As we enter the final hours of the 2012 Presidential Campaign, both candidates are executing a record breaking eleventh hour ad blitz – more money will be spent on more political ads directed at fewer people than ever before. Since the race remains close, there has been growing interest in the impact of political ads on turnout and vote choice. As John Sides wrote last week, these late advertisements are perhaps the most persuasive of the campaign.
Although, Mr. Obama is leading in the national polls by about one percentage point; Mr. Romney is currently out-advertising Mr. Obama in many “must-win” battleground states. Experimental studies suggest that the effects of advertising on turnout are small (2 to 3 points), however this margin could be the difference in a close election. Voters in crucial swing states (Ohio) are experiencing an avalanche of political ads – an average of 333 ads per day have been aired in Columbus, alone. But, are these ads actually being watched?
Forty years ago, if you wanted to avoid watching a commercial, you had to get up off the couch walk across the room and turn the dial. Today, many viewers have complete control of their media environment. Disinterested viewers can simply avoid commercials altogether by changing the channel, pressing the mute button, turning the television off, or fast-forwarding.
Knowing exactly who turns off ads, especially as the incidence rate of time-shifted viewing increases, is paramount to determining the size of advertising effects. After all, if a campaign pays to reach 1,000 viewers and only 200 actually watch the ad – and only about half of those recall anything about it – the real effect of the political ad is significantly diminished.
Given that we know little about how people behave when confronted with political ads, Simon Jackman, Jeff Lewis, Lynn Vavreck and I investigate ad-viewing behavior. We leverage individual level media exposure data, providing second-by-second records of viewers’ television and advertising exposure. Panelists were provided smartphones equipped with audio recognition software that digitally captured all of their television exposure over the course of the 2006 midterm campaign. These data enable an analysis of not only who watches political advertisements, but also what types of people “turn-off” ads and what types of ads are avoided.
To measure ad avoidance we analyze the programs an individual watched (during a time when a political ad was run) to see whether the individual actually watched the political ad during a commercial break. Specifically, we define an “ad avoid” as: watching the 5 seconds prior to the start of an ad (the “lead-in spot”) and subsequently not watching the political advertisement in full. We find that viewers avoid a little less than half of the political ads they encounter.
When it comes to avoiding ads is there anything unique about political ads? After all, if viewers simply skip all ads at similar rates, drawing inferences about political behavior is complicated. The figure above plots number of ads watched as a function of number of ads avoided (muted, fast-forwarded, channel changed) by ad type. Regression lines closer to the y axis indicate more ads avoided than watched, while regression lines closer to the x axis represent more ads watched than avoided. The 45 degree line indicates equal amounts of ads watched and avoided.
The takeaway: individuals are filtering the content they see during commercial breaks. The panelists in our study avoid 66% of the Geico Insurance commercials they encounter — so, could switching to Geico really save you 15% or more on…NEXT! In contrast, viewers avoid only 17% of the Apple commercials they encounter. Political advertisements fall in the middle – viewers avoid approximately 42% of the political ads they encounter.
What kinds of people avoid political ads? The figure below indicates that high and low engagement citizens rarely avoid political ads. Access to technology plays an important role here. Viewers ability to skip an ad is correlated with political engagement – in 2006, 67% of the highly engaged panelists had TIVO or DVR equipment, while only 28% of the least engaged do.
Rather, it is those viewers who are moderately politically engaged who avoid more than half of the ads they come into contact with. Since the low engagement voters probably will not turn out to vote anyway and the high engagement voters likely made up their minds long ago – it is bad news for the candidates that the group they most need to sway is the group most likely to censor them. Perhaps this is why they try so hard to reach them.