDUCATION LEVEL:AGE 24

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 24
		-77: Problematic education history
	
AEDL25  	EDUCATION LEVEL:AGE 25

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 25
		-77: Problematic education history
	
AEDL26  	EDUCATION LEVEL:AGE 26

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 26
		-77: Problematic education history
	
AEDL27  	EDUCATION LEVEL:AGE 27

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 27
		-77: Problematic education history
	
AEDL28  	EDUCATION LEVEL:AGE 28

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 28
		-77: Problematic education history
	
AEDL29  	EDUCATION LEVEL:AGE 29

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 29
		-77: Problematic education history
	
AEDL30  	EDUCATION LEVEL:AGE 30

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 30
		-77: Problematic education history
	
AEDL31  	EDUCATION LEVEL:AGE 31

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 31
		-77: Problematic education history
	
AEDL32  	EDUCATION LEVEL:AGE 32

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 32
		-77: Problematic education history
	
AEDL33  	EDUCATION LEVEL:AGE 33

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 33
		-77: Problematic education history
	
AEDL34  	EDUCATION LEVEL:AGE 34

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 34
		-77: Problematic education history
	
AEDL35  	EDUCATION LEVEL:AGE 35

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 35
		-77: Problematic education history
	
AEDL36  	EDUCATION LEVEL:AGE 36

		Range: 0-14.

		-97: All calendar years left school missing
		-96: Attended primary school after age 36
		-77: Problematic education history
	
AEDL37  	EDUCATION LEVEL:AGE 37

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL38  	EDUCATION LEVEL:AGE 38

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL39  	EDUCATION LEVEL:AGE 39

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL40  	EDUCATION LEVEL:AGE 40

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL41  	EDUCATION LEVEL:AGE 41

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL42  	EDUCATION LEVEL:AGE 42

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL43  	EDUCATION LEVEL:AGE 43

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL44  	EDUCATION LEVEL:AGE 44

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL45  	EDUCATION LEVEL:AGE 45

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL46  	EDUCATION LEVEL:AGE 46

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL47  	EDUCATION LEVEL:AGE 47

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL48  	EDUCATION LEVEL:AGE 48

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL49  	EDUCATION LEVEL:AGE 49

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDL50  	EDUCATION LEVEL:AGE 50

		Range: 0-14.

		-97: All calendar years left school missing
		-77: Problematic education history
	
*****************************************************************************************************
AEDE14 to AEDE50:
		The information from the education history (Questionnair Section A Q10) was used to construct the dichotomous age specific enrollement status for age 14 upto age 50.  The procedure to construct these variables (AEDE14 to AEDE50) was (see agesed.sps):

		Initialize the variables to 0.  Then, at each education level starting from the lowest level primary school to the highest level graduate school,

		1) Compute the age left the given education level by subtracting the calendar year left the given education level (A10D1 to A10D7) from the birth year (BYEAR).  For those who attended a given education level and not graduate, the calendar year left school was estimated from the calendar year left the previous education level and the number of years stayed at the given education level (see A10D1 to A10D7).

		2) Coded 1 for the age left the given education level.

		3) Coded 1 for the ages as many years back as the number of the year stayed in the given education level. 
		
		For example, at the primary school level if a respondent stayed at primary school for 6 years and graduated at age 12 (A10D1-BYEAR), then this respondent would be enrolled at school at ages 12 and at ages between 11 and 6 (12-6=6).  Furthermore, at junior high school level if this respondent stayed at junior high school for 3 years (A10_2B) and graduated at age 17 (A10D2-BYEAR), this respondent would be enrolled in school at age 17 and at ages 16, 15, and 14.  If no further education for this reppondent, this respondent's age-specific enrollment status would be 1 for age 14, 15, 16, and 17 and be 0 otherwise.

*****************************************************************************************************

AEDE14  	EDUCATION ENROLLMENT:AGE 14

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE15  	EDUCATION ENROLLMENT:AGE 15

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE16  	EDUCATION ENROLLMENT:AGE 16

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE17  	EDUCATION ENROLLMENT:AGE 17

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE18  	EDUCATION ENROLLMENT:AGE 18

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE19  	EDUCATION ENROLLMENT:AGE 19

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE20  	EDUCATION ENROLLMENT:AGE 20

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE21  	EDUCATION ENROLLMENT:AGE 21

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE22  	EDUCATION ENROLLMENT:AGE 22

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE23  	EDUCATION ENROLLMENT:AGE 23

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE24  	EDUCATION ENROLLMENT:AGE 24

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE25  	EDUCATION ENROLLMENT:AGE 25

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE26  	EDUCATION ENROLLMENT:AGE 26

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE27  	EDUCATION ENROLLMENT:AGE 27

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE28  	EDUCATION ENROLLMENT:AGE 28

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE29  	EDUCATION ENROLLMENT:AGE 29

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE30  	EDUCATION ENROLLMENT:AGE 30

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE31  	EDUCATION ENROLLMENT:AGE 31

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE32  	EDUCATION ENROLLMENT:AGE 32

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE33  	EDUCATION ENROLLMENT:AGE 33

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE34  	EDUCATION ENROLLMENT:AGE 34

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE35  	EDUCATION ENROLLMENT:AGE 35

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE36  	EDUCATION ENROLLMENT:AGE 36

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE37  	EDUCATION ENROLLMENT:AGE 37

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE38  	EDUCATION ENROLLMENT:AGE 38

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE39  	EDUCATION ENROLLMENT:AGE 39

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE40  	EDUCATION ENROLLMENT:AGE 40

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE41  	EDUCATION ENROLLMENT:AGE 41

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE42  	EDUCATION ENROLLMENT:AGE 42

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE43  	EDUCATION ENROLLMENT:AGE 43

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE44  	EDUCATION ENROLLMENT:AGE 44

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE45  	EDUCATION ENROLLMENT:AGE 45

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE46  	EDUCATION ENROLLMENT:AGE 46

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE47  	EDUCATION ENROLLMENT:AGE 47

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE48  	EDUCATION ENROLLMENT:AGE 48

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE49  	EDUCATION ENROLLMENT:AGE 49

		Dummy 0, 1.

		  0: No
		  1: Yes
		-97: All calendar years left school missing
		-77: Problematic education history
	
AEDE50  	EDUCATION ENROLLMENT:AGE 50

		Dummy 0, 1.
		
		  0: No
		-97: All calendar years left school missing
		-77: Problematic education history
	
-----------------------------------------------------------------------------------

PWT_UA7P   PREDICTED WEIGHT:PRIMARY SAMPLE, URBAN POP

EDU_8G		ATTENDED EDUCATION LEVEL

		Range: 1-8

		1: No formal schooling
		2: Primary school
		3: Middle school
		4: High school
		5: Technical secondary school
		6: Junior college
		7: University
		8: Graduate School

		This measure was obtained from checking the sequence of the variables whether respondents attended each level of schooling (i.e., A10_1A TO A10_7A) conditioning on from the highest level (whether attending graduate school) to the lowest level (whether attending primary school).  If a respondent attended a higher level of schooling without a lower level, say one attended senior high school without attended junior high school, senior high school would be the level to be assigned to EDU_8G.
		
			EDU_8G=8	if A10_7A=1
			EDU_8G=7	if A10_6A=1
			EDU_8G=6	if A10_5A=1
			EDU_8G=5	if A10_4A=1
			EDU_8G=4	if A10_3A=1
			EDU_8G=3	if A10_2A=1
			EDU_8G=2	if A10_1A=1
			EDU_8G=1	if A10_1A=2

		No missing code.
	
AGE_60G		AGE IN 2000:0=1-18, 65=65+

		Range: 0, 19-65

		0: age 18 or younger in 2000 
		65: age 65 or older in 2000

		This measure was obtained from adding 1 to the age in 1999 and then recoding the variable.
		
			AGE_60G=0	if AGE+1 < 19
			AGE_60G=AGE+1	if AGE+1=19 to 64
			AGE_60G=65	if AGE+1=65 or 65+

		No missing code.

CITY_A60	CITY POP DIST:752 GROUPS/CITY

		Range: 52-529628

		This was a contextual variable with 752 categories (2 gender by 8 education groups by 47 age groups) for each city.  The sources of this measure were from the frequency distribution tables of the single year age (for age 6 to 65+) by education (8 groups) by gender (2 groups) for each district in each city from the 2000 China Census Data.  This measure was obtained by summing the populations over all the districts in a given city and selecting the groups for the age 19 or above.  The categories with empty cells in the sample would be dropped when the variable was matched to the respondents' individual record data.
		
		shanghai_pop.xls, wuhan_pop.xls, and xian_pop.xls contained the detail population distributions by age (60 groups) by education (8 groups) by gender (2 groups) for each districts as well as for the city total population in each category, along with the index variables of the grouping, for Shanghai, Wuhan, and Xian, respectively.  shanghai_pop.xls, wuhan_pop.xls, and xian_pop.xls were converted into SPSS data save files: shanghai_pop.sav, wuhan_pop.sav, and xian_pop.sav, respectively.  The data for age 19 or above were selected in each city, and then these three cities' data were merged into one file with 752 groups (47 age groups by 8 education groups by 2 gender) for each city. The variable of the city adult population distribution (i.e., CITY_A60) was then matched to the respondents' individual record data by the index variables of CITY, SEX, EDU_8G, and AGE_60G.
		
		-888887: NA:not sample age range

URBANA60	URBAN POP DIST:752 GROUPS/CITY

		Range: 49-418496

		This was a contextual variable with 752 categories (2 gender by 8 education groups by 47 age groups) for each city.  The sources of this measure were from the frequency distribution tables of the single year age (for age 6 to 65+) by education (8 groups) by gender (2 groups) for each district in each city from the 2000 China Census Data.  This measure was obtained by selecting the groups for the age 19 or above and summing the populations over the urban districts (i.e., the districts in the 3city Survey) in a given city.  The categories with empty cells in the sample would be dropped when the variable was matched to the individual respondent's data.
		
		shanghai_pop.xls, wuhan_pop.xls, and xian_pop.xls contained the population distributions by age (60 groups) by education (8 groups) by gender (2 groups) for each districts as well as for the city total population in each category, along with the index variables of the grouping, for Shanghai, Wuhan, and Xian, respectively.  shanghai_pop.xls, wuhan_pop.xls, and xian_pop.xls were converted into SPSS data save files: shanghai_pop.sav, wuhan_pop.sav, and xian_pop.sav, respectively. The data for age 19 or above were selected and the district populations were summed over the urban districts (i.e., the districts in the 3city Survey) in each city to create the variable of URBANA60. These three cities's data were then merged into one file with 752 groups for each city (47 age groups by 8 education groups by 2 gender). URBANA60 was then matched to each individual respondent's data by the index variables (CITY, SEX, EDU_8G, and AGE_60G).
		
		-888887: NA:not sample age range

CITYTP  	CITY TOTAL POP:AGE 19+ IN 2000

		Codes: 3365952, 6628828, and 11878962.
		 3365952: Xian total population aged 19 or older
		 6628828: Wuhan total population aged 19 or older
		11878962: Shanghai total population aged 19 or older

		This was a contextual variable with 1 value for each city.  The measure was obtained by aggregating CITY_A60 by the index variable of CITY in the combined 3 cities adult population distribution data (with 752 groups for each city).  The variable of the city adult total population (i.e., CITYTP) was matched back to the combined 3 cities population distribution data (with 752 groups for each city) by CITY.  Then, the variable was matched to respondent's individual record by the index variables of CITY, SEX, EDU_8G, and AGE_60G.
		
		-8888887: NA:not sample age range

URBANTP 	CITY TOTAL URBAN POP:AGE 19+ IN 2000

		Codes: 2078775, 3752472, and 9779320.
		2078775: Xian total urban population aged 19 or older
		3752472: Wuhan total urban population aged 19 or older
		9779320: Shanghai total urban population aged 19 or older

		This was a contextual variable with 1 value for each city.  The measure was obtained by aggregating URBANA60 by the index variable of CITY in the combined 3 cities adult population distribution data (with 752 groups for each city).  The variable of the urban adult total population (i.e., URBANTP) was matched back to the combined 3 cities population distribution data (with 752 groups for each city) by CITY.  Then, the variable was matched to respondent's individual record by the index variables of CITY, SEX, EDU_8G, and AGE_60G.
		
		-8888887: NA:not sample age range

PC_A60		CITY POP DIST(752):PROPORTION

		This was a contextual variable with 752 categories (2 gender by 8 education groups by 47 age groups) for each city and was obtained by dividing CITY_A60 by CITYTP.
		
			PC_A60=CITY_A60/CITYTP
		
		Then the variable was matched to the respondents' individual record data by the index variables of CITY, SEX, EDU_8G, and AGE_60G.
		
		-87.0: NA:not sample age range		

PU_A60		URBAN POP DIST(752):PROPORTION

		This was a contextual variable with 752 categories (2 gender by 8 education groups by 47 age groups) for each city and was obtained by dividing URBANA60 by URBANTP.
		
			PU_A60=URBANA60/URBANTP		
		
		Then the variable was matched to the respondents' individual record data by the index variables of CITY, SEX, EDU_8G, and AGE_60G.
		
		-87.0: NA:not sample age range

PC_A7   	CITY POP DIST(252):PROPORTION

		Range: .0007-.0643

		This was a contextual variable with 84 categories (2 gender by 7 education groups by 6 age groups) for each city.  The measure was obtained by aggregating PC_A60 by the index variables of CITY, SEX, EDU_7G, and AGE_7G in the 3 cities detail adult population distribution data (with 752 groups for each city), where PC_A60=CITY_A60/CITYTP.  The aggregated variable of the city adult population proportion distribution with 84 groups for each city (i.e., PC_A7) was matched to respondents' individual records by the index variables of CITY, SEX, EDU_7G, and AGE_7G (see the descriptions of the variables below).  There had only 229 unique valid values because there were 23 empty cells in the sample distribution.
		
		-87.0000: NA:not sample age range

PU_A7   	URBAN POP DIST(252):PROPORTION

		Range: .0007-.0552

		This was a contextual variable with 84 categories (2 gender by 7 education groups by 6 age groups) for each city.  The measure was obtained by aggregating PU_A60 by the index variables of CITY, SEX, EDU_7G, and AGE_7G in the 3 cities detail adult population distribution data (with 752 groups for each city), where PU_A60=URBANA60/URBANTP.  The aggregated variable of the city adult population proportion distribution with 84 groups for each city (i.e., PU_A7) was matched to respondent's individual record by the index variables of CITY, SEX, EDU_7G, and AGE_7G (see the descriptions of the variables below).  There had only 229 unique valid values because there were 23 empty cells in the sample distribution.
		
		-87.0000: NA:not sample age range

EDU_7G  	EDUCATION ATTENDED 7 GROUPS

		Range: 1-7

		1: No formal schooling
		2: Primary school
		3: Middle school
		4: High school
		5: Technical secondary school
		6: Junior college
		7: University or Graduate School

		This measure was obtained from recoding EDU_8G.
		
			EDU_7G=7	if EDU_8G=7 or 8
			EDU_7G=EDU_8G	otherwise

		No missing code.
	
AGE_7G  	AGE IN 2000:7 GROUPS

		Range: 0-6

		0: age 18 or younger in 2000 
		1: age 19 to 24 in 2000
		2: age 25 to 34 in 2000      
		3: age 35 to 44 in 2000
		4: age 45 to 54 in 2000
		5: age 55 5o 64 in 2000
		6: age 65+ in 2000

		This measure was obtained from recoding AGE_60G.
		
			AGE_7G=0	if AGE_60G=0
			AGE_7G=1	if AGE_60G=19 to 24
			AGE_7G=2	if AGE_60G=25 to 34
			AGE_7G=3	if AGE_60G=35 to 44
			AGE_7G=4	if AGE_60G=45 to 54
			AGE_7G=5	if AGE_60G=55 to 64
			AGE_7G=6	if AGE_60G=65 or above

		No missing code.

CITY_A7 	CITY POP FREQ DIST:252 GROUPS

		Range: 2769-638017

		This was a contextual variable with 84 categories (2 gender by 7 education groups by 6 age groups) for each city.  The measure was obtained by aggregating CITY_A60 by the index variables of CITY, SEX, EDU_7G, and AGE_7G in the 3 cities detail adult population distribution data (with 752 groups for each city).  The aggregated variable of the city adult population distribution with 84 groups for each city (i.e., CITY_A7) was matched to respondents' individual records by the index variables of CITY, SEX, EDU_7G, and AGE_7G.  There had only 229 unique valid values because there were 23 empty cells in the sample distribution.
		
		-888887: NA:not sample age range

URBAN_A7	URBAN POP FREQ DIST:252 GROUPS

		Range: 1370-539667

		This was a contextual variable with 84 categories (2 gender by 7 education groups by 6 age groups) for each city.  The measure was obtained by aggregating URBANA60 by the index variables of CITY, SEX, EDU_7G, and AGE_7G in the 3 cities detail adult population distribution data (with 752 groups for each city).  The aggregated variable of the urban adult population distribution with 84 groups for each city (i.e., URBAN_A7) was matched to respondent's individual record by the index variables of CITY, SEX, EDU_7G, and AGE_7G.  There had only 229 unique valid values because there were 23 empty cells in the sample distribution.
		
		-888887: NA:not sample age range

FREQ_S1 	SAMPLE DISTRIBUTION:252 GROUPS

		Range: 1-127
		
		This variable was the observed cell frequencies for the cross table of AGE_7G by EDU_7G by SEX by CITY for the entire 3city survey respondents' individual records and was obtained by the crosstab command in a SPSS program with the write option.  The observed cell frequencies (FREQ_S1) was matched to the file with population distributions and then was matched to respondents' individual record data by the index variables of CITY, SEX, EDU_7G, and AGE_7G.

		-887: NA:not sample age range

WT_CA7  	OBSERVED WEIGHT:CITY POP 252 CELLS

		Range: 495-79252

		This was a contextual variable with 84 categories (2 gender by 7 education groups by 6 age groups) for each city.  The measure was obtained by rounding the value of CITY_A7 divided by FREQ_S1 (see wt_popsmp_a7.do).
		
			WT_CA7=ROUND(CITY_A7/FREQ_S1)
			
  		The variable was then matched to respondents' individual records by the index variables of CITY, SEX, EDU_7G, and AGE_7G.  For those respondents whose age was younger than 18 in 1999, the weights from the corresponding education, gender and city for the age group 19 to 24 in 2000 were used.
  
		No missing code.			
		
WT_UA7  	OBSERVED WEIGHT:URBAN POP 252 CELLS

		Range: 248-70755

		This was a contextual variable with 84 categories (2 gender by 7 education groups by 6 age groups) for each city.  The measure was obtained by rounding the value of URBAN_A7 divided by FREQ_S1 (see wt_popsmp_a7.do).
		
			WT_CA7=ROUND(URBAN_A7/FREQ_S1)
						
  		The variable was then matched to respondents' individual records by the index variables of CITY, SEX, EDU_7G, and AGE_7G.  For those respondents whose age was younger than 18 in 1999, the weights from the corresponding education, gender and city for the age group 19 to 24 in 2000 were used.
  
		No missing code.			

PFRE_S1		PREDICTED SAMPLE DIST:252 GROUPS

		Range: 1-152
		
		This variable was the predicted cell frequency for AGE_7G by EDU_7G by SEX by CITY for the entire 3city survey respondents' individual records and was obtained by rounding the predicted mean value of the log-rate model using FREQ_S1 as the dependent variable and the dummy indicators of AGE_7G, EDU_7G, SEX, and CITY as independent variables with the log of URBAN_A7 as the offset variable (see wt_popsmp_a7.do).
		
			 PFRE_S1=round(y) &
			 y=g_inverse(xb+offset)

			 where g was Poisson distribution, x was the matrix of the independent variables, b was the vector of the estimated coefficients, and offset was the ln(URBAN_A7).

		The predicted cell frequencies (PFRE_S1) was matched to respondents' individual record data by the index variables of CITY, SEX, EDU_7G, and AGE_7G.
		
		-887: NA:not sample age range

PWT_UA7 	PREDICTED WEIGHT:ALL R'S SAMPLE, URBAN POP

		Range: 384-56970

		This was a contextual variable with 84 categories (2 gender by 7 education groups by 6 age groups) for each city.  The measure was obtained by rounding the value of URBAN_A7 divided by PFRE_S1 (see wt_popsmp_a7.do).
		
			PWT_CA7=ROUND(URBAN_A7/PFRE_S1)
						
  		The variable was then matched to respondents' individual records by the index variables of CITY, SEX, EDU_7G, and AGE_7G.  For those respondents whose age was younger than 18 in 1999, the predicted weights from the corresponding education, gender and city for the age group 19 to 24 in 2000 were used.
  
		No missing code.			

FREQ_S1P	PRIMARY R SAMPLE DIST:252 GROUPS

		Range: 1-99
		
		This variable was the observed cell frequencies for the cross table of AGE_7G by EDU_7G by SEX by CITY for the primary respondents (FORM=1 or FORM=2) and was obtained by the crosstab command in a SPSS program with the write option for the primary respondents subsample.  The observed cell frequencies for the primary respondents (FREQ_S1P) was matched to primary respondents' individual record data by the index variables of CITY, SEX, EDU_7G, and AGE_7G.  Then, the observed cell frequencies for  the primary respondents were assigned to the match pair respondents.

		-887: NA:not in sample age range
		-886: NA:Not match pairs (FORM=3/4 & MATCH ne 1)
		
WT_UA7P 	WEIGHT:PRIMARY R URBAN POP 252 CELLS

		Range: 343-70755

		This was a contextual variable with 84 categories (2 gender by 7 education groups by 6 age groups) for each city.  The measure was obtained by rounding the value of URBAN_A7 divided by FREQ_S1P (see wt_popsmp_a7p.do).
		
			WT_CA7P=ROUND(URBAN_A7/FREQ_S1P)
						
  		The variable was then matched to primary respondents' individual records by the index variables of CITY, SEX, EDU_7G, and AGE_7G.  Then, the weights for the primary respondents were assigned to the match pair respondents.
  
		-888886: NA:Not match pairs (form=3/4 & match ne 1)
		
PFRE_S1P	PREDICTED PRIMARY SAMPLE DIST:252 GROUPS

		Range: 1-89

		This variable was the predicted cell frequency for AGE_7G by EDU_7G by SEX by CITY for the primary respondents and was obtained by rounding the predicted mean value of the log-rate model using FREQ_S1P as the dependent variable and the dummy indicators of AGE_7G, EDU_7G, SEX, and CITY as independent variables with the log of URBAN_A7 as the offset variable (see wt_popsmp_a7p.do).
		
			 PFRE_S1P=round(y2) &
			 y2=g_inverse(xb+offset)

			 where g was Poisson distribution, x was the matrix of the independent variables, b was the vector of the estimated coefficients, and offset was the ln(URBAN_A7).

		The predicted cell frequencies (PFRE_S1P) was matched to primary respondents' individual record data by the index variables of CITY, SEX, EDU_7G, and AGE_7G.
		
		-887: NA:not in sample age range
		-886: NA:Not match pairs (FORM=3/4 & MATCH ne 1)
		
PWT_UA7P	PREDICTED WEIGHT:PRIMARY SAMPLE, URBAN POP

		Range: 549-56970

		This was a contextual variable with 84 categories (2 gender by 7 education groups by 6 age groups) for each city.  The measure was obtained by rounding the value of URBAN_A7 divided by PFRE_S1P (see wt_popsmp_a7p.do).
		
			PWT_CA7P=ROUND(URBAN_A7/PFRE_S1P)

  		The variable was then matched to respondents' individual records by the index variables of CITY, SEX, EDU_7G, and AGE_7G.  For those respondents whose age was younger than 18 in 1999, the predicted weights from the corresponding education, gender and city for the age group 19 to 24 in 2000 were used.
  		
		-888886: NA:Not match pairs (form=3/4 & match ne 1)		
-----------------------------------------------------------------------------------------------------
