Polygyny and African couple research

Fertility and family planning research has recently begun to focus on coupl es, rather than solely on women. High levels of polygyny in sub-Saharan Afr ica create two interesting problems for couple analyses. First, some men ma tch with multiple wives in a given sample. Hence observations of a dependen t variable that are a function of the responses from a polygynous man are n ot statistically independent. To correct for this, we propose that research ers use the method of generalized estimating equations. The second problem occurs when survey questions do not properly account for men having multipl e partners. Variables constructed from such questions may contain error. Un til better data are available, we propose that researchers randomly select a wife for each polygynous man to reduce the effects of this error. We illu strate these ideas by studying the determinants of couples having innovativ e family planning behaviors and attitudes in Niger.