As always, such statistics should be treated with caution.

What methodology is used to calculate Statcounter Global Stats?

Statcounter is a web analytics service. Our tracking code is installed on more than 1.5 million sites globally.

Source: https://gs.statcounter.com/faq#methodology

It is assumed that there are more than one billion websites worldwide. It is therefore not exactly unlikely that a Linux user will not access any of these 1.5 million websites.

Furthermore, it is quite common for Linux users to use tools such as Pi-Hole that simply block such statistics scripts. This means that these users would not be counted even if they accessed one of these 1.5 million websites. For my part, I also use computers with Linux that I don't use to access websites. Some of these computers don't even have access to the Internet. They are therefore not counted either.

Finally, let's come to the most important point. Percentage values say not much if you don't know the actual number of users behind them. Let's assume, for example, that 3.5 per cent Linux users were detected in December and only 3 per cent in January. However, if the total number of users was higher in January, it is therefore possible that more users were detected in January.

Source