Edward L. Kain

Jumping on the Bandwagon with the Underdog: The Impact of Attitude Polls on Polling Behavior

An experimental design was used to examine the effect which previous polling information about candidate preference had upon subsequent polling responses. Subjects were given one of three types of information (Carter commanding a lead, Reagan commanding a lead, no information about current poll results) at two different points in time. This yielded nine information conditions. The data clearly demonstrate that both strength of attitude and candidate preference are influenced by knowledge of previous polling results. The patterns of both shift in attitude and switching candidate preference are explained in terms of an oppositional reactivity hypothesis. Stephen J. Ceci is a psychologist and Edward L. Kain is a sociologist. Both are Assistant Professors in the Department of Human Development and Family Studies, Cornell University, Ithaca, N.Y. 14853. Portions of this report were underwritten by a grant from the College of Human Ecology to S. J. Ceci. Public Opinion Quarterly Vol. 46:228-242 ?) 1982 by The Trustees of Columbia University Published by Elsevier Science Publishing Co., Inc. 0033-362X/82/0046-228/

The relation between family context and perceptions of parental efficacy

2.50 This content downloaded from 157.55.39.93 on Wed, 13 Jul 2016 05:53:02 UTC All use subject to http://about.jstor.org/terms IMPACT OF ATTITUDE POLLS ON POLLING BEHAVIOR 229 possible that the absence of any bandwagon or underdog effects is a result of the fact that candidate preference grows more stable as election day approaches. Thus, these studies of bandwagon effects sampled voter opinions at a point in time when the impact of information such as televised election returns would be minimal (Mendelsohn and Crespi, 1970; Navazio, 1977). A further concern with these studies involves the response validity of self-reports of voting behavior. Although two of the studies examined precinct rosters to verify the self-reports of going to the voting booth (Fuchs, 1966, Mendelsohn, 1966) there was no way to validate the self-reports of candidate preference when those votes were cast. A long research tradition has examined response error in surveys of voting behavior (Clausen, 1968; Dinerman, 1948; Miller, 1952; Parry and Crossley, 1950; Traugott and Katosh, 1979). The recent research by Traugott and Katosh (1979) indicates that a fairly large number of respondents misreported their behavior in the 1976 CPS National Election Study. This error was large both in terms of reports of voter registration (15 percent) and reports of voting in the election (14 percent). The substantial majority of misreporting was in the direction of claiming to have voted or registered to vote when this was not, in fact, the case. It appears that some of this misreporting, therefore, was a result of feelings of pressure to give a socially appropriate response. Two further findings of the Traugott and Katosh research are important to note. The authors suggest that for those whose vote could not be verified, there was a distinct preference for Carter, indicating a small postelection bandwagon effect. Also of interest is the finding that being interviewed prior to the election is clearly linked to levels of voter participation. Further, this effect is cumulative, so that respondents who were interviewed in more than one wave of the election study had successively higher rates of validated voter participation (Traugott and Kotosh, 1979: 370-71). Navazio (1977) has suggested that research using an experimental design can help assess the impact of previous information upon polling responses. He used questionnaires including four opinion questions which were sent to both an experimental and control group. The experimental questionnaire differed only in the fact that the questions were preceded by the most recent national poll results on those questions. The response differences between the experimental and control groups did not indicate either an underdog or a bandwagon effect when examined as a whole. When occupational groups were compared, however, consistent differences appeared. The national poll results had been strongly negative in evaluating President Nixon. This content downloaded from 157.55.39.93 on Wed, 13 Jul 2016 05:53:02 UTC All use subject to http://about.jstor.org/terms

Relationship of gender and marital status with symptomatology in psychotic patients.

Factors related to perceptions of parental efficacy—i.e., beliefs concerning the degree to which parents can influence the development of their children—are examined with a national sample of 3000 parents. Results of the national data indicate that parents with little education and/or low incomes are less likely than better educated, financially advantaged parents to believe that parents have a great deal of control over developmental outcomes. Controlling for education and income, white parents are more likely than black parents to believe that parents exert considerable influence on the developing child. Also examined are parents beliefs regarding effective parenting practices. Beliefs regarding effecting practices are related to parents’ perceptions of control over developmental outcomes.