Asking People About Statistics

I don’t know much about polling, but wouldn’t people be biased toward thinking that the correct answer was somewhere in the middle? And yet 35% of people said “More than 25%.” I have to imagine if they’d offered answers in 5% increments all the way up to 100%, the results would have looked even more ridiculous.

Of course since the correct answer is so low, this bias would still lead more people to overestimate the figure…just maybe not by such a ludicrous margin.

> About 3.5% of Americans identify as gay, lesbian, bisexual or transgendered.

That’s not the same as the number of people who are gay.

People seem to be reading more into this than the poll question actually asked. Hence my earlier criticism.

If you ask me, “how many Americans today are X?”, you’re not asking me, “how many people do you know are X?”, “how many people in your community are X?”, or “how many people do you think self-identify as X?”

You’re essentially asking me to make an estimate of the “true nature” of people across the United States, and across age, class and race cohorts that might be well outside my experience, based on my own limited sample and experience.

Let’s give an instructive example.

a) “How many Americans would you say are regular viewers of pornography?”

b) “How many people do you know that are regular viewers of pornography?”

c) “How many people in your community are regular viewers of pornography”?”

d) “How many people would self-identify as regular viewers of pornography?”

I think we can all imagine that the average citizen’s answer to (a) will be much higher than their answer to (b), (c), or (d). The poll asked a question about homosexuality consistent with question (a), yet everybody is assuming that this “means” that people would answer (b), (c) or (d) with the same response as they would (a). If they said 30% of people across the US use pornography, they must think that 30% of people they know use pornography.

And that’s ridiculous, of course.

When you consider what many people think about homosexuality, the the comparison to pornography may be somewhat apt. People who know 2% or 5% of homosexuals in their lives will probably assume that there’s a larger community of people who really are homosexual but won’t admit it, as well as other populations in the US where homosexuality is more common. Indeed, they may be right, as the pornography question and sociology information suggest.

I think the problem comes in as people are being asked not to estimate how many gay people they know or how many gay people are in their community but rather the entire United States.

When confronting this, the responder not only has to understand what the numbers mean (i.e. 25% means 1 out of 4) but he also must estimate the demographics of areas of the United States he is unfamiliar with and their relative importance to the US total population.

Consider, for example, a midwestern farmer who knows very well there are few if any gay people in his small community. He knows that many large cities, though, do have sizeable gay populations. How large? He isn’t sure. Likewise he isn’t quite sure how much of the US population lives in large cities. So say he guesses. He guesses that 75% of the US population is in large cities and they are 33% gay. Well 33% of 75% and 0% of 25% happens to work out to 25% of the total US population. His math skills are perfect, he is just making incorrect guesses about the population distribution of the US.

well all i can say is that the Lesbian Population has certainly increased over the years which it is making it very difficult for us Straight Guys looking for a Good Woman to spend the rest of our life with.