Linux surpassed MacOS in marketshare for the first ever time this month. Let’s go! :)

  • Pigeon@beehaw.org
    link
    fedilink
    arrow-up
    1
    ·
    1 year ago

    Valve isn’t showing us their methodology (sample size, sample selection mechanism, rate at which people decline the surveys, how the hardware information is gathered by steam once the survey is agreed to, etc), so it’s hard to say whether or not a 0.52% increase is accurate, or whether would meet the p-value test of being less than 5% likely to be the result of chance.

    Statistically speaking, sometimes if you flip a coin ten times, it’ll land as heads ten times in a row, even though every flip could go either way. Based on that particular sample of 10 coin flips, it’d look like the chance of a coin landing heads up is 100% - obviously not the case in reality. Another time, you might flip a coin ten times and get ten tails, seemingly the exact opposite and equally dramatically wrong result. Larger sample sizes reduce the risk of such a coincidence occurring across the entire sample, and P-values are helpful for checking the likelihood of whether a fluke such as this may have affected the result of a study.

    Even when studies and surveys do have p < .05 (aka, the likelihood that the result is purely caused by chance is less than 5%, aka statistically significant), there’s still that up-to-5% chance that the result is just random luck affecting the sample. And p-values don’t account for things like sample bias or other methodological errors, like if valve were preferentially serving the survey to more Linux users than Valve users, or if Linux users are just more likely to respond to the steam survey (because they want that Linux number to go up so Linux gets more support) than Windows users, and that kind of thing. Or for p-hacking shenanigans.

    All that said, a small increase makes sense intuitively if it’s including the steam deck, and especially considering the unpopularity of Windows 11, and especially since I’m given to understand the Linux numbers have been slowly increasing for a while? But even so, the point I’m making here is I don’t think we know enough to definitively say that it’s not just luck.