Jump to Any Race
Political Advertising|By Guest Authors Ken Goldstein & Harley Ellenberger, December 2, 2014

The votes have been counted in the 2014 elections and the results revealed there were fewer to count than many had expected. In fact, the proportion of Americans legally eligible to cast ballots in this most recent election was the lowest in more than seven decades.

Every fall of an even-numbered year, the most predictable media story is “the unrelenting negative campaign” and how Americans are supposedly turned off by the negative ads they see on TV. So it was inevitable that when 2014’s low turnout levels became clear, blame would be placed on the nature of the campaigns in general and the tone of the advertising in particular.

In one recent essay, the editorial board of the New York Times summed up the supposed problem and cause: “Over all, the national turnout was 36.3 percent; only the 1942 federal election had a lower participation rate at 33.9 percent. The reasons are apathy, anger and frustration at the relentlessly negative tone of the campaigns.”

But, what if the “shockingly” low turnout in the recently concluded elections was due to the simple fact that there was not much reason for many Americans to vote and—wait for it—that there were too few ads?

Between 36 US Senate races and 36 governor races, just about every state hosted at least one high-profile statewide contest, but most were not competitive. As Election Day approached, the Cook Political Report assessed only 12 of the 36 Senate races and 17 of 36 governor races as being in play (rated Toss-Up or Lean). Ultimately, more than half of all states—29—had no high-profile competitive statewide race on their ballots. This description actually exaggerates the number of eligible voters who resided in states with competitive races because many of the most competitive contests took place in the smallest states—such as Alaska, Arkansas, and New Hampshire.

When we look at the actual population of eligible voters in states (turnout and population data from Michael McDonald’s excellent website), almost two in three (63 percent) of voting-eligible Americans lived in a state that had no competitive statewide race in 2014. The list includes the three largest states in the country: California, New York and Texas. Just over one in four (27 percent) lived in a state with at least one competitive statewide race; only 11 percent of voting-eligible Americans lived in a state with two competitive statewide races in 2014. And while every congressional seat was on the ballot, only 27 out of 435, or 6 percent, were rated as truly competitive by the Cook Political Report.

In the 21 states that held a competitive high-profile statewide race, voter turnout was 43.2 percent. In the 29 states without one, turnout was 33.3 percent. (These figures represent total turnout in each group of states divided by total eligible voters). Comparing the average by state, unweighted by population, turnout in the states with a competitive statewide contest was 44.5 percent; average turnout in states without a competitive race was 35.7 percent.

Turnout, of course, is not only a function of competitiveness but also a function of demographics, political culture, and election laws of a state. Particular states tend to have higher turnout (Minnesota and Wisconsin) or lower turnout (Arizona and West Virginia) no matter the year or political environment. Although it was the case that the 21 states with big competitive races in 2014 also, on average, had higher turnout in 2010 and 2012, the difference was more pronounced in 2014. And the drop-offs from 2010 and 2014, on average, were less pronounced. In fact, the 21 states with big competitive races in 2014 actually experienced no drop-off from their 2010 levels, while those without big competitive races dropped by 6 percentage points. The 2014 “21” also saw less of a drop off from the higher-turnout presidential year than the 2014 “29.”


Not Competitive (29)Competitive (21)
Turnout ‘1042%45%
Turnout ‘1257%61%
Turnout ‘1436%45%
Diff ‘14 to ‘10-6%0%
Diff ‘14 to ‘12-21%-16%

What about negative advertising, the supposed culprit behind the drop in turnout? Using data from Kantar Media CMAG, we calculated the market-weighted number of negative ad occurrences in every state. Since dollars are not comparable as a measure of intensity across states, and spot occurrences within a state are influenced by the number of media markets in that state, the market-weighted average provides a comparable metric. The 10 states with the most negative ad occurrences in Senate and Governor advertising from July 1 to Election Day in 2014 were:

1. Colorado
2. Arkansas
3. Iowa
4. North Carolina
5. Michigan
6. Florida
7. Alaska
8. Illinois
9. Georgia
10. Wisconsin

The turnout rate in these 10 “negative” states was 43.5 percent compared to a turnout rate of 34.7 percent in the less targeted or more positive 40 other states.

The level of competitiveness and the amount of negative advertising in a state are of course related. Competitive contests lead to both more advertising and more negative advertising. Of the 10 states that saw the highest number of negative ad occurrences, all had at least one major statewide contest handicapped as competitive by the Cook Political Report and five of the states had both a Senate race and a governor’s race rated as competitive. States with at least one competitive big statewide race, on average, saw seven times as many negative spots as states that did not host competitive big statewide contests.

To control for pre-existing or organic levels of turnout and look at the independent effects of statewide contests and advertising, we also estimated a simple regression model. The model estimated statewide turnout measured as the percentage of eligible voters who cast ballots as a function of turnout in 2010 and 2012, as well as a dummy variable that took the value of one if the state had a competitive statewide contest. In addition, we included a variable on the level of advertising in the state (measured by spots aired weighted by media market).

Not surprisingly, the regression shows that the single largest influence on turnout in 2014 was turnout in previous elections. States that tended to have higher turnout in previous years had higher turnout in 2014. But, holding those levels of turnout constant, we see that competitiveness and levels of political spots also had a significant impact on turnout. With other factors held constant, a state with a competitive major statewide contest saw voter turnout almost 4 percentage points higher than states without competitive major statewide contests.

Furthermore, when it comes to advertising, even with competitiveness held constant, turnout increased by 0.3 percentage points for every 1,000 negative spots aired. In a state like Colorado, with a market-weighted average of almost 23,000 spots, negative advertising increased turnout by nearly 7 percentage points.

The reflex to blame campaigns and negative advertising for the decline in nationwide turnout in 2014 is not surprising. But the evidence points in the opposite direction. The decline in turnout was due to lack of competitiveness and the excitement, resources and yes, negative advertising that are functions of competitive races. All may not be well with our democracy, but if low turnout is a concern, campaigns are not to blame. In fact, the lack of campaigns is to blame. Competitive races spur negative campaigns and both stimulate voter interest and participation.

Ken Goldstein is a professor of political science at the University of San Francisco; Harley Ellenberger is vice president of analytics at Kantar Media Intelligence.