At wave 2, 24 percent of NSFH
wave 1 respondents had dropped out of the study.
Poverty level was determined using the NSFH
poverty variable multiplied by two.
Indeed, a comparison of diary-based data from the Time Use Survey (TUS) with the NSFH
data employed here indicates substantial differences.
In the NSFH
sample the typical (modal) individual is a white, native-born, married woman in her early forties.
Table 1 Number of Women by Cohort and Survey Birth Cohort OCG SIPP NSFH
1910-14 1152 1915-19 2,033 1920-24 2,483 1925-29 2,818 1,790 292 1930-34 2,640 1,556 254 1935-39 2,647 1,691 338 1940-44 2,976 2,009 400 1945-49 2,613 2,485 567 1950-54 2,703 738 1955-59 2,978 699 1960-64 2,293 660 Total 19,362 17,505 3,948
A comparison of wages by marital and cohabitation status (not shown) using the weighted NSFH
sample sheds some light on the phenomenon to be explained.
Two cycles of the NSFH
were conducted, the first from 1987 through 1988 and the second from 1992 through 1994.
In contrast, the NSFH
offers detailed information on the health and socioeconomic status of the respondent's parents and the number of siblings of the respondent that can be used to identify the caregiving effect.
The data for this study come from the 1992-1994 wave of the NSFH
, a nationally representative sample of the U.
In addition, the NSFH
contains only three waves of data, with interviews occurring 5 or more years apart.
We, too, attempt to deal with the simultaneity of these two uses of time, basing our analysis on a large-scale, nationally representative sample with more extensive relevant measures, the NSFH
includes data collected from a nationally representative sample of 9,643 individuals as well as an oversample of minorities and households containing single-parent families, stepfamilies, recently married couples, and cohabiting couples.