Not sure steam has ever made the claim that the steam survey is random nor have they ever publicly disclosed the algorithm used to select participants.
Selecting participants in a survey is just one part of what makes a survey accurate, the how, when, and why are all equally important. Particularly when you are talking about polling virtual participants, where you can not exactly validate that each survey response is from a system that hasn't been polled before. In addition steam has a very strict data collection policy, which surely makes fingerprinting unique systems even harder.
Add to that, we know for a fact that steam has had ongoing issues and has revised multiple past surveys due to inaccuracies. At least the ones they have caught as far as they know. The steam survey is only really useful as puff pieces for tech and gaming news websites, anyone using it otherwise is doing so because they are either ignorant or it validates their pre-concieved notions.
There are three problems with this conclusion
1) Blizzard's products were disabled in China on January 23rd. Why would we see a jump now and not in the two months in between?
2) You assume that every person switching from Activision / Blizzard's platform are being polled. If the survey is random, it would not poll all the new Chinese users at once and not within such a specific timeframe. If it is then it's not remotely random.
3) You assume that said users didn't already have a steam account. The overlap between users who have a Activision / Blizzard account who also have a steam account is likely extremely high.
Thank you for your pointless rant.
Firstly, Valve never reversed any survey due to being inaccurate, it was always accurate within the tracked population, the issue was that sometimes, the tracked population included also additional hw which was not supposed to be there, for example, internet cafes or different regions.
No one is whining either, that among gaming machines, there are also included work, and school laptops/pcs which are also inflating and potentially skewing the data.
It doesn't matter that some machines are counted more than once, the entry doesn't have to be unique. You do nothing but create pointless conspiracy theories because your favourite hw manufacturer's market share decreased. Deal with it, no one cares.
If you have a large enough sample, and do the process randomly or you use the same methodology which is semi-random, you get the correct result closely minoring reality within the confidence interval and the fact, they got it SOMETIMES "wrong" despite doing the survey for years, it literally proves they do IT RIGHT, because that's how statistics and probability works and you are only publicly exposing your serious lack of knowledge. Now go complain that politics pools cannot be trusted, because they are sometimes wrong despite THAT BEING THE FEATURE AND ISSUE DUE TO METHODOLOGY. If you had any clue what you are talking about, you would UNDERSTAND, that the data presented are not hard facts and there is a chance they are completely wrong and that every time we do such surveys we do talk about probabilities being within the range of the population we try to analyze and survey.
OH NO, they were "wrong" in the past. THAT'S THE FEATURE and limitation of statistics. If you wanted to be accurate 99,99999% of the time, Valve would have to provide you with such ridiculous % intervals, which would tell you literally nothing. In your case, if I were to estimate your height and the height of the people here, in order to have extremely high confidence in the result, I would have to say that you height is somewhere between 10cm and 250cm.
By all means, if you are such an expert, provide us with a valid methodology which is right all the time when you are surveying only a very small sample. No, there isn't such a thing. Bye. Any discussion about the matter with you is entirely pointless because you do not know the fundamentals.
