3 ms·

> it seems at least plausible from the data that much of the difference in age groups (and, for that matter, partisan affiliation) could disappear if controlle

> it seems at least plausible from the data that much of the difference in age groups (and, for that matter, partisan affiliation) could disappear if controlled for different identification rates among people's actual peer groups This assumes that homosexuality is evenly distributed across age groups, something that is directly contradicted by the data. Your bias makes you think that this is caused by unacceptability of identification, but you would need to show that the change in acceptability has only affected younger generations. As a counter-example, people who have worn LA lights sneakers are disproportionately under 30. It has become culturally acceptable to wear LA Lights sneakers over the last 15 years, perhaps older people are under-reporting their wearing of the sneakers? > "How many Americans have ever cheated on a test?" and asking another set "Have you ever cheated on a test?" and then publishing an article about how "Americans Greatly Overestimate How Many Americans Have Cheated on a Test" This metaphor seems directly contradicted by: > the change in the acceptability of identifying as LGBT over the last 70 years is pretty profound, The change in ever cheating on a test over the last 70 years does not seem pretty profound to me. It's clear that your personal bias is that there is an iceberg of non-identifying homosexuals who still feel oppressed. Even in SF, the number barely reaches 15% [1] 1: https://en.wikipedia.org/wiki/LGBT_demographics_of_the_United_States#By_locality https://en.wikipedia.org/wiki/LGBT_demographics_of_the_Unite...

abathur 7y ago
Sigh. I probably didn't phrase something clearly enough. Or perhaps we're not occupying the same universe. Since I was feeling free to impute sloppiness to Gallup last night, I'll do my best to take your imputation of bias in stride. > This assumes that homosexuality is evenly distributed across age groups, something that is directly contradicted by the data. Your bias makes you think that this is caused by unacceptability of identification, but you would need to show that the change in acceptability has only affected younger generations. Exactly the opposite. It is possible that a significant fraction of the age correlation the poll observed might disappear if they controlled for different identification rates at different ages. They don't try to use the data to investigate the possibility that the rates young people guess and the rates old people guess are actually pretty similar (even if over-estimated!) relative to identification rates in their peer group. It directly assumes the rate (at least of identification) can differ. > This metaphor seems directly contradicted by... The change in ever cheating on a test over the last 70 years does not seem pretty profound to me. The point of the metaphor isn't a time-correlated rate change. It's that the narrative of the release (Americans don't know Americans) is trying to compare quantities that might be expected to differ (because they measure different things) and then making news out of the fact that they aren't the same. > It's clear that your personal bias is that there is an iceberg of non-identifying homosexuals who still feel oppressed. Even in SF, the number barely reaches 15% [1] I have no strong opinion on or suspicion about what portion of the population does or doesn't identify as gay. If anything, I think the entire project of ontologizing our sexuality is fraught.

abathur 7y ago

Sigh. I probably didn't phrase something clearly enough. Or perhaps we're not occupying the same universe. Since I was feeling free to impute sloppiness to Gallup last night, I'll do my best to take your imputation of bias in stride. > This assumes that homosexuality is evenly distributed across age groups, something that is directly contradicted by the data. Your bias makes you think that this is caused by unacceptability of identification, but you would need to show that the change in acceptability has only affected younger generations. Exactly the opposite. It is possible that a significant fraction of the age correlation the poll observed might disappear if they controlled for different identification rates at different ages. They don't try to use the data to investigate the possibility that the rates young people guess and the rates old people guess are actually pretty similar (even if over-estimated!) relative to identification rates in their peer group. It directly assumes the rate (at least of identification) can differ. > This metaphor seems directly contradicted by... The change in ever cheating on a test over the last 70 years does not seem pretty profound to me. The point of the metaphor isn't a time-correlated rate change. It's that the narrative of the release (Americans don't know Americans) is trying to compare quantities that might be expected to differ (because they measure different things) and then making news out of the fact that they aren't the same. > It's clear that your personal bias is that there is an iceberg of non-identifying homosexuals who still feel oppressed. Even in SF, the number barely reaches 15% [1] I have no strong opinion on or suspicion about what portion of the population does or doesn't identify as gay. If anything, I think the entire project of ontologizing our sexuality is fraught.