local areas in the original design of the study, but it did not prove feasible for him to include that material in the analysis in his book.18)
The Patterson study made a significant contribution to the systematic study of the media's role in elections but for this discussion of methods, a couple of limitations in that study should also be reviewed. The first is the omission of any substantial analysis of political ads from the study (which was surprising since Patterson's previous book had compared the content of ads with that of TV news in the 1972 election19). The lack of such analysis, and the lack of questions on the public's responses to ads, makes it impossible to be certain whether some of the impacts he found were due entirely to news or whether ads actually accounted for some of those impacts or produced others not asked about. (This is not intended to cast general.doubts on the findings of that study; the evidence and logic of much of Patterson's findings are so powerful that they have, indeed, become a foundation of this field of study.) The second limitation is that the study did not specifically analyze the visual dimension of TV news.
Landmark Study II: The Intensive Approach by Graber
The second landmark study that provides an important lesson in basic methods is Doris Graber's Processing the News study. Inspired by a study done years before by Robert Lane,'0 Graber sought to delve more deeply into how people received and mentally processed campaign communications than was possible using a large-scale survey approach. She used a more intensive approach, which provided a more thorough exploration of the way people "tame the information tide" of election communications and greater insight into the patterns of their political thinking after having been exposed to election messages. It also yielded information on these citizens' "psychological, social, and informational settings."21
Graber's study involved 21 registered voters of varying demographic char¬acteristics and of varying levels of interest and access to news media. She con¬ducted in-depth interviews with each of these people, averaging two hours each, ten times during a presidential election. The interviews were "arranged on short notice, allowing little time for respondents to prepare themselves for anticipated questions,"'- thus minimizing artificially induced attention and knowledge levels. Importantly, the interviews were conducted in the respon¬dent's home or place of business, instead of in a university setting, so that the citizen would be in a more "natural" setting; this also minimized artificial responses. The questions were not those typical of opinion surveys wherein the researcher uses her own wording and categories for the questions-and-answer options (closed-ended questions); rather, they were open-ended questions to allow the citizens to use their own words and categories to respond to the
276
The Media in Elections: Methods ot Study
information and impressions they were garnering and the meaning they were assigning to that information.
Besides asking about peoples' understanding of various issues and candi¬dates, she also asked questions that tested their recall of specific news stories during that time, along with questions about what news media were used in what frequency. Questions about the respondents' backgrounds, past experiences, and current social setting were also asked. Some questions were asked repeatedly in the same form in successive interviews, and some of the same questions were asked in different form. This enabled her not to mistake an atypical answer, stimulated by a unique setting, condition, and timing, for a "normal" one; it also enabled her to detect an answer that masked a thought pattern (such as a racial bias) and/or an answer that was designed to tell the researcher what the respon¬dent thought she wanted to hear, instead of the respondent's own thinking. Finally, Graber had a majority of the respondents keep daily diaries in which they reported on "what they remembered about news stories that had recently come to their attention" and their reactions to those stories—after a lag time to allow for some of the forgetting that occurs in normal life circumstances.'*
The results of this study were a very significant contribution to our under¬standing of how people receive, process, and make meaning out of election communications. As Graber notes, however, there are some cautions and inher¬ent limitations in the in-depth interview approach. First, although the study obtained a good range of demographics, political interest, and news access in the respondents, it is still the case that only 21 citizens from one local area in the nation cannot accurately represent all people. Correspondingly, one must be cautious about the generalizability of the findings. Second, the sensitization problem clearly can exist in this method as well—particularly with the use of daily diaries, which would carry an even larger potential to artificially heighten the attention level of respondents. To test for sensitization, Graber periodically interviewed people from the original pool of 200 from which the final 21 people were selected; she found little evidence of sensitization. Even so, this is an important factor to monitor. Finally, in an interview situation of this length (2 hours) and intensity, the personality and speaking pattern of the individual interviewers can affect the nature of the response and engender a noncharacter-istic response from the citizen (it is also generally helpful to have an interviewer of the same race as the respondent to maximize comfort and trust), as can changes in question wording and sequence.
Further Steps in the Comprehensive Study
of Media and Elections: The "Democracy '92" Study
Political scientists Marion Just, Ann Crigler, Timothy Cook, Darrell West, and the author of this book, along with communications scholar Montague Kern,
277
Chapter 10 The Media in Elections I
sought to make further progress in developing methods of studying the media in elections and to apply their methods to a study of the 1992 presidential election. The book from that study is titled Crosstalk: Citizens, Candidates, and Media in a Presidential Campaign.2* They were inspired by and sought to combine the best of the Patterson and Graber approaches, as well as adding refinements of their own. They also placed their study, which was called the "Democracy' '92" proj¬ect, in the broader context of one of the main enterprises in political science: election studies (not just media and election studies).
Election Studies and the Role of Theory The Democracy '92 study team noted that political scientists have found four basic factors to be the principal shapers of the public's voting decisions: (1) the political party people identify with, a long-term factor that parents pass on to children; (2) policy issues of general concern; (3) retrospective assessments of incumbents' performance in office; and (4) candidate evaluations, peoples' evaluations of the candidates' qualifications for office, leadership capacities and personal qualities.
The research team noted that most of the leading studies of elections over the years had focused on one or the other of the first three of those factors. Thus, the landmark study by Campbell, Converse, Miller, and Stokes, The American Voter,23 employed a social-psychological approach and discussed how partisan identification was a long-lasting orientation that helped rather poorly informed people sort out election information and judge candidate options by reference to the party cue. In the early years of voting studies (the 1950s and early 1960s), while people still largely identified with the parties, this theory provided a powerful explanation of the vote. But the parties' role in the Ameri¬can system declined through the following years, and more and more questions were raised about how to best explain peoples' voting decisions. Nie, Verba, and Petrocik, in The Changing American Votei;26 thought they had found evidence that peoples' focus on policy issues had significantly increased starting in 1964, but later critiques suggested there were problems in the methods they used and most studies continued to find low levels of public knowledge and understand¬ing of policy issues. Finally, Fiorina's Retrospective Voting in American National Elections2' found that people, rather continuously, evaluate the performance in office of the parties in general and especially the president (focusing on that more than on promises for the future). Fiorina's study and some others28 also presented evidence on how assessments of the state of the economy are particu¬larly important in voters' evaluation of incumbents' past performance.
There was little focus in these studies on the candidate evaluation factor in voting decisions. This was surprising since a close review of the data in both The American Voter and The Changing American Voter provided evidence of major impact of the personal appeal of presidential candidates. For example, from examining the expressed "likes" and "dislikes" of the candidates in National
278
The Media in Elections: Methods ol Study
Election Study data, Nie, Verba, and Petrocik, found that citizens' references to the personal characteristics of candidates were the most frequent type of response and had been consistently high in election after election.29
Schema theory (which we discussed in Chapter 2) certainly expanded the social-psychological approach to more fully analyze the way people mentally process political communications and make sense of the political world. But the Democracy '92 team thought, in general, a richer—yet realisticconception was needed of the citizen's role in elections and in the reception of and response to political communications. The team focused on the construction of candidate images in peoples' minds—images in a rather inclusive s