(Steve Helber/AP)

Over at Vox, Danielle Kurtzleben reports on a new poll:

In a recent Ipsos-MORI poll, 1,001 Americans were asked, “Out of every 100 people of working age, how many do you think are unemployed and looking for work?” Their average response was 32. That’s almost 26 percentage points higher than the 6.1-percent jobless rate in August, when the poll was conducted.

There are two issues here.  First, this one single poll is not in line with the published research on Americans’ ability to estimate the unemployment rate.

In this 1986 article, political scientists Lee Sigelman and Ernest Yanarella report the results from a 1982 survey, conducted when the unemployment rate was 11 percent.  They found that a substantial majority, two-thirds, stated that the unemployment rate was 10 percent, 11 percent, or 12 percent — a substantial degree of accuracy.

In this 2014 article (earlier ungated version), political scientists Steven Ansolabehere and Brian Schaffner report the results of a 2010 survey, conducted when the unemployment rate was 9.7 percent.  They found that, depending on whether respondents took the survey on the phone or over the Internet, approximately 40-50 percent of respondents could estimate this rate within 1 percentage point.

In this 2014 article, Eric Lawrence and I report the results of another 2010 survey.  We asked people to estimate the unemployment rate, in a similar fashion to the Ipsos-MORI poll.  Here is the survey question:

The federal government defines unemployment to be when people do not have a job, have actively looked for work in the prior 4 weeks, and are currently available for work. What percent of Americans do you think currently are unemployed? If you do not know, you can just give your best guess.

Here are the results:


Although some people do give estimates that are far away from the truth, most respondents gave fairly accurate estimates — which is reflected in the median.

This research suggests that the public is not nearly as ignorant about the unemployment rate as the Ipsos-MORI poll suggests.

The second issue concerns this argument from Kurtzleben:

This all doesn’t just matter because people are off. It matters because the degree to which people perceive problems guides how they make political decisions.

Lawrence and I actually found that correcting erroneous estimates of the unemployment rate had almost no effect on public attitudes toward spending on student loans, aid to the poor, job training, unemployment benefits, or food stamps.

Of course, our study is just one study. Perhaps there are other kinds of attitudes that would be affected by giving people correct information about the unemployment rate.  Or perhaps there are other ways of providing this information that would have a larger impact.

Regardless, it seems premature to argue that Americans vastly overestimate the unemployment rate, or that the inability to estimate the unemployment rate really matters for political decisions.