Abstract
In survey methods research, the context effects of preceding questions on responses to survey items have been the subject of many interesting studies (see, for instance, Abelson, 1984; Bishop, Oldendick, & Tuchfarber, 1985; Bradburn, 1983; Carpenter & Blackwood, 1979; McClendon & O’Brien, 1988; McFarland, 1981; Perreault, 1975; Schuman, Presser, & Ludwig, 1981; Schuman & Presser, 1981; Sigelman, 1981; T. W. Smith, 1981c; and the chapters in this volume). In particular, effects of question order have been investigated. In most of these studies, the effects of only one or two preceding questions were considered, or the focus was on part-whole combinations of questions. Schwarz and his associates also studied the effects on responses when different ranges of response categories are offered (e.g., Schwarz & Hippler, 1987).
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsPreview
Unable to display preview. Download preview PDF.
Notes
Therefore, we prefer to use the term “skewness effect” instead of “frequency effect.”
Only the size of the projection screen in these experiments might serve as a maximum.
Low and high probabilities were determined from mean ratings in the preliminary study.
In a control condition with equally spaced stimulus values and a full range, one half of the 40 respondents were presented the “left” order of focal stimuli, the other half the “right” order. There were no significant differences in mean ratings for the focal stimuli between respondents with one order of presentation and respondents with the other order (none of the F values came even close to significance). Therefore, we may conclude that differences in ratings due to range-frequency effects in this study are not confounded with order effects within the set of focal stimuli.
The number of refusals was 49; the remaining 63 did not cooperate because of other reasons (mostly “not at home”).
In this analysis, the ratings on the 9-point scale of conditions (LS9 and RS9) are linearly transformed into a scale from 1 to 5 to make these ratings comparable with those in LS5 and RS5.
Comparison of LS9 with LR (or of RS9 with RR) is less obvious because of the main effect of Number of Categories described above.
We expected range-skewness effects to be stronger in our face-to-face mode (with visual aid) than in the telephone mode. The major reason for this expectation was “the ability to see an entire set of questions in mail and other self-administered surveys … and the lack of this ability in telephone surveys” (Tarnai & Dillman, chap. 9, this volume). However, we found no significant mode effects.
By “frequentistic” stimuli, we mean events that occur in a repetitive fashion and thereby permit specification of likeliness in terms of relative frequency of occurrence. By nonfrequentistic stimuli, we mean events that, while uncertain, are non-repetitive and unique (Howell & Burnett, 1978).
As part of a follow-up experiment, we interviewed 40 respondents (out of the same pool of volunteers and with the same demographic characteristics as those in the present study). We asked them to judge the dissimilarity between focal and contextual Stimuli. More specifically, we asked them to pick 2 out of 10 focal stimuli that they believed to be the least similar to a specific contextual stimulus (7 of the 10 focal stimuli were the ones common to all conditions in the present study, the 3 other ones were relevant to the follow-up study). Respondents had to perform this task six times: They compared the 10 focal stimuli with 6 contextual stimuli (randomly chosen out of the contextual sets: there were 4 different samples of 6 stimuli, 2 drawn from the “left” contextual sets and 2 from the “right” sets). It appeared that the “opinion” focal stimuli are chosen most often as the most different from the contextual stimuli. These 3 focal stimuli (numbers 3, 5, and 6) were chosen significantly more often than chance (the z scores corrected for continuity are 2.67, 3.96, and 6.38, respectively, all p <.01).
We had our focal stimuli preceded by series of 30 or 32 contextual stimuli (depending on experimental conditions), but psychophysical experiments show that range-frequency effects also occur in shorter series.
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1992 Springer-Verlag New York, Inc.
About this chapter
Cite this chapter
Daamen, D.D.L., de Bie, S.E. (1992). Serial Context Effects in Survey Interviews. In: Schwarz, N., Sudman, S. (eds) Context Effects in Social and Psychological Research. Springer, New York, NY. https://doi.org/10.1007/978-1-4612-2848-6_8
Download citation
DOI: https://doi.org/10.1007/978-1-4612-2848-6_8
Publisher Name: Springer, New York, NY
Print ISBN: 978-1-4612-7695-1
Online ISBN: 978-1-4612-2848-6
eBook Packages: Springer Book Archive