PUMA stands for Public Use Microdata Area. PUMAs are areas designated by the Census Bureau for statistical reporting. They each have between 100k and 200k residents, don’t cross state lines, and follow county and city boundaries when possible. Their big advantage over county maps is letting us see much more detail in urban areas, but it does come at the cost of being able to present all the information in one image. Additionally, many counties often have very high margins of error for survey data like this. The relatively consistent size of PUMAs ensures reasonably low margins of error across the whole map.
Is this the percentage of people that *live* in these places, or is it that the job is *based* in the location?
Paying SF rents but not willing to go to the SF office is peak SF.
The middle of the country works remotely at some of the lowest rates in America because its economy is built on hands-on jobs rather than laptop-based work.
Obviously see spikes in cities which makes sense.
The one area in the second chart that surprises me is Cincinnati being so low. I would expect a lot of financial jobs to have WFH opportunities.
“Working from home” means 100% working from home? Hybrid is not counted?
Source is from *2023*. These numbers are likely greatly inflated compared to the current year.
Aren’t all farmers working from home?
This is hilarious if you look closely at Oregon there’s a light purple blurb in the middle of the state with a dark purple small dot. I live there with close to 50% us working from home. I forget how rare we are.
Love to see a table by cities ranked for future job search
I RTO and sit on Teams/zoom with others in my office, even the woman in the next cube. We never use the conference rooms. Work has changed but management can’t accept it.
Could easily be 50% but we would rather live in hell than gives workers even a glimmer of hope in a dark and bleak world
Working from home for how long? As in you never go to the office or are you there 1-5 days a week?
wow, it kinda seems like you’re letting the data make you work instead of making the data work. What’s with the variable number of decimals? Why the arbitrary binning (where your bins stop just above the midpoint of the highest value in your dataset)? Let your color scale do the work for you — smooth it into a continuous color scale into of bins, or at least break it into twice as many bins. If you just can’t help yourself from labeling all the states individually (personally I think if you need to label every data point you should just make a table), round to the nearest integer (or technically to the nearest hundredth, as these are percentages).
What even are those boundaries in Slide 3? What exactly do Wellesley, Natick, Sherborn, Holliston, and Hopkinton have in common that Framingham doesn’t with any of them, but instead shares with Marlboro? Also, why are some towns grouped together like Belmont, Arlington, and Watertown, but some are just on their own like Milton, Quincy, Somerville, Cambridge, and Lowell?
Denver’s interesting because the Front Range is pretty much entirely blue. All those little mountain towns are chock full of WFH software engineers from the coasts who moved to ski / MTB / climb semi-fulltime around a handful of work meetings.
They couldn’t afford to live there if they worked a Denver job, so it has to be remote.
Source: A former Denverite who got laid off from a tech job and absolutely could not find anything in Denver remotely close to what I used to get paid.
I’m actually surprised by how similar the state by state numbers are, particularly if you look at the number of people *not* working from home.
84% of people in California don’t work from home; 90% of people in Indiana don’t work from home.
All the cheap, stormy, remote places have very few remote workers. Looks like at least for now, you have to actually be in LA or New York or whatever, get an office job, then talk your boss into letting you work from home some days a week
“Work from home” = do 25% of your job while you play on your phone and get high
Interesting that it turned out to just be another population density map
Well, this explains why everyone has been saying “RTO is over, it’s like before COVID now” and I think “what the hell are they talking about? I work from home and so do half the people I know, the downtown office buildings are still all like 40% empty!”
I live in one of the dark blue counties.
In the SF Bay Area the map is basically no one wants to cross the bridges, and in San Diego, North County just doesn’t want to take the 5 or 15.
I wonder how much of these are ultimately just inverse maps of traffic patterns?
![[OC] Percent of Workers Working From Home in the US](https://www.europesays.com/wp-content/uploads/2025/12/otmbbcb08g4g1-1080x1024.jpg)
22 comments
Source: IPUMS USA Census Microdata – 2023 1-year data
Citation: IPUMS USA, University of Minnesota, [www.ipums.org](http://www.ipums.org)
Tools: R (packages: dplyr, ggplot2, sf, usmap, tools, ggfx, grid, scales, cowplot, showtext, sysfonts)
You may be wondering what a PUMA is…
PUMA stands for Public Use Microdata Area. PUMAs are areas designated by the Census Bureau for statistical reporting. They each have between 100k and 200k residents, don’t cross state lines, and follow county and city boundaries when possible. Their big advantage over county maps is letting us see much more detail in urban areas, but it does come at the cost of being able to present all the information in one image. Additionally, many counties often have very high margins of error for survey data like this. The relatively consistent size of PUMAs ensures reasonably low margins of error across the whole map.
Is this the percentage of people that *live* in these places, or is it that the job is *based* in the location?
Paying SF rents but not willing to go to the SF office is peak SF.
The middle of the country works remotely at some of the lowest rates in America because its economy is built on hands-on jobs rather than laptop-based work.
Obviously see spikes in cities which makes sense.
The one area in the second chart that surprises me is Cincinnati being so low. I would expect a lot of financial jobs to have WFH opportunities.
“Working from home” means 100% working from home? Hybrid is not counted?
Source is from *2023*. These numbers are likely greatly inflated compared to the current year.
Aren’t all farmers working from home?
This is hilarious if you look closely at Oregon there’s a light purple blurb in the middle of the state with a dark purple small dot. I live there with close to 50% us working from home. I forget how rare we are.
Love to see a table by cities ranked for future job search
I RTO and sit on Teams/zoom with others in my office, even the woman in the next cube. We never use the conference rooms. Work has changed but management can’t accept it.
Could easily be 50% but we would rather live in hell than gives workers even a glimmer of hope in a dark and bleak world
Working from home for how long? As in you never go to the office or are you there 1-5 days a week?
wow, it kinda seems like you’re letting the data make you work instead of making the data work. What’s with the variable number of decimals? Why the arbitrary binning (where your bins stop just above the midpoint of the highest value in your dataset)? Let your color scale do the work for you — smooth it into a continuous color scale into of bins, or at least break it into twice as many bins. If you just can’t help yourself from labeling all the states individually (personally I think if you need to label every data point you should just make a table), round to the nearest integer (or technically to the nearest hundredth, as these are percentages).
What even are those boundaries in Slide 3? What exactly do Wellesley, Natick, Sherborn, Holliston, and Hopkinton have in common that Framingham doesn’t with any of them, but instead shares with Marlboro? Also, why are some towns grouped together like Belmont, Arlington, and Watertown, but some are just on their own like Milton, Quincy, Somerville, Cambridge, and Lowell?
Denver’s interesting because the Front Range is pretty much entirely blue. All those little mountain towns are chock full of WFH software engineers from the coasts who moved to ski / MTB / climb semi-fulltime around a handful of work meetings.
They couldn’t afford to live there if they worked a Denver job, so it has to be remote.
Source: A former Denverite who got laid off from a tech job and absolutely could not find anything in Denver remotely close to what I used to get paid.
I’m actually surprised by how similar the state by state numbers are, particularly if you look at the number of people *not* working from home.
84% of people in California don’t work from home; 90% of people in Indiana don’t work from home.
All the cheap, stormy, remote places have very few remote workers. Looks like at least for now, you have to actually be in LA or New York or whatever, get an office job, then talk your boss into letting you work from home some days a week
Few [Kansas](https://www.realtor.com/realestateandhomes-detail/306-W-5th-NA_Longton_KS_67352_M93334-11298) residents are able to get those jobs in the first place
“Work from home” = do 25% of your job while you play on your phone and get high
Interesting that it turned out to just be another population density map
Well, this explains why everyone has been saying “RTO is over, it’s like before COVID now” and I think “what the hell are they talking about? I work from home and so do half the people I know, the downtown office buildings are still all like 40% empty!”
I live in one of the dark blue counties.
In the SF Bay Area the map is basically no one wants to cross the bridges, and in San Diego, North County just doesn’t want to take the 5 or 15.
I wonder how much of these are ultimately just inverse maps of traffic patterns?
Comments are closed.