|
How Should We Measure Media Exposure?
Editor's Note: Scott Althaus and David Tewksbury recently completed an ANES pilot study report on a new set of media use measures that were tested in the 2006 ANES pilot. They (and I) thought that PCR readers would appreciate the opportunity to (1) hear about the results of the pilot study report, which is available on the ANES web site but is otherwise not being distributed to the political communication community; (2) comment on the strengths and weaknesses of the measurement strategy that we propose in the pilot study; and,(3) brainstorm on additional measures of media use that would be of general benefit to the political communication community.
This third benefit is perhaps the most important, because the new “Online Commons” format used by ANES empowers the broader research community to lobby for the inclusion of specific items. The next issue of PCR will feature a commentary from Scott and David and a roundtable of essays on this topic.
Below you will find the executive summary of the report. The full report can be obtained here. We encourage you to read through these materials and offer to participate in next issue's roundtable. To nominate yourself contact David Ryfe directly.
The executive summary of Althaus and Tewksbury's report now follows:
Executive Summary
The degree to which people seek and retain information about politics is a key variable for understanding why people think, feel, and act as they do politically. But measuring information acquisition has proven to be fraught with challenges. As a consequence, in recent years political scientists have shifted their measurement strategies to focus on information retention, most commonly in the form of factual knowledge questions. Interest in this approach has grown so much that some political scientists have begun to question whether the traditional media exposure measures are still worth asking. We argue that while the existing ANES media exposure measures may be problematic on methodological grounds, it is important to continue asking questions about the process of information acquisition. A measurement strategy based on information retention, we contend, requires survey instrumentation that is election-specific and unlikely to be valid over long stretches of time. The resulting problems of longitudinal continuity make this approach unsuitable as a stand-alone measurement strategy for the ANES.
The current battery of media exposure items can be traced back to the 1980 and 1984 ANES surveys. Many items had been introduced in 1980 and refined in 1984 to improve upon the variety of media use measures that had been employed up to that time without much consistency or continuity (Traugott 1985; cf Erbring and Clark 1979). The validity of these items was rigorously tested for the first time in the 1989 ANES pilot study (Price & Zaller 1993) and the items were subsequently updated after the 1995 pilot (Bartels 1996, 1996; Buhr, Crigler et al. 1996; Zaller 1996). In recent years, additional media use questions have been added to account for the growing importance of the Internet and local news as sources of information about national political campaigns. However, our review of the available pilot study reports, technical reports, and methods publications detailing these media use measures suggests that there has been no overall reassessment of the purpose for and basic measurement strategy underlying this battery of questions since it was first introduced a quarter century ago.
Our consideration of the best ways to measure media use today and in the future raises serious questions about how we measure political information acquisition from mass media sources. We are concerned that the ANES has not kept up with important developments in our understanding of how people process information, and that it is not well equipped to react to changes in the media environment that are already happening today. Recent research has underlined the place of media use and interpersonal talk (as well as their interaction) in the mobilization of citizens (e.g., Nisbet & Scheufele 2004; Shah & Scheufele 2006), but the ANES may not be well equipped to assess how these patterns change over time. What is needed is a better way of measuring information exposure, one that not only addresses important methodological concerns with the existing media exposure and interpersonal talk measures, but more importantly one that can adapt to the changing media landscape without requiring changes in question wording or the addition of new questions. In other words, we believe the time is past due for the ANES to overhaul its existing strategy for assessing information exposure to bring it in line with advances in our understanding of the psychology of information acquisition as well as to better match the changing media landscape of the 21st century.
Adapting to a Changed Media Environment
The 2004 American National Election Studies contained a battery of eight media exposure items. These traditional items suffer from two important limitations. First, because these items have been added and adjusted at different points in time over the last 25 years, the questions for different media are often inconsistently worded and thus difficult to compare.[1] <#_ftn1> Second, the existing media exposure questions are unevenly distributed across the contemporary media environment, with a heavy emphasis on newspapers and television programs but little coverage of the Internet and radio news sources that attract such large audiences today.[2] <#_ftn2>
The 2006 ANES pilot study featured a new set of media exposure questions designed to replace those used in prior ANES instruments. These measures included questions assessing the number of days in a typical week and minutes per day that Americans obtain news from newspapers, television, the Internet, and radio.
Key findings from the 2006 ANES pilot study:
- Extending the traditional “how many days per week” questions to include not only television and newspapers but also radio news and the Internet yields a superior map of media exposure patterns compared to traditional ANES questions. The average pilot study respondent reported seeking news sources on the Internet 2.5 days per week, reading a newspaper 3.7 days per week, watching television news programs 5.2 days per week, and listening to radio news 3.0 days per week.
- Measuring the number of minutes per day that respondents report using a news medium adds no new information beyond that obtained from measuring how many days per week a respondent reports using a news medium.
- None of the standard measures of respondent demographics, knowledge levels, or party identification is a strong predictor of self-reported exposure to the four news media considered here.
The traditional focus in National Election Studies on newspaper and television news exposure may help to explain why previous analyses tended to conclude that news exposure had little unique explanatory power once political knowledge levels are controlled. Our analysis replicates this general pattern for newspaper and television news exposure, but also finds that Internet and radio news exposure have unique effects on knowledge acquisition, perceptions, and frequency of political talk even after controlling for levels of political knowledge. This finding is likely a result of the increased content and audience diversification in the news industry that is found in 2006 relative to previous pilot study data collection efforts in 1995 and 1989.
Integrating Current Theories of Information Processing
The new media options available today have segmented citizens into media products generating different types of political content. Not only have the content options on cable television differentiated along ideological lines (Pew Research Center 2004), but the radio and online news environments are rapidly changing as well, with increased diversity of information flows as a result. Understanding where and how people learn about politics today may be just as important as knowing what they learn, because we now understand that the effects of campaign learning are conditioned both by cognitive processing goals and by the likelihood that citizens actively consider and reflect on the information they find, particularly in the context of interpersonal conversations about politics. Thus, there appears to be a pressing need for the ANES to assess how people process political information as they are exposed to it.
Kruglanski and colleagues (Kruglanski, Webster, & Klem 1993; Webster & Kruglanski 1994) developed and validated a Need for Closure scale with several subscales for tracing the impact of processing goals on a variety of judgmental tasks. Decisiveness and closed-mindedness are two subscales that seem particularly well suited for assessing individual-level processing goals in the ANES. Both represent defensive processing goals and should therefore be associated with increased polarization of perceptions about candidates and parties. High levels of defensive processing goals should reliably predict greater perceived issue distances between candidates and parties. Similarly, higher levels of elaboration likelihood should be associated with smaller perceived issue distances, as accuracy goals lead citizens to reflect at greater length and detail upon the campaign information they have received.
Single-item measures of decisiveness and closed-mindedness were included in the 2006 pilot, and a single-item measure of need for cognition was included in the 2004 ANES. Key findings from the 2006 ANES pilot study:
- Decisiveness was positively and significantly related to the perceived issue distances separating the presidential candidates and national parties, even after controlling for closed-mindedness, political knowledge and media exposure.
- Closed-mindedness had a marginally significant positive relationship with perceived party distance, and a nonsignificant but positive relationship with perceived candidate distance.
- Need for cognition was negatively and significantly associated with perceived issue distances separating the national parties, but the same relationship with perceived issue distances separating the presidential candidates fell just outside marginal levels of significance.
- Although none of the three variables was a consistent predictor of media exposure, need for cognition had a significant positive association with Internet news use, and closed-mindedness had a significant negative relationship with television news use
Recommendations
- Continue the use of self-reported media exposure questions along with questions that measure political knowledge, since each has unique effects on a range of dependent variables even when controlling for the other.
- Reconfigure the traditional battery of media exposure measures to better map the current contours of the media landscape by focusing on exposure to newspapers, television news, radio news, and news sources on the Internet.
- Standardize the measures of exposure to each of the four news media as days in a typical week.
- Add a new media exposure question that asks respondents to identify where they have been getting most of their information about the presidential campaign. Given content differences between newspapers, television news, radio news, and Internet news sources, prioritizing the relative importance of these media would allow the NES community to better assess the political consequences of exposure to particular media.
- Retain a measure of political discussion formatted to match the days per week scale of the media exposure measures.
- Consider adding longer scales that measure information processing goals (i.e., decisiveness and closed-mindedness) and need for cognition in order to further clarify the relationships of these variables to knowledge acquisition and a wide range of political attitudes, perceptions, and behaviors.
|
