Secondary Analysis of Survey Data

Secondary analysis of survey data is the reanalysis of existing survey responses with research questions that differ from those of the original research. With an increasing number of data sets available to researchers, the role of surveys in social research has expanded, and, with improved data quality and access, secondary analysis has assumed an increasingly central position. Secondary survey data are available on a wide range of topics covering health, attitudes, population characteristics, criminal behavior, aging, and more, and many existing data sets are useful for both cross-sectional and temporal analyses. A major advantage of secondary survey analysis is resource economy. Both time and money are saved, and the need for extensive personnel is eliminated. Because the data already exist, the analysis is typically noninvasive, although discussions regarding human subjects implications are ongoing. The availability of multiple data sets simplifies comparative analyses—most often of nations or over time.

The limitations of this method are usually those of validity, relating to the appropriateness of existing [Page 1009]data for particular research questions. Because the secondary researcher did not design the original survey, it is unlikely that the data set will contain all variables of interest. Even those that do appear may have been measured in ways that diminish their utility. With existing data sets, researchers must be certain that samples represent the population of interest or have enough cases of the needed subpopulation that can be easily extracted. The unit of analysis must also be compatible with the study. Although many existing data sets were developed with a specific focus, there has been some effort to promote omnibus surveys, that is, those covering a wide range of topics. The most widely known of these is the General Social Survey, which is conducted by the National Opinion Research Center (NORC) at the University of Chicago and contains an array of social indicator data.

Secondary survey data are now widely available on disk, CD-ROM, and the Internet. Many data repositories exist—some with a huge number of holdings covering a variety of topics, and others that are smaller and more specialized. Although academic archives, government agencies, and public opinion research centers (such as Roper and Gallup) are obvious sources of secondary survey data, any organization that stores such data (e.g., private research firms, foundations, and charitable organizations) can be considered a resource for the secondary analyst. A leading data source for academic social science researchers is the Inter-university Consortium for Political and Social Research (ICPSR) at the University of Michigan, which can be accessed on the Internet at http://www.icpsr.umich.edu. The ICPSR is an excellent starting point for locating desired data. In addition to their substantial holdings, the Website provides links to other archives.

To make the best use of existing survey data, researchers should merge general substantive interests with familiarity of existing data and be able to evaluate primary data source(s) with respect to survey design and administration, issues of sampling, and data set limitations. Furthermore, researchers must be creative in their approach to combining data both within and across surveys, and by incorporating data from outside the data set. Although inherent challenges persist in the use of existing survey data, the opportunities for social research are enormous.

...