Using the RAND HRS Data and RAND-Enhanced Fat Files. Sample Programs for HRS Summer Institute Workshop

Transcription

1 Using the RAND HRS Data and RAND-Enhanced Fat Files Sample Programs for HRS Summer Institute Workshop This document is intended to provide users with some examples of how to both set up and perform some simple descriptive analyses in SAS, SPSS, and Stata, using the RAND HRS data products. The sample programs assume that the data being used are set up in a folder called C:\RandHRS, and that any files you create are stored in a separate folder called C:\MyPaper. There are four programs, each of which should be run in the specified order: Part #1: Run some descriptive analyses, using variables from the RAND HRS dataset (rndhrs_j). o Runs tables on variables that describe the interview response status at each wave. o INW1 - INW9: These are flags that indicate whether an individual responded at a given wave, where 0 = Non-response and 1 = Response, Alive. o RwIWSTAT: These variables provide (where w is the wave, e.g., R9IWSTAT for Wave 9) another way to examine interview status. In addition to telling you whether the individual responded or not, these variables also indicate mortality status. Part #2: Create a file that contains a subset of variables from the RAND HRS dataset (rndhrs_j). o HACOHORT: This variable identifies the entry cohort subsample (e.g., HRS, Ahead, CODA, WB, EBB). o RAGNDER: This variable identifies the gender of the respondent, where 1 = Male and 2 = Female. o RARACEM: This variable identifies the race of the respondent, where 1 = White/Caucasian, 2 = Black/African American, and 3 = Other. Note that respondents who are Hispanic can be identified using the RAHISPAN variable. o R7PENINC - R9PENINC: These variables indicate, for Waves 7 (2004), 8 (2006), and 9 (2008), whether the respondent is currently receiving any pension income. The corresponding variables for the spouse are S7PENINC - S9PENINC. o R7LBRF - R9LBRF: These variables summarize, for Waves 7 (2004), 8 (2006), and 9 (2008), the labor force status for the respondent at each wave as working full-time, working part-time, unemployed, partly retired, retired, disabled, or not in the labor force. The corresponding variables for the spouse are S7LBRF - S9LBRF. o R7AGEY_E - R9AGEY_E: These variables, for Waves 7 (2004), 8 (2006), and 9 (2008), indicate the age of the respondent (in years) at the final interview date. The corresponding variables for the spouse are S7AGEY_E - S9AGEY_E. Part #3: Run some descriptive analyses using the dataset created in Part #2. o Generate summary statistics for respondents pension income receipt in Waves 7-9. o Produce some tables that examine/compare the following: o Respondents pension income receipt by cohort for Wave 9. o Respondents pension income receipt to that of the spouse for Wave 8. o Respondents Wave 8 pension income receipt to that in Wave 9. o Respondents labor force status by pension income receipt for Waves 8 and 9. o Respondents pension income receipt and labor force status in Wave 9 by gender.

2 Part #4: Merge the dataset created in Part #2 with the RAND-Enhanced Fat Files for Waves 8 and 9. The raw variables that are selected from the fat files (described below) include questions that ask respondents whether their employers offer pension plans. o Merge the dataset created in Part #2 with the fat files for Wave 8 (h06f2a) and Wave 9 (h08e1a). There are some important things to note: o For sorting and merging you can use RAHHIDPN, HHIDPN, or HHID & PN together. The results should be the same regardless of which format of the respondent identifier you use. o The rndhrs_j dataset and the fat files are all sorted by RAHHIDPN (HHIDPN and HHID & PN). However, Stata requires you to resort the files before merging, and only allows two files to be merged at once. o Keep RAHHIDPN, HHIDPN, HHID and PN from the fat files, as well as the KJ325 (Wave 8) and LJ325 (Wave 9) variables described below. o In the final merged dataset, keep only the individuals who responded to Waves 8 and 9. o Produce some tables that examine/compare the following: o Response status flags for Waves 8 and 9. o Questions in Waves 8 and 9 about whether employers offer pension plans. o Questions in Waves 8 and 9 about whether employers offer pension plans by gender. ================================================================================ KJ325 DOES EMPLOYER OFFER ANY PLANS Section: J Level: Respondent Type: Numeric Width: 1 Decimals: 0 Ref: SecJ.CURRENTPENSIONNEW.J325_ Does your employer offer any [such plans? / pension or retirement plans on your job?] YES NO DK (Don't Know); NA (Not Ascertained) 3 9. RF (Refused) Blank. INAP (Inapplicable); Partial Interview ================================================================================ LJ325 DOES EMPLOYER OFFER ANY PLANS Section: J Level: Respondent Type: Numeric Width: 1 Decimals: 0 Ref: SecJ.CURRENTPENSIONNEW.J325_ Does your employer offer any (such) retirement plans? YES NO DK (Don't Know) 3 9. RF (Refused) Blank. INAP (Inapplicable); Partial Interview

3 SAS Code Part #1: /* The formats.sas7bcat and rndhrs_j.sas7bdat files are stored in c:\randhrs */ libname library "c:\randhrs"; proc freq data=library.rndhrs_j; table inw1-inw9 r7iwstat r8iwstat r9iwstat / missprint; Part #2: libname mylib "c:\mypaper"; /* this is where the output file will be stored */ data mylib.wkshop; set library.rndhrs_j (keep=inw7-inw9 hhidpn rahhidpn hacohort ragender raracem r7iwstat r8iwstat r9iwstat r7peninc r8peninc r9peninc r7lbrf r8lbrf r9lbrf r7agey_e r8agey_e r9agey_e s7peninc s8peninc s9peninc s7lbrf s8lbrf s9lbrf s7agey_e s8agey_e s9agey_e); where inw7=1 or inw8=1 or inw9=1; proc contents data=mylib.wkshop; Part #3: proc means data=mylib.wkshop; var r7peninc r8peninc r9peninc; proc freq data=mylib.wkshop; table r9peninc*hacohort r8peninc*s8peninc r8peninc*r9peninc r8peninc*r8lbrf r9peninc*r9lbrf ragender*(r9peninc r9lbrf) ragender*r9peninc*r9lbrf /missprint;

4 Part #4: data mylib.wkplus; merge mylib.wkshop (in=inrnd) library.h06f2a (in=i06 keep=hhidpn rahhidpn hhid pn kj325) library.h08e1a (in=i08 keep=hhidpn rahhidpn hhid pn lj325); by rahhidpn; in06=i06; in08=i08; if inw8=1 and inw9=1; proc freq data=mylib.wkplus; table in06*inw8 in08*inw9 kj325 lj325 ragender*kj325 ragender*lj325 /missprint;

5 SPSS Code Part #1: get file="c:\randhrs\rndhrs_j.sav". frequencies var=inw1 inw2 inw3 inw4 inw5 inw6 inw7 inw8 inw9 r7iwstat r8iwstat r9iwstat. Part #2: get file="c:\randhrs\rndhrs_j.sav" /keep=inw7 inw8 inw9 hhidpn rahhidpn hacohort ragender raracem r7iwstat r8iwstat r9iwstat r7peninc r8peninc r9peninc r7lbrf r8lbrf r9lbrf r7agey_e r8agey_e r9agey_e s7peninc s8peninc s9peninc s7lbrf s8lbrf s9lbrf s7agey_e s8agey_e s9agey_e. select if inw7=1 or inw8=1 or inw9=1. save outfile="c:\mypaper\wkshop.sav" /map. Part #3: get file="c:\mypaper\wkshop.sav". descriptives r7peninc r8peninc r9peninc. crosstabs tables=r9peninc by hacohort /tables=r8peninc by s8peninc /tables=r8peninc by r9peninc /missing=include. crosstabs tables=r8lbrf by r8peninc /tables= r9lbrf by r9peninc /cells=count col. crosstabs tables=r9peninc by ragender /tables=r9lbrf by ragender /tables=r9peninc by r9lbrf by ragender /cells=count col.

6 Part #4: match files file="c:\mypaper\wkshop.sav" /in=inrnd /file="c:\randhrs\h06f2a.sav" /in=in06 /file="c:\randhrs\h08e1a.sav" /in=in08 /keep= inw7 inw8 inw9 hhidpn rahhidpn hhid pn ragender raracem r7iwstat r8iwstat r9iwstat r7peninc r8peninc r9peninc r7lbrf r8lbrf r9lbrf r7agey_e r8agey_e r9agey_e s7peninc s8peninc s9peninc s7lbrf s8lbrf s9lbrf s7agey_e s8agey_e s9agey_e kj325 lj325 /by rahhidpn /map. save outfile="c:\mypaper\wkplus.sav". select if inw8=1 and inw9=1. frequencies var=kj325 lj325. crosstab tables= in06 by inw8 /tables=in08 by inw9 /tables=kj325 by ragender /tables=lj325 by ragender /cells=count col.

7 Stata Code Part #1: set memory 200m set maxvar #delimit use inw* hhidpn rahhidpn hacohort ragender raracem r7iwstat r8iwstat r9iwstat r7peninc r8peninc r9peninc r7lbrf r8lbrf r9lbrf r7agey_e r8agey_e r9agey_e s7peninc s8peninc s9peninc s7lbrf s8lbrf s9lbrf s7agey_e s8agey_e s9agey_e using "c:\randhrs\rndhrs_j" #delimit cr tab inw1 tab inw2 tab inw3 tab inw4 tab inw5 tab inw6 tab inw7 tab inw8 tab inw9 tab r7iwstat tab r8iwstat tab r9iwstat Part #2: keep if inw7==1 inw8==1 inw9==1 save "c:\mypaper\wkshop", replace Part #3: use "c:\mypaper\wkshop" sum r7peninc r8peninc r9peninc tab r9peninc hacohort tab r8peninc s8peninc tab r8peninc r9peninc tab r8lbrf r8peninc, col tab r9lbrf r9peninc, col tab r9peninc ragender, col tab r9lbrf ragender, col tab r9lbrf r9peninc if ragender==1, col tab r9lbrf r9peninc if ragender==2, col

8 Part #4: use hhidpn rahhidpn hhid pn kj325 using "c:\randhrs\h06f2a" save "c:\mypaper\h06x", replace use hhidpn rahhidpn hhid pn lj325 using "c:\randhrs\h08e1a" save "c:\mypaper\h08x", replace clear use "c:\mypaper\wkshop" merge rahhidpn using "c:\mypaper\h06x", _merge(mrg06) tab mrg06 merge rahhidpn using "c:\mypaper\h08x", _merge(mrg08) tab mrg08 save "c:\mypaper\wkplus", replace keep if inw8==1 & inw9==1 tab mrg06 inw8 tab mrg08 inw9 tab kj325 tab lj325 tab kj325 ragender tab lj325 ragender