Robert Richards and Dr. Christopher Karpowitz, Department of Political Science
Background Research, Hypotheses
A great deal of political research has analyzed how people form their opinions about candidates and issues. Two models, drawing heavily from cognitive psychology have emerged: the memory-based model (Zaller and Feldman 1992; Huckfeldt et al 1999) and the on-line model (Taber, Lodge, and Glathar 2001). Though the mechanics differ, both models assume a person recalls some sort of information when forming or reporting an opinion. Thus, political opinions depend largely on recall of political information.
Recall of information is a topic that has received much treatment in the field of psychology (for example, see Eagly et al 1999). Thus, there is a good foundation for understanding the factors involved in recall generally. However, political science seems to have mostly ignored the concept of recall. Exploring this topic would allow political scientists to take advantage of the framework for understanding recall laid out in psychological research. I propose an experiment to examine recall rates for different kinds of information, to see which might be more effective in helping people form their opinions.
Hypotheses
According to research presented by Eagly and her colleagues (1999), attitudes affect recall of information, although the nature of these effects is still being debated. Much of this debate centers around whether so-called congeniality effects boost recall of information that supports (is congenial to) a person’s preexisting attitudes. The growing consensus seems to be that while congenial information is recalled more easily, so is information that opposes one’s views (Eagly et al 2001). Despite this ongoing debate, it seems reasonable to assume based on psychological research that information related in some way to a person’s attitudes would be recalled more easily. Indeed, political science shows that there is a link between preexisting opinions and attitude formation (Huckfeldt et al 1999, Lavine and Gschwend 2007), which, as discussed above, depends on recall. Thus, it seems that some types of information are more easily recalled than others. Biographical information is not as salient to a person’s views, and thus may not trigger any congeniality (or uncongeniality) effects. I initially hypothesized that people would have higher rates of recall for information on a candidate’s policy views than for biographical information. I also expected this effect to be magnified among those with strong partisan and ideological leanings.
Procedures
To test these hypotheses, I conducted a survey experiment on the February 2012 Utah Voter Poll (UVP), conducted by BYU’s Center for the Study of Elections and Democracy. A pretest section collected information on respondents’ issue attitudes and political sophistication. Information about other personal characteristics (partisanship, religion, race, gender, etc.) was also collected on the UVP.
The treatment involved showing respondents four items of information about a hypothetical candidate for the Utah House of Representatives named James Anderson. Respondents were not told the candidate was hypothetical until the very end of the survey. Three different kinds of information were presented randomly. In one treatment condition, respondents saw four items of biographical information about James Anderson. In the other two treatments respondents saw a list of four policy positions supported by the candidate. In one of these the positions were all typical conservative positions, and in the other the candidate supposedly held the opposite liberal positions. For example, one of the items dealt with gay marriage. In the conservative treatment, respondents were told that James Anderson favored a Constitutional amendment defining marriage as being between a man and a woman. In the liberal treatment, James Anderson supported an amendment protecting the right of homosexuals to marry. To keep all three treatments as parallel as possible, the items in the biographical treatment corresponded with the types of issues in the conservative and liberal treatments. For example, the biographical item parallel to the gay marriage issue was a statement that James Anderson is married with two children.
Near the end of the survey (after respondents had answered other unrelated questions on the survey), respondents were asked to report what they remembered about James Anderson. No other cues to recall were given aside from the name. After the data had been collected, I coded these open-ended responses to see how many items people remembered (correctly) about James Anderson, with a minimum of one item and a maximum of four. These coded values served as my primary dependent variable.
Preliminary Results
I am still in the process of analyzing the results of this study, but one finding is striking. The levels of recall in the biographical treatment were indeed different from that of the two issue treatments, but in the opposite direction of my hypothesis. In other words, people who were given seemingly mundane biographical information about the candidate remembered more than people who were given more ideologically charged information about the candidate’s policy positions. This treatment effect is statistically significant at the 0.05 level (the standard level of significance in political science), and was present even among many subgroups, such as among Republicans, conservatives, and those with high interest in/ awareness of politics. Issue attitudes also did not seem to affect levels of recall. The two issue treatments had statistically similar levels of recall.
Because the four items were parallel across the treatment groups, I could analyze the effects of the treatment on recall of specific topics. The data suggests that the type of issue matters in some cases. People in the issue treatment were less likely to remember James Anderson’s policy on job creation than people in the biographical treatment were to remember his career background. The same was true for James Anderson’s educational policy compared to his educational background. On the gay marriage/ family information item, people in the biographical and conservative treatments had similar levels of recall, while the liberal treatment alone had significantly lower levels of recall. The treatment effects disappear for the public lands issue item; people remembered that item at the same rate as people in the biographical treatment remembered his love of outdoor recreation. What these findings suggest is that different topics have different dynamics in terms of recall. Vague policies on job creation, which few voters really understand in depth, are not as likely to be recalled as personal information about the candidate’s career. This has implications for the way candidates run their campaigns, suggesting that they need to focus on helping voters get to know them rather than their policies if they want to be remembered. However, these results say nothing about how this translates into voting behavior.
Further Analysis
I plan to delve deeper into the data collected from my survey experiment. I would like to specifically dig deeper on the treatment effects for the individual items. The results of this further analysis could be presented at a conference or published in the future.
References
- Eagly, Alice H., Serena Chen, Shelly Chaiken, and Kelly Shaw-Barnes. 1999. “The Impact of Attitudes on Memory: An Affair to Remember.” Psychological Bulletin 125, no. 1: 64-89. PsycARTICLES, EBSCOhost (accessed October 13, 2010).
- Eagly, Alice H., Patrick Kulesa, Serena Chen, and Shelly Chaiken. 2001. Do attitudes affect memory? tests of the congeniality hypothesis. Current Directions in Psychological Science 10, no. 1: pp. 5-9.
- Huckfeldt, Robert, Jeffrey Levine, William Morgan, and John Sprague. 1999. Accessibility and the Political Utility of Partisan and Ideological Orientations. American Journal of Political Science 43, no. 3 (Jul., 1999).: pp. 888-911.
- Lavine, Howard and Thomas Gschwend. 2007. Issues, party and character: The moderating role of ideological thinking on candidate evaluation. British Journal of Political Science 37, no. 1: pp. 139-163.
- Taber, Charles S., Milton Lodge, and Jill Glathar. 2001. “The Motivated Construction of Political Judgments.” In James H. Kuklinski (ed.), Citizens and Politics. New York: Cambridge University Press.
- Zaller, John and Stanley Feldman. 1992. “A Simple Theory of the Survey Response: Answering Questions versus Revealing Preferences.” American Journal of Political Science 44:579-616.