The stereotype is accurate enough in that the areas of the country with Republicans are fatter, but it's not that accurate for individuals.
By the way, from the county study, "1% increase in county-level support for the Republican candidate corresponds to a 0.02% increase in age-adjusted obesity rates" (adjusted for their covariates). In other words, imagining two counties that are otherwise demographically identical, but one which is magically 100% Republican voting and one that is 100% Democrat voting, the difference would be 100 * 0.02%points, or 2%points. Basically zero.
If I had to guess, I would say New Jersey Republicans are probably the least fat conservatives (despite Chris Christie). Conservatism is associated with rural Christianity, whereas liberalism is associated with urbanism and education. Obesity is also negative correlated with intelligence.
I don't have any education in statistics, but the highest p-value is 0.049, which does look suspicious. I need some kind of AI tool that I can feed a study into and it will flag P-values for me.
"controlling for country of origin, age (using a spline), and gender did not change the results."
Country of origin? That's a strange thing to control for. What about race?
It's probably black and Hispanic fat people balancing out the fat white people, who are Republicans. It is sort of like IQ, where racial differences obscures differences among whites, which is what most people are thinking of. There is just no way on earth that there isn't a relationship among whites given what we know about how groups vote.
>Country of origin? That's a strange thing to control for. What about race?
Unfortunately, the ESS does not have a question that identifies people by race. I mistakenly wrote 'country of origin' instead of 'country of residence', which is what the variable was measuring.
The GSS after a certain year started asking people if they were Hispanic; after restricting the sample to only non-Hispanic Whites, the positive correlation still existed (r = .031) but was not still statistically significant (p = .10). Based on priors, the true correlation should be a bit stronger than that, though the observed association does betray how prominent the stereotype is.
If we are going to discount the lumpenproletariat that Dems turn out, should we not do the same for the GOP?
I'd be interested in seeing white obesity rate by ideology controlling for IQ and family size. I doubt liberals maintain an edge after accounting for these things. Though I have no doubt libs win the "skinny fat" category.
It's just very hard for me not to notice the ugliness of liberals versus conservatives amongst similar SES people.
When you do not have the data, do not analyse it. I do believe that fat Republicans and obese Democrats exists, and I do believe they are not the same group, but I cannot prove it: I do not have the data. And you cannot refute this claim without the data.
We have no reliable data on obesete and American political identification. However, we know that American political polls are sampling about 5% of the autoselective part of the population who are willing to answer surveys. Chances are high that body-shamed Republican men hardly get to those samples that do not record their weight, anyways.
The ESS survey is a red herring. It does not record political affiliation for 25 years, the left-right placement in European multi-party democracies is a very bad proxy for anything and useless for American political identification.
Now, it would be very interesting to know more about obese people, their identities, affiliations, because it would also help fighting obesity. However, to make your claims, you would have to overcome the following barriers:
1. You need to prove that obese people vote or identify politically. We do not know. Because obesity is widespread, we can assume that some do.
2. You need to prove that political polls include obese people, and if not, you would have to create a reliable sampling that includes them. It will be very costly, difficult, but worth the effort.
3. You would have to see if those socio-democraphic, psychometric, environmental variables that contribute to obesity are independent of political identities and party affiliation. I dare to say that you will not be able to that, it would contradict a large body of research on both obesity and political behavior.
I think it would be far more fruitful to get data on obese people, and try to find out why some of them affiliate Democrat and Republican. Because Democrats and Republicans are a quiet well separated group demographically, I think that you will find a culturally, demographically different obese Democrat and obese Republican electorate. My hypothesis is that the 'fat Republican' exists and it is a distinct group. Yet studying them would require first a qualitative analysis of how obesee people treat politics and political polling, then a good survey that perhaps even overweights this group, and then doing the analysis.
In the absence of the data, I would not analyse data, because that is only misleading.
No, I think that your uninformed data gimmicks cast negative light on survey scientists, political scientists, and public health professionals but arbitrarily mixing up completely unrelated dataset that are incompatible by design. No survey specialist would endorse the way you interpret and mix their data: retrospective survey harmonisation is hard. You seem to be uninformed about the ESS survey and the left placement variable that has a body of literature why it is ill-designed (no place on its scale for the centrist or median voter) and has no connection party affiliation. You disregard the differences in party affiliation in Europe & US, and absolutely have no clue on the non-genetic factors of obesity. What you do is that you invent a catchy heading that fits in well with the prejudices of many people, then mimick data analysis and come to a conclusion that is highly counter-intuitive. Call me pedantric, but at least I know how to work ethically and in a scientific manner with survey data.
It's interesting to see the lack of evidence. I think that stereotype comes from age. But analyzing self reported heights and weights strikes me as naïve.
Can you explain why you include the comparison before we adjust for covariates, rather than after? Wouldn't the bivariate logistic regression have a spurious coefficient?
The Brazilian stereotype of Americans is that they're often overweight. Even expats here tend to be noticeably heavier than most locals. Emil recently posted about IQ and body weight, suggesting there's no clear correlation between the two.
If you control for income it might be a stronger relationship. Poor low intelligence people can dodge obesity by not having enough to spend on extra food
The stereotype is accurate enough in that the areas of the country with Republicans are fatter, but it's not that accurate for individuals.
By the way, from the county study, "1% increase in county-level support for the Republican candidate corresponds to a 0.02% increase in age-adjusted obesity rates" (adjusted for their covariates). In other words, imagining two counties that are otherwise demographically identical, but one which is magically 100% Republican voting and one that is 100% Democrat voting, the difference would be 100 * 0.02%points, or 2%points. Basically zero.
If I had to guess, I would say New Jersey Republicans are probably the least fat conservatives (despite Chris Christie). Conservatism is associated with rural Christianity, whereas liberalism is associated with urbanism and education. Obesity is also negative correlated with intelligence.
Yes, but apparently these associates are too weak to result in much of a difference. The BMI and IQ link is quite weak to begin with.
In this article, I cite a study from Britain showing a 42% obesity gap between labor and conservatives. What do you make of that?
https://deepleft.substack.com/p/conservatives-are-mentally-ill
You mean this one? https://www.nature.com/articles/s41366-024-01569-5/tables/2 Look at the p values. This study is p-hacked.
I don't have any education in statistics, but the highest p-value is 0.049, which does look suspicious. I need some kind of AI tool that I can feed a study into and it will flag P-values for me.
"controlling for country of origin, age (using a spline), and gender did not change the results."
Country of origin? That's a strange thing to control for. What about race?
It's probably black and Hispanic fat people balancing out the fat white people, who are Republicans. It is sort of like IQ, where racial differences obscures differences among whites, which is what most people are thinking of. There is just no way on earth that there isn't a relationship among whites given what we know about how groups vote.
>Country of origin? That's a strange thing to control for. What about race?
Unfortunately, the ESS does not have a question that identifies people by race. I mistakenly wrote 'country of origin' instead of 'country of residence', which is what the variable was measuring.
The GSS after a certain year started asking people if they were Hispanic; after restricting the sample to only non-Hispanic Whites, the positive correlation still existed (r = .031) but was not still statistically significant (p = .10). Based on priors, the true correlation should be a bit stronger than that, though the observed association does betray how prominent the stereotype is.
If we are going to discount the lumpenproletariat that Dems turn out, should we not do the same for the GOP?
I'd be interested in seeing white obesity rate by ideology controlling for IQ and family size. I doubt liberals maintain an edge after accounting for these things. Though I have no doubt libs win the "skinny fat" category.
It's just very hard for me not to notice the ugliness of liberals versus conservatives amongst similar SES people.
When you do not have the data, do not analyse it. I do believe that fat Republicans and obese Democrats exists, and I do believe they are not the same group, but I cannot prove it: I do not have the data. And you cannot refute this claim without the data.
We have no reliable data on obesete and American political identification. However, we know that American political polls are sampling about 5% of the autoselective part of the population who are willing to answer surveys. Chances are high that body-shamed Republican men hardly get to those samples that do not record their weight, anyways.
The ESS survey is a red herring. It does not record political affiliation for 25 years, the left-right placement in European multi-party democracies is a very bad proxy for anything and useless for American political identification.
Now, it would be very interesting to know more about obese people, their identities, affiliations, because it would also help fighting obesity. However, to make your claims, you would have to overcome the following barriers:
1. You need to prove that obese people vote or identify politically. We do not know. Because obesity is widespread, we can assume that some do.
2. You need to prove that political polls include obese people, and if not, you would have to create a reliable sampling that includes them. It will be very costly, difficult, but worth the effort.
3. You would have to see if those socio-democraphic, psychometric, environmental variables that contribute to obesity are independent of political identities and party affiliation. I dare to say that you will not be able to that, it would contradict a large body of research on both obesity and political behavior.
I think it would be far more fruitful to get data on obese people, and try to find out why some of them affiliate Democrat and Republican. Because Democrats and Republicans are a quiet well separated group demographically, I think that you will find a culturally, demographically different obese Democrat and obese Republican electorate. My hypothesis is that the 'fat Republican' exists and it is a distinct group. Yet studying them would require first a qualitative analysis of how obesee people treat politics and political polling, then a good survey that perhaps even overweights this group, and then doing the analysis.
In the absence of the data, I would not analyse data, because that is only misleading.
You are pedantic.
No, I think that your uninformed data gimmicks cast negative light on survey scientists, political scientists, and public health professionals but arbitrarily mixing up completely unrelated dataset that are incompatible by design. No survey specialist would endorse the way you interpret and mix their data: retrospective survey harmonisation is hard. You seem to be uninformed about the ESS survey and the left placement variable that has a body of literature why it is ill-designed (no place on its scale for the centrist or median voter) and has no connection party affiliation. You disregard the differences in party affiliation in Europe & US, and absolutely have no clue on the non-genetic factors of obesity. What you do is that you invent a catchy heading that fits in well with the prejudices of many people, then mimick data analysis and come to a conclusion that is highly counter-intuitive. Call me pedantric, but at least I know how to work ethically and in a scientific manner with survey data.
It's great we have geniuses such as yourself to teach us.
It's interesting to see the lack of evidence. I think that stereotype comes from age. But analyzing self reported heights and weights strikes me as naïve.
Can you explain why you include the comparison before we adjust for covariates, rather than after? Wouldn't the bivariate logistic regression have a spurious coefficient?
The Brazilian stereotype of Americans is that they're often overweight. Even expats here tend to be noticeably heavier than most locals. Emil recently posted about IQ and body weight, suggesting there's no clear correlation between the two.
It's quite weak relationship, yes. Which is curious because very few people in my smart people network are fat.
https://www.emilkirkegaard.com/p/are-fat-people-dumb
If you control for income it might be a stronger relationship. Poor low intelligence people can dodge obesity by not having enough to spend on extra food
The European stereotype is Americans are fat, regardless of political affiliation