Abstract
J. Thompson and R. Parmer (1994) "A
Comparison of Nonparametric Methods With Parametric Methods for
the CPS CATI/CAPI Mode Effects Analysis," Proceedings of
the Section on Survey Research Methods, American Statistical
Association.
The analysis for mode effects in the CATI/CAPI Overlap
Study has relied heavily on the two sample t-test. This test
is popular because it is easily interpretable and fairly
robust to the assumption of normality. The latter assumption
is, however, difficult to verify with complex survey data.
Moreover, the variance estimate used in the parametric
analysis is not distributed as a Chi-Squared random variable.
Therefore, we apply a variety of nonparametric methods to
split panel data from the Basic CPS and Parallel Survey, and
show a comparison of the results to the normal theory based
results.
Last Modified Date: July 19, 2008
|