LGBTQ+ population density data by city?
May 27, 2022 10:13 AM Subscribe
Where can I find a dataset containing the population or pop density of LGBTQ+ individuals *by US city and/or metropolitan area* (not just counties or states)? I can only find data on the largest 50 cities and I need coverage of as many cities as possible.
Gallup (via the Williams Institute) report data on the 50 largest cities only.
The Williams Institute also has county-level data.
I believe the Census collects this info via the American Community Survey and the Household Pulse Survey but I can't find a full tabulation of them.
It would be extremely wonderful if these data were available in a spreadsheet rather than an interactive map.
Gallup (via the Williams Institute) report data on the 50 largest cities only.
The Williams Institute also has county-level data.
I believe the Census collects this info via the American Community Survey and the Household Pulse Survey but I can't find a full tabulation of them.
It would be extremely wonderful if these data were available in a spreadsheet rather than an interactive map.
Best answer: FWIW that county-level data map is only measuring same-sex cohabitating couples, which I think might be the only data you can get from the American Community Survey (I am not a census expert, I could be totally wrong!). On Census.gov there are some prepopulated tables with info about same-sex couples, or you can dive into the "microdata" tables and see all the available fields from the ACS and find data to do your own analysis.
posted by mskyle at 11:39 AM on May 27, 2022 [5 favorites]
posted by mskyle at 11:39 AM on May 27, 2022 [5 favorites]
Best answer: Yes, seconding mskyle, the Census only collects info about same-sex couples who live together. There are no nationally representative sexual orientation/gender identity data at a sub-state, so far. The Census Bureau planned to add one for the 2020 Census, but you can blame the Tr*mp admin for the fact that it didn't happen.
posted by acridrabbit at 1:03 PM on May 27, 2022 [2 favorites]
posted by acridrabbit at 1:03 PM on May 27, 2022 [2 favorites]
Response by poster: I totally said metropolitan area when I meant municipal area. Sheesh...it's been a long week.
Anything that even roughly quantifies population prevalence of LGBTQ+ individuals is workable even if extremely imperfect, as in the case of same-sex couples.
It sounds like there's nothing from the Census. Is there anything from Gallup or other non-gov orgs? It seems like Gallup might have the data for more than those 50 cities based on their methodology- any ideas of how to access it?
posted by quiet coyote at 1:56 PM on May 27, 2022
Anything that even roughly quantifies population prevalence of LGBTQ+ individuals is workable even if extremely imperfect, as in the case of same-sex couples.
It sounds like there's nothing from the Census. Is there anything from Gallup or other non-gov orgs? It seems like Gallup might have the data for more than those 50 cities based on their methodology- any ideas of how to access it?
posted by quiet coyote at 1:56 PM on May 27, 2022
There is data on same-sex partners available from the Census down to the Census tract level (average of around 4k people). Good luck!
posted by acridrabbit at 5:03 PM on May 27, 2022 [1 favorite]
posted by acridrabbit at 5:03 PM on May 27, 2022 [1 favorite]
Best answer: Hi. I'm fairly qualified to answer this question. It’s going to take multiple comments.
Here are some things to think about, though by no means all the things. A more extensive explanation is literally a couple hundred pages long.
1) Note that cities, metropolitan statistical areas, and municipalities capture different geographic areas. The city boundaries I live in, the water district, and the MSA are all different places, for example. You're going to need to pinpoint exactly what geographic area you're after.
1a) Please note that if you decide to do what acridrabbit suggests and piece together places from Census tracts, that you're in for a metric fuckton of work (it's a good idea! Just a lot of work. Way more work than you may realize if you’re not used to wrangling this data). There are 84,414 Census tracts in the United States. My dissertation chair wanted a national analysis with Census tracts, and I politely told her to go to hell (or rather, ArcMap did, as it was already choking on same with the 3,143 counties or county-level equivalents in the U.S.)
2) If you do decide that you want to use Census data (with the issues already discussed and that I will discuss), you needed to decide if you want to use Decennial Census data or the American Community Survey data. The advantage of the Decennial (the every 10 years Census) data is that it is much closer to an actual count. The disadvantage is that we do it every 10 years, and by the time that the poor overworked sods at the Census Bureau are able to get the data cleaned and posted for human consumption, it’s a couple years old.
(Even with ACS data, there’s a delay; the 2016-2020 rolling file just dropped a couple months ago, though that’s more of a Rona thing than a data takes awhile to clean and assemble thing).
ACS data, on the other hand, is available for more frequent time periods; there’s 1 year data files and 5 year rolling data files. However, the ACS is a *survey*, which means that it is estimated data. With small populations (and us queer folks are a fairly small population) error in estimates is going to be greater than in bigger populations.
Either way, we know there’s undercounting in the data because not everyone is out, even now in 2022.
Additionally, as already mentioned, Census data only captures same-sex households that live together. Households are visible, individuals are not, and the Census doesn’t ask about sexuality. Additionally, the Census only started coding and reporting data about same sex households in 1990. Before that, no matter how you identified yourself, you were coded as roommates. In addition, if you use Census data prior to 2013, anyone who indicated that they were married was recoded as an unmarried couple, because the Census’ lawyers were told that coding married couples in the dataset would constitute federal recognition of Same-sex marriage. (How the VA, which recognized SSM earlier than the federal government, got away with that, I don’t know, but I am forever grateful as my wife is a vet).
3) You say LGBTQ+ individuals… so now you’re putting sexuality and gender in the same bucket. How do you plan on counting someone who is, say, gay AND trans? Do they double count? Even if you had the data to do that (and the Census doesn’t collect data on trans folks, the Census just asks for sex and you pick what you want to put down). You’d likely have to hand clean that data too, if you had data that let you have access to anonymized individual cases (and the General Social Survey does do that), so you could figure out how was trans and gay or lesbian or whatever and only count them once). This is why we generally run gender and sexuality separately, UNLESS you’re wanting to compare say SS male households and SS female households, which you can do with Census and ACS data. Gallup is doing that making sure they don’t double count people behind the scenes (“You can select as many as apply.”)
4) The data you want just doesn’t exist. No one has counted us as individuals yet for national level data. That said, it looks like you might be able to access Gallup Data (https://www.gallup.com/analytics/213617/gallup-analytics.aspx). The Pride Study is working on it (https://pridestudy.org/research).
posted by joycehealy at 8:32 AM on May 28, 2022 [4 favorites]
Here are some things to think about, though by no means all the things. A more extensive explanation is literally a couple hundred pages long.
1) Note that cities, metropolitan statistical areas, and municipalities capture different geographic areas. The city boundaries I live in, the water district, and the MSA are all different places, for example. You're going to need to pinpoint exactly what geographic area you're after.
1a) Please note that if you decide to do what acridrabbit suggests and piece together places from Census tracts, that you're in for a metric fuckton of work (it's a good idea! Just a lot of work. Way more work than you may realize if you’re not used to wrangling this data). There are 84,414 Census tracts in the United States. My dissertation chair wanted a national analysis with Census tracts, and I politely told her to go to hell (or rather, ArcMap did, as it was already choking on same with the 3,143 counties or county-level equivalents in the U.S.)
2) If you do decide that you want to use Census data (with the issues already discussed and that I will discuss), you needed to decide if you want to use Decennial Census data or the American Community Survey data. The advantage of the Decennial (the every 10 years Census) data is that it is much closer to an actual count. The disadvantage is that we do it every 10 years, and by the time that the poor overworked sods at the Census Bureau are able to get the data cleaned and posted for human consumption, it’s a couple years old.
(Even with ACS data, there’s a delay; the 2016-2020 rolling file just dropped a couple months ago, though that’s more of a Rona thing than a data takes awhile to clean and assemble thing).
ACS data, on the other hand, is available for more frequent time periods; there’s 1 year data files and 5 year rolling data files. However, the ACS is a *survey*, which means that it is estimated data. With small populations (and us queer folks are a fairly small population) error in estimates is going to be greater than in bigger populations.
Either way, we know there’s undercounting in the data because not everyone is out, even now in 2022.
Additionally, as already mentioned, Census data only captures same-sex households that live together. Households are visible, individuals are not, and the Census doesn’t ask about sexuality. Additionally, the Census only started coding and reporting data about same sex households in 1990. Before that, no matter how you identified yourself, you were coded as roommates. In addition, if you use Census data prior to 2013, anyone who indicated that they were married was recoded as an unmarried couple, because the Census’ lawyers were told that coding married couples in the dataset would constitute federal recognition of Same-sex marriage. (How the VA, which recognized SSM earlier than the federal government, got away with that, I don’t know, but I am forever grateful as my wife is a vet).
3) You say LGBTQ+ individuals… so now you’re putting sexuality and gender in the same bucket. How do you plan on counting someone who is, say, gay AND trans? Do they double count? Even if you had the data to do that (and the Census doesn’t collect data on trans folks, the Census just asks for sex and you pick what you want to put down). You’d likely have to hand clean that data too, if you had data that let you have access to anonymized individual cases (and the General Social Survey does do that), so you could figure out how was trans and gay or lesbian or whatever and only count them once). This is why we generally run gender and sexuality separately, UNLESS you’re wanting to compare say SS male households and SS female households, which you can do with Census and ACS data. Gallup is doing that making sure they don’t double count people behind the scenes (“You can select as many as apply.”)
4) The data you want just doesn’t exist. No one has counted us as individuals yet for national level data. That said, it looks like you might be able to access Gallup Data (https://www.gallup.com/analytics/213617/gallup-analytics.aspx). The Pride Study is working on it (https://pridestudy.org/research).
posted by joycehealy at 8:32 AM on May 28, 2022 [4 favorites]
Best answer: (The Williams Institute also has county-level data. Hover over that map, it's couples, same data the rest of us are using and crying into our coffee because queer demography is a bitch because of data availability.)
5) So you decide to wade into the wide world of crunching this data on your own, which is great! But if you haven't done this kind of thing before, it's a lot, just a heads up. Mskyle linked to the Census microdata tables, which are great. I prefer the National Historical GIS interface, however: https://www.nhgis.org It's just about what you're more familiar with.
Don't be put off by the fact that it says GIS; you still download spreadsheets, not shapefiles, the spreadsheets just include geographic identifiers (and in fact, you have to, so you can clean and join your data so you can make maps if you were going to make maps, which you said is not what you want, just explaining why there are spreadsheets and shape files, both).
Once you make an account, you have to choose a geographic level and choose a year. Let's say you pick county and the 2015-2019 data (it looks like 2016-2020 doesn't have gender and household status in it yet, they're still adding data). You're going to need then go digging a bit. Click on topics and pick households, relationship to householder, and household and family size, because I can't remember which topic it's under. Cuss and change to "or" in the dropdown next to topics. In the next table, you want Household Type (Including Living Alone) by Relationship, which will give you spouses and unmarried partners.
So then you download the data and look at the codebook. ALU6E011 is households with same-sex spouses, ALU6E013 is households is same-sex unmarried partners, and then you cuss because this group of data includes the total population and the population living in households but not the number of households. When you back and out and get it, it's variable AMQRE001.
(ALU6E011 + ALU6E013) / AMQRE001
So that's how you get a percentage of same-sex households of all households by county (I haven't even gotten to, when you say population density, do you mean by, everyone in the population? Just adults? Just households? Something else? What do you want of your denominator), which isn't exactly what you were looking for, but an example of the kinds of gymnastics you're looking at to get to I believe the Census collects this info via the American Community Survey and the Household Pulse Survey but I can't find a full tabulation of them.
This, friends, is why my next couple of projects will be qualitative (which has its own headaches!)
So play with the data and have fun, but this is all just a heads up that it's not as simple as, I'm going to be able to download some spreadsheets and whiz up some numbers.
posted by joycehealy at 8:59 AM on May 28, 2022 [3 favorites]
5) So you decide to wade into the wide world of crunching this data on your own, which is great! But if you haven't done this kind of thing before, it's a lot, just a heads up. Mskyle linked to the Census microdata tables, which are great. I prefer the National Historical GIS interface, however: https://www.nhgis.org It's just about what you're more familiar with.
Don't be put off by the fact that it says GIS; you still download spreadsheets, not shapefiles, the spreadsheets just include geographic identifiers (and in fact, you have to, so you can clean and join your data so you can make maps if you were going to make maps, which you said is not what you want, just explaining why there are spreadsheets and shape files, both).
Once you make an account, you have to choose a geographic level and choose a year. Let's say you pick county and the 2015-2019 data (it looks like 2016-2020 doesn't have gender and household status in it yet, they're still adding data). You're going to need then go digging a bit. Click on topics and pick households, relationship to householder, and household and family size, because I can't remember which topic it's under. Cuss and change to "or" in the dropdown next to topics. In the next table, you want Household Type (Including Living Alone) by Relationship, which will give you spouses and unmarried partners.
So then you download the data and look at the codebook. ALU6E011 is households with same-sex spouses, ALU6E013 is households is same-sex unmarried partners, and then you cuss because this group of data includes the total population and the population living in households but not the number of households. When you back and out and get it, it's variable AMQRE001.
(ALU6E011 + ALU6E013) / AMQRE001
So that's how you get a percentage of same-sex households of all households by county (I haven't even gotten to, when you say population density, do you mean by, everyone in the population? Just adults? Just households? Something else? What do you want of your denominator), which isn't exactly what you were looking for, but an example of the kinds of gymnastics you're looking at to get to I believe the Census collects this info via the American Community Survey and the Household Pulse Survey but I can't find a full tabulation of them.
This, friends, is why my next couple of projects will be qualitative (which has its own headaches!)
So play with the data and have fun, but this is all just a heads up that it's not as simple as, I'm going to be able to download some spreadsheets and whiz up some numbers.
posted by joycehealy at 8:59 AM on May 28, 2022 [3 favorites]
One last thing: ACS does also count male male and female female households and then you're looking at (MM married + MM unmarried + FF married + FF unmarried) / total households, I just can't remember which variables it is off the top of my head, and oh, heads up, the variable identifies change between data sets. But the data is there if you go digging around.
posted by joycehealy at 9:01 AM on May 28, 2022
posted by joycehealy at 9:01 AM on May 28, 2022
This thread is closed to new comments.
posted by madcaptenor at 11:10 AM on May 27, 2022