Hear no numbers, see no numbers: math and the partisan mind

in Uncategorized by

There’s a famous book titled “How to Lie with Statistics” — but according to a breaking study, lying is only half of it: partisans, it turns out (and particularly those who are good at math!) interpret numbers as they want them to be, not as they really are.

An experiment conducted by researchers from Yale, Ohio State, Cornell, and Oregon, finds that people who are otherwise good at math are far more likely to lose their reasoning abilities when calculating solutions that run contrary to their political beliefs.

A total of 1,111 participants engaged in the study, which began with a series of questions to gauge mathematical reasoning ability, referred to as “numeracy”, and political orientation.

Participants were then asked to solve a problem that required interpreting the results of a mock scientific study.

For all participants, the data itself was identical. Some participants were told that the numbers represented the effectiveness of a new skin cream, while others were told it described the effectiveness of a ban on concealed handguns.

Half of the “skin care” group received data suggesting that the skin cream worked, and half of the “handgun” group saw data suggesting that handgun bans are effective.  The other halves saw numbers suggesting the opposite.

Participants’ accuracy in interpreting their given data was way, way different between the skin care and handgun groups, yet in terms of numbers, everyone was tasked with answering the identical problem.

Even more interesting, it wasn’t the people who stink at math who had the greatest difference in performance.  For people low or average in “numeracy” it didn’t much matter which topic they received.  In other words, for people bad at math, political beliefs didn’t take over where math skills left off.

Rather, the study found, people who are good at math are the most susceptible to letting political beliefs skew their logic.

Regardless of political orientation, people who scored high in numeracy were all far more likely to get the skin care question correct.  Yet when the same numbers were said to relate to guns, things got squirrely.

Highly numerate liberal Democrats scored nearly perfectly when given a dataset showing that gun bans work — but only about half got the correct answer when the numbers showed increased crime following a gun ban.

Likewise, about 80 percent of conservative Republicans answered correctly when given data that gun bans don’t work, while about half answered correctly when the numbers showed a gun ban working.

2013.09.11 numeracy partisanship 

The study’s lead author Dan Kahn thinks the results reflect a blind spot that partisan people have when identifying results that undermine their views.

But no matter the case, can we all agree that when it comes to partisan logic, sixty percent of the time it works every time?

Karen Cyphers, PhD, is a public policy researcher, political consultant, and mother to three daughters. She can be reached at karen@cyphersgroup.com.