What Must My Interest in Politics Be If I Just Told You "I Don't Know"?

Abstract

How much people think they follow what's going on in government and public affairs depends upon the context in which they are asked the question. If asked immediately after difficult questions about what they know of their congressman's record, they are much less likely to think they pay attention to public affairs than if they are asked, first, how they are in such matters. More important, data from two independent experiments show that this context effect cannot be eliminated, or significantly reduced, by interposing a buffer of questions on unrelated topics between the items that are known to affect one another. The authors discuss the psychological significance of these findings and their implications for survey research. George F. Bishop is Associate Professor of Political Science and a Senior Research Associate in the Institute for Policy Research at the University of Cincinnati. Robert W. Oldendick is Assistant Director and Alfred J. Tuchfarber is Director of the Institute for Policy Research, University of Cincinnati. This research was supported by a grant from the National Science Foundation (SES81-11404). The authors want to thank the anonymous referees for their helpful comments on revising the manuscript. Public Opinion Quarterly Vol. 48:510-519 ? 1984 by the Trustees of Columbia University Published by Elsevier Science Publishing Co., Inc. 0033-362X/84/0048-510/$2.50 This content downloaded from 207.46.13.120 on Wed, 14 Sep 2016 05:51:58 UTC All use subject to http://about.jstor.org/terms INTEREST IN POLITICS AND DON'T KNOW 511 search of their long-term memory for all pertinent political experiences and then give a considered response based upon that search. Nor do they appear to take even a representative sample of such experiences in saying, for example, that they follow politics of the or only now and then. Instead, the evidence indicates that they respond largely with the first thing that comes to mind from their most relevant, recent experience. And that, we find, depends heavily on the immediate context in which the question is asked.' In a previous experiment with this item, for example, we found that people were much less likely to say they follow what's going on in government and public affairs when asked about it immediately after, rather than just before, a difficult pair of questions about what they knew of their congressman's record (Bishop, et al. 1982). In another experiment, reported in the same article, we discovered that respondents, particularly the less informed, were more likely to see themselves as very interested in following the political campaigns (in 1980) when the question about campaign interest was posed after, instead of before, other items concerning the closeness and importance of the election. In neither case, moreover, did we contrive the experiment merely to demonstrate that such context effects could be created. On the contrary, we designed each exp-eriment to test hypotheses about the causes of an apparently precipitous decline of interest in politics at the time of the 1978 American National Election Study. Both experiments confirmed our hypothesis: namely, that the decline was due to changes in question order and context.2 The results thus have direct implications for the design of the biennial national election surveys. What can we do about the problem? Methodologically, the solution seems to lie in separating questions that are known or suspected to affect one another. Indeed, probably everyone who designs survey items assumes that by interposing a buffer of questions on unrelated topics, context effects can be minimized, if not eliminated. But can they? To our knowledge, this conventional wisdom has rarely, if ever, been tested systematically.3 Here we will describe the results of a I For a review of the literature on context effects, see Schuman and Presser (1981), Ch. 2; see also McFarland, 1981; Sigelman, 1981; Smith, 1982. 2 We do not wish to suggest that all of the apparent changes in political interest at the time of the 1978 election study were an artifact of changes in context. But the evidence we have presented elsewhere indicates that much of the change was indeed artifactual (see Bishop, et al., 1982). 3 Recently, we learned that Howard Schuman and his associates had, independently, conducted such an experiment on a different topic-reciprocity for Communist and American reporters (see Schuman, et al., 1983), showing results much like those demonstrated here. This content downloaded from 207.46.13.120 on Wed, 14 Sep 2016 05:51:58 UTC All use subject to http://about.jstor.org/terms 512 GEORGE F. BISHOP, ET AL. split-ballot experiment designed precisely to test that assumption, a replication of that experiment, and its implications for survey re-