In lieu of an abstract, here is a brief excerpt of the content:

3 A Brief Primer on Data and Research Design I N THIS SHORT CHAPTER, we describe the data used in the analysis for Chapters 4–6. (In Chapter 7, we switch the focus somewhat and leave the discussion of the methodology employed there to that chapter.) One of the key advantages of our approach in this book is methodological. Most existing research in the study of ad effects relies on laboratory experiments, which have been critically important in helping scholars understand the process by which ads might work (see, e.g., Ansolabehere and Iyengar 1995; Chang 2001; Kahn and Geer 1994; Meirick 2002; Pinkleton 1997, 1998; Valentino, Hutchings, and Williams 2004). It is very difficult, however, to make the transition from the laboratory to what happens in the real world of a campaign. Seeing one or two ads in quick succession just is not the same as seeing a multiplicity of ads from various sources over the course of a lengthy political campaign. To overcome this challenge, we rely on a combination of survey data and real-world ad-tracking data. We can, therefore, test the effects of exposure to multiple ads in multiple contexts (Senate and presidential races; competitive and non-competitive elections; open seats and races with incumbents). We can also look for these relationships at the individual level, which allows voters in the same media market, state, or county to have very different levels of advertising exposure. 38 Î Chapter 3 In short, we estimate levels of exposure to advertising for each respondent in our survey data, and that level will vary depending on how much television that person reports having watched and how many ads were aired where that person lives. This approach was used originally by Freedman and Goldstein (1999) to study the effect of ad exposure on turnout and has been used by them, and others, quite extensively to study a range of ad effects (Franz, Freedman, Goldstein, and Ridout 2007; Freedman, Franz, and Goldstein 2004; Martin 2004; Ridout and Franz 2007; Stevens 2009).1 Given our examination of a wide variety of real-world political races using this detailed measure of ad exposure, we believe our research contributes something important to the investigation of ad persuasion. Consider briefly the body of research on the persuasive force of negative campaigns. According to Lau, Sigelman, and Rovner (2007), as of 2007, sixty-one studies had been identified that investigate the relationship between negative campaigns (all-encompassing, including media and ad negativity) and persuasion. Of these, fortyone have used experiments (twenty-nine of which involved college students as participants). Of the other twenty, thirteen have been published , while seven were papers presented at conferences. Of the thirteen non-experimental published papers, none measured exposure in the way we do here, and only four looked for ad effects in either multiple years or multiple political contexts.2 Partly because of the data we employ, we believe this book offers the most comprehensive examination of ad persuasion effects to date. Of course, many scholars prefer to rely on experiments to test the influence of political advertising. The experiment-versus-survey debate is the classic fault line in the tradeoff between securing external validity (the ability to speak beyond the specific data used) and internal validity (the confidence one has in identified causal relationships). 1 Its usage has generated considerable and fruitful debate within the field of political advertising . For a longer discussion of the measure’s validity, see Ridout, Shah, Goldstein, and Franz 2004; Stevens 2008. For a more critical assessment, see Huber and Arceneaux 2007. 2 These were Bartels 2000; Fridkin and Kenney 2004; Kahn and Kenney 2004; Lau and Pomper 2004. The empirical research on ad persuasion more generally (without regard to ad characteristics) is a bit more limited in volume but has the same reliance on experiments or aggregate-level models: see Huber and Arceneaux 2007; Johnston, Hagen, and Jamieson 2004; Shaw 1999, 2006. A Brief Primer on Data and Research Design D 39 We believe it is paramount that questions about advertising persuasion have high external validity, however. In plain terms, knowing the influence of ads in actual campaigns—and in multiple contexts— speaks more strongly to the place of advertising in contemporary American elections than testing the marginal impact of a few ads in highly artificial contexts. We begin by briefly listing the data sources used in the three chapters that follow. We then describe how we estimated ad exposure and describe the model...


Additional Information

Related ISBN
MARC Record
Launched on MUSE
Open Access
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.