120 likes | 230 Vues
This lecture focuses on the methodological aspects of studying marriage events through logistic regression analysis, particularly for individuals not currently married. It discusses how to examine the effects of time-varying characteristics and the importance of parental background in understanding marriage dynamics. The lecture also explores data management techniques using STATA to merge parental characteristics with individual data, enabling deeper insights into marriage history, remarriage rates, and social mobility.
E N D
CMGPD-LNMethodological Lecture Day 5 Marriage
MarriageEvent history • Outcome: marriage (or remarriage) in the next three years • Logistic regression • Restrict to people not currently married • Useful for considering the effects of time-varying characteristics • Current characteristics of kin, etc. • Can be carried out for daughters, subject to caveats related to omission of records • Discrete-time event-history via logit or complementary log-log • AT_RISK_MARRY and NEXT_MARRY • AT_RISK_REMARRY and NEXT_REMARRY
MarriageCurrent marital status • Alternatively, can look at current marital status as an outcome • Fine if right-hand side of variables of interest are fixed • Childhood characteristics etc. • Can construct flag variables for outcomes from MARITAL_STATUS • Ever married (MARITAL_STATUS != 2 & MARITAL_STATUS >= 1) • Currently married (MARITAL_STATUS == 1) • Currently remarried (MARITAL_STATUS==4) • Would want to restrict to observations of people who were MARITAL_STATUS == 3 | MARITAL_STATUS == 4
MarriageVariables for parents’ characteristics • For the study of marriage, and social mobility, parental characteristics are important • In STATA, it is easy to extract information from records of parents, and add it as a variable to an individual’s observation • Basic principle is to merge based on FATHER_ID, MOTHER_ID and so forth • Can also extract information from other kin
Preparing a file with father’s information to be merged in use "C:\Users\Cameron Campbe\Documents\Baqi\CMGPD-LN from ICPSR\ICPSR_27063\DS0001\27063-0001-Data.dta", clear keep PERSON_ID YEAR HAS_POSITION rename PERSON_ID FATHER_ID rename HAS_POSITION FATHER_HAS_POSITION bysort FATHER_ID YEAR: keep if _n == 1 save father_information, replace
Merging the file and carrying out a tabulation use "C:\Users\Cameron Campbe\Documents\Baqi\CMGPD-LN from ICPSR\ICPSR_27063\DS0001\27063-0001-Data.dta", clear merge m:1 FATHER_ID YEAR using father_information drop if _merge == 2 label variable FATHER_HAS_POSITION "Father has position" tab HAS_POSITION FATHER_HAS_POSITION if SEX == 2 & AGE_IN_SUI > 16 & AGE_IN_SUI <= 55 & PRESENT & FATHER_HAS_POSITION >= 0, col generate byte IS_MARRIED = MARITAL_STATUS != 2 tab IS_MARRIED FATHER_HAS_POSITION if SEX == 2 & AGE_IN_SUI > 16 & AGE_IN_SUI <= 35 & PRESENT & FATHER_HAS_POSITION >= 0 & MARITAL_STATUS >= 1, col
tab HAS_POSITION FATHER_HAS_POSITION if SEX == 2 & AGE_IN_SUI > 16 & AGE_IN_SUI <= 55 & PRESENT & FATHER_HAS_POSITION >= 0, col +-------------------+ | Key | |-------------------| | frequency | | column percentage | +-------------------+ Has Official | Father has position Position | No Yes | Total ---------------+----------------------+---------- No | 218,254 9,256 | 227,510 | 98.84 86.88 | 98.29 ---------------+----------------------+---------- Yes | 2,561 1,398 | 3,959 | 1.16 13.12 | 1.71 ---------------+----------------------+---------- Total | 220,815 10,654 | 231,469 | 100.00 100.00 | 100.00
. tab IS_MARRIED FATHER_HAS_POSITION if SEX == 2 & AGE_IN_SUI > 16 & AGE_IN_SUI <= 35 & PRESENT & FATHER_HAS_POSITION >= 0 & MARITAL > _STATUS >= 1, col +-------------------+ | Key | |-------------------| | frequency | | column percentage | +-------------------+ | Father has position IS_MARRIED | No Yes | Total -----------+----------------------+---------- 0 | 68,210 2,452 | 70,662 | 40.20 30.27 | 39.75 -----------+----------------------+---------- 1 | 101,447 5,648 | 107,095 | 59.80 69.73 | 60.25 -----------+----------------------+---------- Total | 169,657 8,100 | 177,757 | 100.00 100.00 | 100.00
Creating one record per father and using it as the basis for the merge use "C:\Users\Cameron Campbe\Documents\Baqi\CMGPD-LN from ICPSR\ICPSR_27063\DS0001\27063-0001-Data.dta", clear bysort PERSON_ID (HAS_POSITION): generate father_ever_position = HAS_POSITION[_N] bysort PERSON_ID: keep if _n == 1 keep PERSON_ID father_ever_position rename PERSON_ID FATHER_ID save father_ever_position use "C:\Users\Cameron Campbe\Documents\Baqi\CMGPD-LN from ICPSR\ICPSR_27063\DS0001\27063-0001-Data.dta", clear keep if SEX == 2 merge m:1 FATHER_ID using father_ever_position, keep(match master) label variable father_ever_position "Father ever held position" tab HAS_POSITION father_ever_position if PRESENT & HAS_POSITION >= 0 & father_ever_position >= 0 & AGE_IN_SUI >= 16 & AGE_IN_SUI <= 55, column
Results +-------------------+ | Key | |-------------------| | frequency | | column percentage | +-------------------+ | Father ever held Has Official | position Position | 0 1 | Total ---------------+----------------------+---------- No | 429,455 22,841 | 452,296 | 98.76 86.02 | 98.03 ---------------+----------------------+---------- Yes | 5,390 3,711 | 9,101 | 1.24 13.98 | 1.97 ---------------+----------------------+---------- Total | 434,845 26,552 | 461,397 | 100.00 100.00 | 100.00
Moving information about a husband into a wife’s record use "C:\Users\Cameron Campbe\Documents\Baqi\CMGPD-LN from ICPSR\ICPSR_27063\DS0001\27063-0001-Data.dta", clear keep if SEX == 2 keep PERSON_ID YEAR AGE_IN_SUI rename PERSON_ID HUSBAND_ID rename AGE_IN_SUI HUSBAND_AGE bysort HUSBAND_ID YEAR: keep if _n == 1 save husband_information use "C:\Users\Cameron Campbe\Documents\Baqi\CMGPD-LN from ICPSR\ICPSR_27063\DS0001\27063-0001-Data.dta", clear keep if SEX == 1 merge m:1 HUSBAND_ID YEAR using husband_information, keep(match master) generate AGE_DIFFERENCE = HUSBAND_AGE - AGE_IN_SUI if HUSBAND_AGE > 0 & AGE_IN_SUI > 0 recode AGE_DIFFERENCE min/-10=-10 10/max=10 bysortPERSON_ID:keep if _n == 1& MARITAL_STATUS > 0 & MARITAL_STATUS != 2 histogram AGE_DIFFERENCE, width(1) discrete scheme(s1mono) xtitle("Husband's age - wife's age") fraction