STABILITY OF BOUNDARIES BETWEEN RESPONSE OPTIONS OF RESPONSE SCALES
Does ‘very happy’ remain equally happy over the years?


Tineke DeJonge, Wim Kalmijn, Ruut Veenhoven and Lidia Arends
Social Indicators Research 2015, 123 (1) 241-266, ISSN 1573-0921, DOI 1007/s11205-014-0735-2, Online since September 7 2014

Abstract
The differences between response scales in number and wording of response options make it hard to compare data from survey research and to perform research syntheses. A recent method that we have developed to tackle this problem is rooted in the idea that the transition points on a bounded continuum, on which verbal response options from a primary scale transit from one point to another, for instance from ‘happy’ to ‘very happy’, remain unchanged over time. The idea behind this is that although people may change their perception of, for example, their own happiness intensity over time, they are assumed not to change the degree of appreciation they attribute to the terms used to label response options. This is an important assumption for research syntheses that requires that everything remains unchanged, except for the change of interest. It means that if our method is applied to measurements at distinct points in time, differences in estimates of the mean and standard deviation can be attributed solely to changes in the frequency distributions on the primary scale. In this paper we apply the method to happiness and show that it is reasonable to assume that the transition points between the response options are stable over time.

Keywords: verbal response scales; comparability; scale transformation; beta distribution; reference distribution; research synthesis

full text