Beware of statistics


“Hispanics love Pawlenty! Poll shows 69% approval of Hispanics” reads the headline in another local publication. I couldn’t believe it. Why would the governor whose divisive and hostile stance on immigrants be so well-loved by Hispanics?

So I investigated. And I found that the statistics are suspicious if not downright misleading. 

The headline and article were based on a “tracking poll” conducted by an outfit called Survey USA. The poll was sponsored by television stations in Rochester (KAAL), Alexandria (KSAX), Duluth (WDIO) and Minneapolis (KSTP).

Survey USA polled 600 Minnesota adults and reported that overall, Pawlenty’s approval rating was 54 percent. Now, here’s where it gets interesting. The 600 adults included only 2 percent identified as “Hispanics.” Survey USA’s Web site said that was 14 individuals. So the statement that “Hispanics love Pawlenty” was based on a survey of 14 Hispanic individuals! Can a survey of 14 individuals provide a legitimate basis for conclusions about the opinions of a Latino population estimated at somewhere between 169,000 and 193,000 in 2005? 

Let’s look at some more numbers. Survey USA says it interviewed 14 Hispanics, or 2 percent of its entire polling sample. But then it reports 15 answers given by the 14 people. How does that add up?

Then there’s the matter of numbers. According to estimates from the Minnesota State Demographic Center, the population of the state is estimated to be 3.42 percent Latino in 2005. That means that Latinos are underrepresented in Survey USA’s sample by 50 percent. (The pollsters claim to have included 2 percent Hispanics in their sample, but they should have included at least 3.42 percent.)

Does this strange story of Survey USA mean that statistics are useless and unreliable? Not at all. Some statistics are useful, but these statistics were misrepresented and misused. There is no way that a poll of 14 people can represent the opinion of 193,000 people. To say that this poll represents the views of Hispanics in Minnesota is just wrong.

We do not know whether the people who wrote this article looked at the poll numbers before they wrote it. Maybe they did not understand the numbers. Maybe they didn’t care, but just wanted a headline. Maybe—just maybe—they intended to advocate for the governor and his anti-immigrant agenda. Without looking into their minds, we really do not know.

What we do know is that all of us—writers, editors, and readers—need to look carefully at statistics and sources, and to consider how reliable they have proven over time. I know that I will look very carefully and suspiciously at any future poll results from Survey USA and at any future reporting from the sources that funded and/or misused this poll. As my daddy told me years ago, “Figures don’t lie, but liars sure do figure.”