Can you elaborate on why population size skews working from home distributions? It isn't obvious to me.

It's effectively cherry-picking data.

If you want to encourage something as public policy (like working from home, recycling, public transit, etc), it's a lot easier to deploy a new program at a 'city level' where everyone more or less lives and travels in similar way than than at a US Federal level where you are dealing with over 300 million people who live in all kinds of geography and all kinds of different situations. The US covers everything from highly urban areas to vast, unpopulated deserts and mountains.

The Netherlands is about the size (population wise) of a large city like Shanghai and it's fairly consistent in geography and urbanization throughout the entire country. That's just much easier situation to influence than a vast country like the US. It's a much simpler problem.

It would be more fair to compare the US with all of Europe, including Spain, Greece, Croatia, etc. What percent of people are working from home in Croatia? And how does that compare to a Netherlands-sized cherry picked area in the US, like the New York City metro area? See how that's an unfair comparison?

According to the article, the US vs. the entire EU rates are not too far off.

> It would be more fair to compare the US with all of Europe, including Spain, Greece, Croatia, etc. What percent of people are working from home in Croatia?

These are all shown on the map. Only a handful small Eastern European countries (plus Greece) have lower rates than the US. Spain has 4.3% people working from home. Countries both with lower and higher population density than the US have higher numbers.

> According to the article, the US vs. the entire EU rates are not too far off.

EU is 5.2% and the US 2.8%, that's about half of the EU.

> it's a lot easier to deploy a new program at a 'city level' where everyone more or less lives and travels in similar way than than at a US Federal level where you are dealing with over 300 million people who live in all kinds of geography and all kinds of different situations.

Why ? You can just treat the US a a collection of smaller communities and implement it independently in each of those. All you need to do at the federal level is to instruct the local level to tackle the issue.

Small sample sizes are much more likely to produce extreme results, because it's easier to get an extreme result by chance. In the US, the highest rates of cancer occur in the most rural counties. So do the lowest rates of cancer.

Think about it like grabbing from a giant bag of M&M's. The fewer you select, the higher the chances your results will be skewed towards one color.

But we are not sub-sampling here. We are comparing the whole of the Netherlands with the whole of the US right?

We might take a big bag of M&M's and a smaller bag of M&M's, and count both of their contents, and find that the small bag has proportionally significantly more blue M&M's than the bigger bag.

1. it’s good to flag clearly false statements 2. it’s not a “fact it’s easy to have good ratios when your country is tiny”?

Jelly1's comment adds even less. If your observation about correlations is true, can we assume that Belgium and Albania have even higher ratios of employees that work from home?

Belgium = 6.6%. Given that Belgium and the Netherlands are neighbors, with a majority Flemish population, this means simple size isn't the primary factor. (Albania isn't even listed.)

Switzerland is only slightly larger than the European part of the Netherlands, and yet it has a rate of 4.1%.

Population density might be a factor, but simple density isn't enough, as Finland's rate is 13.3% while the US has a density about twice that of Finland.

Since size isn't a clear factor, and density isn't a clear factor, I could not makes sense of Jelly1's comment. It did not come across as an informed jab but a factually incorrect statement, which I tried to highlight, without a verbose analysis like this one.