Considerations before Pooling Data from Two Different Cycles of the General Social Survey

The General Social Survey Program has been gathering data for some time now in a series of independent annual household surveys. A rich collection of data now exists for social science and other researchers. One interesting potential tool is the integration of data across two or more cycles. This means using the data from two or more cycles to estimate quantities of interest. There are two ways of doing this (which, in general, give different answers). One is to compute separate estimates by cycle and combine them. Another way is by simply pooling the data sets from different cycles together and computing estimates using the pooled data. This document provides a brief description of situations when the second method, pooling, is appropriate as well as some things to take into account for the researcher interested in pooling. Additionally, some basic, practical recipes are provided to aid the researcher in the integrating exercise. Perusing the contents immediately below will provide a broad overview of the considerations and recipes. In essence, the researcher must decide whether or not the two data sets should be pooled and can be pooled in order to conduct their analysis project.