I do wonder if there’s any selection bias in the hardware survey steam does. I’m sure they sample randomly, but I think a user on Linux might be much more eager to participate in the survey than a Windows user, simply because Linux users tend to have a desire to be more vocal about their OS use than Windows users.
Linux users are also more likely to be private people and not willing to share info about their system. Both of those two effects are probably pretty small and cancel eachother for this survey.
It’s not a manually filled out survey. It’s just a box that pops up on Steam, you click OK to share info, and that’s it. I think there’s very little bias involved in it.
There’s the bias between users willing to share that data and those not willing. All sorts of correlations could be hiding there.
I can’t find out what the rate of refusals are, but I’d guess them to be rather low, maybe 10% at most. The survey is incredibly easy to perform and very transparent about colleted data.
The important thing to realise is that the survey is very consistent, so while the baseline may have some bias, the trends are very representative.
I do wonder if there’s any selection bias in the hardware survey steam does. I’m sure they sample randomly, but I think a user on Linux might be much more eager to participate in the survey than a Windows user, simply because Linux users tend to have a desire to be more vocal about their OS use than Windows users.
Linux users are also more likely to be private people and not willing to share info about their system. Both of those two effects are probably pretty small and cancel eachother for this survey.
It’s not a manually filled out survey. It’s just a box that pops up on Steam, you click OK to share info, and that’s it. I think there’s very little bias involved in it.
There’s the bias between users willing to share that data and those not willing. All sorts of correlations could be hiding there.
I can’t find out what the rate of refusals are, but I’d guess them to be rather low, maybe 10% at most. The survey is incredibly easy to perform and very transparent about colleted data.
The important thing to realise is that the survey is very consistent, so while the baseline may have some bias, the trends are very representative.