Democratic Presidential candidates face a situation much like stepping in as a pinch hitter with two strikes against you; unless Ross Perot commits a catcher’s balk, things don’t look good. The problems, similar to those I faced in elementary school, are geography and math. A Democrat must nearly sweep the Northeast, the West Coast, and win major states in the Midwest. After defending their "home turf," our nominee must win two of the three large swing states: Florida, Ohio, and Pennsylvania. Simply put, as things stand every four years we play a dangerous game of chicken with Rev. Ronald Podhoretz-Falwell; and the odds are stacked against us.
This is the second installment of a three part series addressing the issue of national voting patterns. The first analyzed experiments and data underlying major theories of voting behavior. This diary will analyze media effects. The final episode will examine the political landscape of 2008 and comment on the Obama campaign with regards to the possibility of a Western realignment.
I encourage everyone to look at Part I; it's a fine read and you'll need it for Part III (and the end is the good part!)
The fundamental question that underlies political science is, quite simply, what matters in politics? The Kos community is founded on the idea that fundraising, campaign organization, campaign gaffes, campaign victories, and (even) the candidate him/herself play vital if not the constitutive role in electoral success. I imagine that many would be surprised to find, however, that this position is by no means universally accepted. Many political scientists believe that objective circumstances such as the economy, a sitting President's popularity, and incumbency actually determine the outcome of elections. The values of objective metrics such as economic indicators and presidential approval rating could then be measured and plugged into a formula to determine the outcome any election (political scientists have all sorts of fun running regressions to figure out better formulas).
This may seem somewhat, no, very silly. However, the results of these models are jaw dropping. In their article Why are American Presidential Election Campaign Polls So Variable When Votes Are So Predictable, Andrew Gelman and Gary King compare the performance of polls to forecast models from the same time period. The following table from their article shows the accuracy of various predictions in the months leading up to the 1984, 1988 and 1992 Presidential elections:
Gelman and King, 413:
The forecast models outperform the polls considerably. As an aside, I imagine some of you are wondering what forecast models predict for 2008. It is, generally speaking, to early for forecasts to be made because economic indicators are not yet in. However, Michael Lewis-Beck and Charles Tien, in a good article on forecast models project values on which to run two forecast models. You should really read the article but I'll spoil the end. In both models the Democrat wins; the margin is large in one and narrow in the other. However, in 2000 forecast models weren't just off, they were embarassingly wrong predicting, on average, that Gore would win 56% of the popular vote. Given this failure and the counterintuitiveness of the forecasting hypothesis we will assume that campaigns do matter and ask how?
In this diary I will focus on one aspect of this question; what effect, if any, does the media have on a campaign? I will look at this in the context of three different but related ideas; information reception, political persuasion, and media markets. This diary will be far less speculative than Part I. It will focus on specific experiments and features of the electoral landscape that illuminate the role of media in campaigns. I will not discuss media bias, the tone of news coverage or issue framing. I will instead focus on two main areas; media's effect on voter knowledge and vote preference. I have decided to leave media markets for Part III.
Media Variety and Differences in Learning
(or, everything your parents told you is wrong)
Before we head full on with an inquiry into how the media affects elections, it makes sense to ask to what extent and how voters retain information from media sources? In Citizens, Knowledge and the Information Environment Jennifer Jerit, Jason Barabos, and Toby Bolson examine the relationship between education and political learning. They begin by explaining that there is "knowledge gap" in current events, often in correlation with various socioeconomic factors, between those with different levels of formal education. Those who have less "background knowledge" about how (for example) government works will have greater difficulty processing and retaining information about government. Therefore they are less likely to benefit from new information. Jerit et al. demonstrate this by making a metric for the amount of news coverage various events receive and quantifying what percent of "high education" and "low education" voters are able to answer questions on that subject correctly.
Jerit et al., 273:
As the hypothesis suggested on issues with low levels of news coverage the percent able to provide correct answers was nearly the same between "high education" and "low education" voters. As the amount of media coverage increased, however, a gap emerged between the two survey groups. Whether it is possible to close this gap is a question of great importance to political campaigns. Therefore, Jerit et al. investigate whether different types of media effect the size of the knowledge gap. Using a complex regression they determine that, holding all other variables constant, while an increase in newspaper coverage greatly expands the knowledge gap, it remains almost constant with an increase in television coverage.
Jerit et al., 276:
This gives evidence for the somewhat surprising conclusion that television is, in fact, a more effective medium for increasing voter awareness. Even more surprising, research by Russel Neuman, Marion Just and Ann Crigler, published in the book Common Knowledge shows that television is, in fact, the most effective at educating voters about issues perceived as "distant" and "complex."
"The audience judged television more "attention grabbing" overall than magazines or newspapers. In experiments, a media's attention-grabbing quotient translated to more learning into stories that were low in salience- apartheid in South Africa, Star Wars [SDI]- or high in difficulty- the stock market crash." (Neuman, Just, Crigler 89)
They explain:
"Our experiments suggest that television's exciting, involving qualities and explanatory visuals break the attention barrier and make the personal relevance of an issue apparent. People with low initial interest, or who worry that the issue is too difficult to understand, can be drawn into a television story." (Neuman, Just, Crigler 92)
We see, therefore, that candidates- particularly those who want to educate voters on complex or distant issues- get the best result in terms of reception by using television. This is true even when all other variables are held constant.
Media Effects on Voting Behavior
(or, it's 3am, the phone rings; will you vote for who the TV tells you to vote for?)
Whether and to what extent the media actually influences elections is a difficult and disputed question. There is a plethora of seemingly contradictory information on the subject. For most of the latter half of the 20th century the predominant view was that media had only "minimal effects" because there was not a clear correlation between media metrics and voting behavior. However, the majority was by no means a consensus. For example, political scientists Shanto Iyengar and Donald Kinder contend that media has little direct effect on voter preference but has tremendous indirect influence through priming (associating specific issues and character traits with candidates) and agenda setting (establishing what issues are important for the election).
However, a breakthrough occurred with John Zaller's seminal work The Nature and Origins of Mass Opinion. His model held two factors determined the level of impact a media event will have on a voter; reception and acceptance(those familiar with the work would no doubt correct me that it is a three part model; the third is sampling which occurs when an opinion poll is taken. However, sampling is not relevant to the following discussion so it will not be discussed further). Zaller classifies voters by "habitual news consumption" which is measured through a test on current events. Reception, the probability that a voter will hear and understand given message is directly proportional to habitual news consumption; both because one who consumes a high amount of news media is more likely to hear a message and have the background knowledge to understand it. Acceptance, or yielding, is the probability one will be persuaded by a candidate's message, is inversely proportional to habitual news consumption. This is theorized based on the hypothesis that voters who consume a greater amount of news will have a larger basis of knowledge from which to pick up partisan cues and "counter-argue" against other positions. Thus, high news consumers are most likely to receive a message but least likely to yield to it. Low news consumers are the least likely to receive a message but the most likely to accept it.
From here I will focus on Zaller's 1996 essay "The Myth of Massive Media Impact Revived" from Political Persuasion and Attitude Change. In it, Zaller considers the impact of candidate's messages in a hypothetical two person race. Using his earlier theory, he posits that earlier studies on media affects have failed by looking for a linear correlation between media intensity and voter preference. He instead posits we consider "reception gaps," or the probability of receiving one candidates message but not the other.
Zaller, 30:
There are three assumptions. First, an incumbent will be much better known and have advantages that will give him/her a significant initial advantage in news reception. Second, increased media use/coverage for a challenger will result in increased media use/coverage for an incumbent as a campaign intensifies. If either of these assumptions fails the candidate with the greater media use/coverage gains a clear advantage (e.g. IN-08 in 2006). Third, increased news coverage is positive. Negative media about one candidate should be counted as positive media for the opponent. However, what is notable is that as campaign intensity increases the maximum of the difference function moves to the left. If the intensity continues to increase (it becomes a high intensity campaign) the function will achieve a maximum at x=0; voters with the least media use. This is important because incumbency advantage is proportional to the area under the difference function and voters with lower use are less likely to vote.
The second step, yielding, stands in the opposite relation to habitual media use; those with the greatest media use are least likely to be persuaded. However, as the above graph shows the function mapping the reception gap also decreases as habitual media use increases. Therefore, the step moving from reception to persuasion only reinforces the trends already established. He tests his hypothesis by examining three sets congressional elections.
Zaller, 36:
The data fits his model stunningly well. Media effects on voter preference, therefore, correlate to the "reception gap" between candidates. In other words, voters who hear an incumbent's message but not the challenger's are more likely to vote for the incumbent. More intense campaigns- those which receive (free or paid) media attention because they are in high profile districts, have a well known challenger, or for which prominent party members spend time to campaign- move the maximum of the persuasion function towards the origin; to those people who are least likely to vote. In these the incumbency advantage can disappear or even reverse. Therefore, the "minimal effects" thesis of the media's influence is thoroughly refuted. The media, in fact, plays a great role in influencing the outcome of elections by serving as a tool for candidates to promote their message.
A final remark should be made on the always important topic of money. Analysis of the 2000 Senate races done by Gary C. Jacobson found, not unsurprisingly, that spending by challengers correlated to success. Sufficient spending was able to close the recognition gap between an incumbent and challenger and even reverse the recall gap. In the context of Zaller's analysis we see that campaign spending is incredibly important.
Jacobson, 214:
In summary, what have we learned?
1.) Television is the most effective medium for increasing political knowledge; particularly those with less political education.
2.) Television can make voters aware of stories which seem too complex or distant
3.) The probability of receiving one candidate's message but not the other's is directly correlated to voter preference; and the intensity of the campaign determines what group of voters is most likely to receive only one message.
4.) Campaign spending by a challenger directly correlates to closing recall and recognition gaps
In the third and final installment of this series I will examine the political landscape of the 2008 election using the theoretical background from this diary and Part I. In it, I will suggest that the Democratic party, particular on the coattails of a 50-State Obama campaign, has the potential to induce a realignment in the Western States which would dramatically change the politics of the Electoral College.