Skip Header

Working Paper Number SM1996-12
Wendy Davis, Tracy R. Wellens, and Theresa J. DeMaio
Component ID: #ti1831183238


The Center for Survey Methods Research conducted experimental research to evaluate the response scales used in the Diet and Health Knowledge Survey telephone questionnaire. Specifically, three characteristics of the response scale were examined. The length of the scale (i.e., the number of points on the scale), the extent of verbal labeling used, (i.e., partially labeled with only the end points labeled, or fully labeled with all points labeled) and the type of scale format used (i.e., branching with the direction and magnitude of an opinion collected in two questions; or standard with both the direction and magnitude collected in one question). The literature addressing these issues is often inconclusive. For example, several studies examining the type of scale format to use in a telephone interview have contradictory results. Some studies suggest that data quality is higher with a branching scales (Groves, 1979) and other studies suggest data quality is higher with a standard scale format (Miller, 1984). There are similar contradictions regarding the extent of labeling. There is more consensus in the literature regarding the number of points to use on a scale, but a range of points is offered as the optimal number of points (between 3 and 9 points) rather than a specific number of points. The study presented here examined each of these issues plus an additional survey characteristic currently used in the Diet and Health Knowledge Survey interview – providing respondents with a show card to refer to during the interview.

Back to Header