Also keep in mind the limitations and biases of the CanIUse stats themselves - last I checked, their data was from StatCounter, and so based on people who visit

Also keep in mind the limitations and biases of the CanIUse stats themselves - last I checked, their data was from StatCounter, and so based on people who visit websites that use the StatCounter service. The country-wise sample sizes [1] make the geographic biases clear - China is pretty underrepresented for example, presumably because Chinese websites don't use StatCounter as much. There may also be other kinds of biases in the data that are not so easily visible. This is not to say that the stats should be disregarded; just that like any other data source, it should be treated as one view of (some of) the data, with its own shortcomings. [1] https://gs.statcounter.com/faq#sample-size https://gs.statcounter.com/faq#sample-size