This data frame contains 2787 rows and 10 columns. These data are
pertinent to an investigation of the way that earnings changed,
between 1974-1975 and 1978, in the absence of training. Data for the
experimental treatment group in nswdemo are combined with the
psid1 control data from the Panel Study of Income Dynamics
(PSID) study.
psid1This data frame contains the following columns:
a numeric vector identifying the study in which the subjects were enrolled (0 = Control, 1 = treated).
age (in years).
years of education.
(0 = not black, 1 = black).
(0 = not hispanic, 1 = hispanic).
(0 = not married, 1 = married).
(0 = completed high school, 1 = dropout).
real earnings in 1974.
real earnings in 1975.
real earnings in 1978.
The cps1 and psid1 data sets are two non-experimental
"control" groups, alternative to that in nswdemo, used in
investigating whether use of such a non-experimental control group can
be satisfactory. cps2 and cps3 are subsets of cps1,
designed to be better matched to the experimental data than cps1.
Similary psid2 and psid3 are subsets of psid1,
designed to be better matched to the experimental data than
psid1. nswpsid1 combines data for the experimental
treatment group in nswdemo with the psid1 control data
from the Panel Study of Income Dynamics (PSID) study.
Dehejia, R.H. and Wahba, S. 1999. Causal effects in non-experimental studies: re-evaluating the evaluation of training programs. Journal of the American Statistical Association 94: 1053-1062.
Lalonde, R. 1986. Evaluating the economic evaluations of training programs. American Economic Review 76: 604-620.
Smith, J. A. and Todd, P.E. "Does Matching overcome. LaLonde?s critique of nonexperimental estimators", Journal of Econometrics 125: 305-353.
Dehejia, R.H. 2005. Practical propensity score matching: a reply to Smith and Todd. Journal of Econometrics 125: 355-364.