…don’t understand fractions. A recent Gallup poll asked people what percentage of Americans were gay and lesbian. The results? 52 percent estimated that twenty percent or more of the population is gay or lesbian:

Keep in mind that most estimates put the lesbian and gay population at around *three* percent. While the Gallup pollsters and other commentators ascribe this to increasing acceptance, I think it’s something more basic–a lot of people don’t really instinctively make the link between fractions and percentages:

The arithmetic gap is the most obvious one: profs over a certain age (and some immigrant profs) were drilled in mental math; Canadian students under a certain age haven’t been. Some implications of the arithmetic gap are familiar: profs who can’t understand why students insist on using calculators; students who can’t understand why their profs are so unreasonable.

But the mental arithmetic gap has more subtle implications. Mental calculations often require intuition about, and comfort with, the use of fractions. Pre-calculator: 1/3+1/3=2/3. Calculator era: 0.3333….+0.3333….=0.6666…. Pre-calculator: “To multiply by twenty-five, divide by four and add two zeros (25*Y=1/4*100*Y)” Calculator: Multiply by twenty-five. Back in the day, fractions were easier than – or at least not much more difficult than – decimals. Calculators make fractions obsolete.

I think the answers would have been much more reasonable–keep in mind *one of out three respondents* thought *25% or more* of the U.S. is gay or lesbian–had the question been asked in terms of fractions. It’s a lot easier to think, “I know X gay people out of Y” and answer that question directly than transform it to percentages–some people aren’t good at doing long division in their heads. If you look at how various demographic groups answered, the less educated and poorer respondents had much higher estimates of the gay and lesbian percentage of the population. Younger respondents also scored had much higher estimates; if Frances Woolley is correct, this shouldn’t be surprising either.

I think this odd result might have something to do with the way the question is phrased.

**An aside:** My guess if you asked people about American Jews, we would also be ridiculously overestimated (we’re about two percent of population, depending on how exactly one does the counting). I think once a minority is no longer denigrated or invisible (and those two states are related), that combined with a cognitive bias against choosing extremely high or low frequencies would yield overestimates. That still doesn’t explain so many people believing one out of five or more Americans are gay or lesbian…