Experimental Methods in Survey Research

I am enormously flattered to be asked to supply a preface for this path-breaking volume of essays about experiments embedded in surveys. I had hoped to contribute a chapter to the volume with my friend and long-term collaborator, Stephen Fienberg, but Steve, active to the end, lost his long battle with cancer before he was able to make the time to work on the chapter we had planned to write. So, I would like to use this opportunity to write something about the work we had done, and had hoped to continue, on the parallels between experimental and survey methodology, on experiments embedded in surveys (and vice-versa), and some of the considerations for analysis occasioned by such embedding. We had long noted (e.g. Fienberg and Tanur 1987, 1988, 1989, 1996) that there are a great many parallels between elements of survey design and experimental design. Although in fact surveys and experiments had developed very long and independent traditions by the start of the twentieth century, it was only with the rise of ideas associated with mathematical statistics in the 1920s that the tools for major progress in these areas became available. ᑖe key intellectual idea was the role of randomization or random selection, both in experimentation and in sampling, and both R.A. Fisher and Jerzy Neyman utilized that idea, although in different ways. ᑖe rich- ness of the two separate literatures continues to offer new opportunities for cross-fertilization of theory and tools for survey practice in particular.