CIRCLE (The Center for Information and Research on Civic Learning and Engagement)
conducts research on the civic and political engagement of young Americans.
The Center for Information & Research on Civic Learning and Engagement

Estimating Youth Turnout from Census and Exit Poll Data

May 10th, 2013
Email to a Friend

According to our analysis of Census Current Population Survey (CPS) voting data released on Wednesday, youth turnout in the 2012 election was 45% for ages 18-29 and 41.2% among 18-24s. Those numbers represent declines compared to 2004 and 2008, although youth turnout was higher last November than it had been in 1996 and 2000.

Immediately after the 2012 election, CIRCLE released our own estimates of the turnout rate: 50% for ages 18-29 and 48% for ages 18-24. We used the following method to reach those estimates: multiplying the number of ballots officially counted by that date by the proportion of voters who were young (according to the National Exit Polls) and then dividing that result by the number of young citizens (from Census data). In previous elections going back to the origins of the modern exit polls in the late 1990s, this method had always closely tracked the results of CPS turnout estimates. The discrepancy is a new phenomenon in 2012.

Like other organizations that study voting, we will now switch to using the CPS, which provides the largest sample sizes and the longest and most consistent historical trends. That means that we now presume that youth turnout fell significantly in 2012.

It is important to note, however, that all estimates of youth turnout are subject to bias and error. The Census CPS is a survey of a random sample of Americans, conducted shortly after the election. This year, it suggests that 133 million Americans voted. Counts of the actual number of ballots cast in 2012 vary, but no one has estimated more than 129 million votes. If the people who are contacted for the CPS and agree to participate do not precisely represent the public, the sample may be skewed. Although the sample size is large, there can still be a systematic bias–and the bias can be different for particular demographic groups, such as young adults. Also, respondents may claim that they voted when they did not. They may even accurately recall voting, but if their ballots were not counted, there would be a discrepancy between the CPS turnout estimates and the number of votes.

CIRCLE’s method that uses the Exit Polls has several compensating advantages. It is much faster than the CPS, yielding an estimate as soon as most ballots have been cast. It does not rely on self-reports of voting, because everyone in the Exit Poll actually voted. However, the Exit Polls use a complex sampling method whose main purpose is not to estimate the ages of voters. If the Exit Polls report an inaccurate proportion of young voters, that would introduce error in our turnout estimates.

Therefore, CIRCLE’s analysis of the 2012 election will henceforth be based on the CPS, and our previous releases on the 2012 election are now superseded. But we remind readers that the CPS also yields an estimate of youth turnout that is subject to error.

2 Responses to “Estimating Youth Turnout from Census and Exit Poll Data”

  1. youth voting declined in 2012 « Peter Levine Says:

    […] surprisingly. The CPS data suggests that there was actually a decline. This CIRCLE blog post explains the […]

  2. Sorry Democrats, Marijuana Doesn’t Bring Young Voters to the Polls — (fascinating stats, charts, and numbers for the political junkie) | Gazelle Says:

    […] the CPS more than the exit polls. The latter aren’t designed to estimate the ages of voters, as the Center for Information and Research on Civic Learning and Engagement has pointed out. That’s not to say CPS estimates are immune to margin of error. But a third source of […]