Anastasia Semykina and Jeffrey M. Wooldridge, "Binary Response Panel
Data Models with Sample Selection and Self Selection," Journal of
Applied Econometrics, Vol. 33, No. 2, 2018, pp. 179-197.

The paper uses National Longitudinal Survey of Youth 79 (NLSY79) data,
which were obtained from

https://www.bls.gov/nls/nlsy79.htm (public-use files)

The sample is limited to white women ages 25 and older, survey years
1990-1994. All observations that have missing values for variables
used in the analysis were deleted. The only exception is 'pension'
variable, which has missing values for all nonworking women. Only
women who remained in the sample during the entire period (1990-1994)
were included.

The sample includes five-year histories for 1668 women, 8340
observations in total. The data are in the file NLSY79_91_94.csv,
which is an ASCII file in DOS format. It is zipped in sw-data.zip.
Unix/Linux users should use "unzip -a".

Stata do-files are available in the file sw-do-files.zip, which
includes three files:

  param_regs.do was used to perform parametric estimation

  flex_param_regs.do was used to perform flexible parametric estimation

  semiparam_regs.do was used to perform semiparametric estimation


Variable definitions:

id              woman's unique ID#
year            survey year
RACE            race
SEX             gender
afqt_1          AFQT score, ability measure
age             age in years
married         =1 if married, 0 if not
educ            years of schooling, truncated at 20 years
working         =1 if working, 0 if not
hrswrk52        number of hours worked last year
pension         =1 if has employer provided pension benefits
mmarried        individual time mean for married
td2             year=1991 indicator
td3             year=1992 indicator
td4             year=1993 indicator
td5             year=1994 indicator