More of us and more of them? Or, the perils of interpreting social research

[bi_janus]

Study: Polls may underestimate anti-gay sentiment and size of gay, lesbian population

http://www.pewresearch.org/fact-tank/2013/10/09/study-polls-may-underestimate-anti-gay-sentiment-and-size-of-gay-lesbian-population/

[Gee Whillickers]

Very interesting, but desperately needs corroboration and peer review, and maybe more controls? I'd really like to hear more about this study.

The study recruited participants online. How was this done? Sounds already like a self-selected group, which means biased results.

[Graeme]

The study recruited participants online. How was this done? Sounds already like a self-selected group, which means biased results.

From what I read, they allowed for that in their results. They're not claiming the percentages reported are indicative of national trends. What they're reporting is that when given a chance to respond anonymously, a higher percentage of people indicated that they weren't heterosexual, and a higher percentage of people indicated that they had homophobic views.

Given that it was the same self-selected group that gave both results, it was very interesting. The percentage differences appeared quite significant, too, so it wasn't a case of data mining to find unusual results.

So the result isn't what the percentages were, but the fact that a significant number of people are uncomfortable with saying that they're not heterosexual AND others are uncomfortable admitting that they have anti-gay views.

By the way, this would partially explain the Proposition 8 result in California. When surveyed before the election people didn't give anti-gay views, but when they had the opportunity to do so anonymously, more came forward to vote for Prop 8 than expected from the pre-poll surveys.

[Gee Whillickers]

From what I read, they allowed for that in their results. They're not claiming the percentages reported are indicative of national trends. What they're reporting is that when given a chance to respond anonymously, a higher percentage of people indicated that they weren't heterosexual, and a higher percentage of people indicated that they had homophobic views.

Given that it was the same self-selected group that gave both results, it was very interesting. The percentage differences appeared quite significant, too, so it wasn't a case of data mining to find unusual results.

So the result isn't what the percentages were, but the fact that a significant number of people are uncomfortable with saying that they're not heterosexual AND others are uncomfortable admitting that they have anti-gay views.

By the way, this would partially explain the Proposition 8 result in California. When surveyed before the election people didn't give anti-gay views, but when they had the opportunity to do so anonymously, more came forward to vote for Prop 8 than expected from the pre-poll surveys.

Okay, that makes sense. But as for the California Proposition 8 results, how does that explain the results from the votes in Canada and in other countries that were fairly strongly in favor of gay marriage?

[Graeme]

The votes in Canada and other countries were, I believe, by politicians and not by the general populous. Politicians also don't do secret ballets so this research isn't applicable.

Where this research comes into play is when the general populous is voting on gay rights. If the research is accurate, the pre-poll surveys would have over-estimated the support for gay rights, because of the percentage of people who wouldn't express their views in the survey, but would in the ballot box.

[ChrisR]

[...] Politicians also don't do secret ballets so this research isn't applicable. [...]

Perhaps the best Freudian keystroke I've ever seen!

[JamesSavik]

In my field research the incidence of same-sex sexual behavior seems to increase in direct proportion to alcohol intake.

More research is definitely called for.

A paper should be written and submitted to a scholarly journal on the subject.

Manhunt or the Advocate perhaps.