Describing Stratified Multiple Responses for Sparse Data
Surveys often contain qualitative variables for which respondents may select any number of the outcome categories. For instance, for the question “What type of contraception have you used? ” with possible responses (oral, condom, lubricated condom, spermicide, and di-aphragm), respondents would be i...
Main Author: | |
---|---|
Other Authors: | |
Format: | Text |
Language: | English |
Published: |
2004
|
Subjects: | |
Online Access: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.625.3909 http://www.mcs.vuw.ac.nz/research/publications/reports/mscs/mscs04-04.pdf |
Summary: | Surveys often contain qualitative variables for which respondents may select any number of the outcome categories. For instance, for the question “What type of contraception have you used? ” with possible responses (oral, condom, lubricated condom, spermicide, and di-aphragm), respondents would be instructed to select as many of the outcomes that apply. This type of response is called multiple responses. Bilder and Loughin (2002) proposed a Cochran-Mantel-Haenszel (MH) type method to test whether the choice of type of contraception is marginally independent of an explanatory variable given a stratification variable (known as conditional multiple marginal independence, CMMI). We apply the generalized MH type es-timators (Greenland, 1989) to estimate the conditional group effects among the c outcome categories and follow the bootstrap method to estimate the variances and covariances for the estimators. The method can also be used for data with dependent observations across strata. It performs well even for highly sparse data. Key words: bootstrap method; Mantel-Haenszel estimator; multiple responses; odds ratio. 1 1 |
---|