One of the challenges in business surveys is to achieve sufficiently high response rates at reasonable cost. High response rates not only increase the number of cases available for analysis, but also provide some insurance against nonresponse bias. A response rate is typically defined as the number of completed surveys divided by the total number of sample cases (including partial surveys, but excluding ineligibles).

Because some respondents quit the survey before it is finished (“partial interviews”), one strategy to increase the response rate is to convert partial interviews to complete interviews. This also makes good sense from a cost perspective since drawing in new respondents is often more difficult (and costly) than keeping respondents going.

This presentation discusses some strategies to minimize the relative number of partial surveys. More specifically, the effects of questionnaire design will be discussed in light of “human computer interaction” (“usability”). The main thesis is that good design leads to higher quality responses and to more completed (and less partial) surveys. We will discuss “Internet survey paradata”. This is a tool that enables researchers to detect design problems and to optimize the design of the survey instrument. A number of design issues, such as progress indicators and error messages, will be discussed from the perspective of paradata.