OPENING REMARKS
CORK: Let me introduce myself. My name is Daniel Cork, and I am the study director for this, the [National Research Council (NRC)] Workshop on Survey Automation. This workshop is being conducted by the Committee on National Statistics of the NRC, with sponsorship from the U.S. Census Bureau. The agenda in the agenda books identifies Chet Bowie, who was going to give opening remarks on behalf of the Census Bureau; he is unable to make it, so actually we are going to have our first speaker—Pat Doyle, also of the Census Bureau—give those remarks in Chet Bowie’s stead.
DOYLE: Welcome; I’m really thrilled that you all could come today and share with us your expertise on what we believe to be a very pressing set of issues. Basically, our task for the two days is to address our rather overzealous entry into computer-assisted personal interviewing (CAPI).1 Automation is a wonderful thing; it allows us to basically take instruments to the point beyond which we can comprehend them. And we have certainly taken up that challenge.
We have instruments that provide a great deal of precision in measurement, and that’s excellent for the quality and the statistics we can produce from our surveys.2 It’s also allowed us—when we try—to reduce the burden by targeting our questions precisely to the individuals [to] whom the questions would be relevant.
But all of this comes at the cost of complexity, and that complexity complicates our comprehension of the instrument. It complicates the testing of the instrument. It increases the time and the resources needed up front to get started. It prohibits an interpretable image of the instrument that’s being fielded, i.e., the questionnaire. And basically it did away with the free good of instrument documentation. When we were in paper, we had a free good; we had documentation.
Now, we don’t really believe [that] the testing challenges we face are new; we think they have been faced in other disciplines. And what