Is Steam Hardware Survey completely fake?

[C1REX]

I always wondered how accurate Steam Hardware stats were.
Some stats are highly questionable, such as the 4080 being more popular than the 4090.

But the most dubious statistic is that Chinese is the most popular language by far. Chinese accounts for 45.9%, compared to English at 26.4%.

Like, really?

https://store.steampowered.com/hwsurvey/Steam-Hardware-Software-Survey-Welcome-to-Steam

[KalGimpa]

Postat inițial de C1REX:

Postat inițial de KalCuey:

at some point you either believe what is published or you don't

I believe that steam is sharing the true numbers they collect. I also believe these numbers can have many very serious errors.

Like Chinese speaking players make 25% of steam players in November. 45% of total steam players in October. All this while making 17% of global population.

Do you believe these numbers?

why wouldn't i?

what theory makes you believe that there is something wrong with them?

could they be off by +/- % points?

maybe

this is not some third party group taking a survey for a company

questioning a fraction of the users and extrapolating from the data

this is a company giving us their information directly

what reason would they have to skew it?

[plat]

Yes, I don't believe the results are tampered with to forward some secret agenda but that's my opinion.

"Fake" maybe isn't the word. "Questionable"? "Representative"?

[C1REX]

Postat inițial de KalCuey:

why wouldn't i?

what theory makes you believe that there is something wrong with them?

could they be off by +/- % points?

maybe

Okay, so how can you explain that players who speak Chinese were 45% of all Steam users in October and 25% in November? Why such a big difference from month to month? China has 17% of the world's population.

[KalGimpa]

Postat inițial de C1REX:

Postat inițial de KalCuey:

why wouldn't i?

what theory makes you believe that there is something wrong with them?

could they be off by +/- % points?

maybe

Okay, so how can you explain that players who speak Chinese were 45% of all Steam users in October and 25% in November? Why such a big difference from month to month? China has 17% of the world's population.

why do you keep bringing up their pop size like it means something?

i have no clue why it fluctuates so much

it could be something as easy to describe as that is just how many logged on and when

you have not given a reason why they would skew/manipulate the data

what purpose would it achieve?

[C1REX]

Postat inițial de KalCuey:

why do you keep bringing up their pop size like it means something?

i have no clue why it fluctuates so much

it could be something as easy to describe as that is just how many logged on and when

you have not given a reason why they would skew/manipulate the data

what purpose would it achieve?

I don't believe they intentionally skew or manipulate data.
I think it's full of errors and unexplainable fluctuations.
We also don't know whether a simple number of logins can change global data to such insane numbers.
I simply think the data is often not very accurate.

[Illusion of Progress]

These unexplainable fluctuations happened (so far as I know) only twice in years and years. Yes, there's something to them, but they can be cast aside.

They're not even that unexplainable either. The data is highly suggestive as to the cause (although maybe not why) if you know how to interpret it.

Postat inițial de C1REX:

China has 17% of the world's population.

This is irrelevant. Steam isn't sampling the entire population; only the gaming population.

[PopinFRESH]

Postat inițial de Illusion of Progress:

These unexplainable fluctuations happened (so far as I know) only twice in years and years. Yes, there's something to them, but they can be cast aside.

They're not even that unexplainable either. The data is highly suggestive as to the cause (although maybe not why) if you know how to interpret it.

Postat inițial de C1REX:

China has 17% of the world's population.

This is irrelevant. Steam isn't sampling the entire population; only the gaming population.

One step further...

"This is irrelevant. Steam isn't sampling the entire population, Steam isn't even sampling the entire gaming population, Steam is sampling only the gaming population of Steam users."

EDIT: which yes, is likely to be highly correlated with "the gaming population". But they don't have 100% marketshare...

[C1REX]

Postat inițial de Illusion of Progress:

This is irrelevant. Steam isn't sampling the entire population; only the gaming population.

Is it irrelevant? No other single country can reach 25% ot total steam users.

Also from what I know 32% of steam users were speaking Chinese in September.
So 32% ->45% ->25%.
(checking historical data is not easy what I also don't like)

How should I read these numbers?
What percentage of steam users use Chinese language?

[cSg|mc-Hotsauce]

Postat inițial de C1REX:

Postat inițial de Illusion of Progress:

This is irrelevant. Steam isn't sampling the entire population; only the gaming population.

Is it irrelevant? No other single country can reach 25% ot total steam users.

Also from what I know 32% of steam users were speaking Chinese in September.
So 32% ->45% ->25%.
(checking historical data is not easy what I also don't like)

How should I read these numbers?
What percentage of steam users use Chinese language?

Just look for yourself...

https://web.archive.org/web/20230000000000*/https://store.steampowered.com/hwsurvey/Steam-Hardware-Software-Survey-Welcome-to-Steam

[C1REX]

Postat inițial de cSg|mc-Hotsauce:

Just look for yourself...

https://web.archive.org/web/20230000000000*/https://store.steampowered.com/hwsurvey/Steam-Hardware-Software-Survey-Welcome-to-Steam

:summercat2023:

Thank you

[Illusion of Progress]

Postat inițial de C1REX:

Postat inițial de Illusion of Progress:

This is irrelevant. Steam isn't sampling the entire population; only the gaming population.

Is it irrelevant? No other single country can reach 25% ot total steam users.

Yes, it is irrelevant. Otherwise, explain how that is relevant when gaming population does not equal population. You premise here is that results must be skewed because they are too far from matching the population of the country, and you seemingly don't like that, yet you're not understanding that's an entirely arbitrary thing.

Countries are land that fall within human made borders. That's it. And Steam does not care about that, nor should it. As long as it hands out surveys evenly amongst the gaming population, then it would be doing it correctly. An expectation that Steam should hand them out evenly per country would, ironically, be an incorrect sampling bias that would skew the results.

If you can provide some reasoning that Steam is consistently doing this for China, maybe there's a discussion to be had about that (the two months in particular where the numbers jumped are acknowledged as being skewed so there's little point beating that dead horse). It's no secret or anything that China has a massive gaming population, or that there's less of a console gaming population there siphoning some of their gamers away from the PC.

Postat inițial de C1REX:

Also from what I know 32% of steam users were speaking Chinese in September.
So 32% ->45% ->25%.
(checking historical data is not easy what I also don't like)

How should I read these numbers?

Again, the two months where the numbers were clearly off has been acknowledged as such.

Therefore, when I said the data is suggesting something, I'm not talking about one of those shifts.

[C1REX]

Postat inițial de Illusion of Progress:

Yes, it is irrelevant. Otherwise, explain how that is relevant when gaming population does not equal population.

I wasn't clear about what I meant. My point isn't that China should represent exactly 17% of the Steam chart, but rather, due to its size, it's likely the most significant influencer on the data. Any errors in data collection in this one country could significantly skew the overall global chart. On the other hand, a country like Sweden is unlikely to ever account for 25% of the chart. In this context, the size of the country does matter.

I've reviewed historical data for China for the last two years. Typically, China accounts for about 25% of the players. However, there are exceptions:

October 2022: 32.17%
March: 51.63%
September: 32.22%
October: 45.93%

While it's feasible for China to make 25% of total players, I have some doubts. I'm certain that errors have impacted the global data at least four times. Three times this year.

[Illusion of Progress]

The size of a country itself doesn't matter because Steam should (key word) be simply randomly offering the survey without bias, but differences in certain factors (like culture) within countries may indeed be relevant. What factors does China have that the Western world doesn't have as much of? A much smaller console population, meaning more of its gaming population will be PC gamers, combined with being more reliant on net cafes. We might not know exactly why certain places in China get sampled more only once in a Blue moon (though it's almost surely net cafe related), but we can still see that is what is happening. Again, patterns of data can suggest (key words, admittedly) a lot.

The two outlier months (I'm not aware of any beyond that, but you're now showing there's two others to a lesser degree I guess) are apparently (key word) erroneous and can be tossed from consideration, if you will.

Those months aside, the results are typically too consistent to suggest they are wrong, and being "correct" here means whatever Valve wants the data to be, and not what any random person wants the data to measure up to being. Even if what Valve wants is a higher tolerance for being a bit inaccurate than you might wan.. That's an important distinction here and I already pointed this out. The main thing the survey is meant to be used for is for game developers to grasp the broad market, not "X hardware spec/part has precisely Y market share", so people wanting to say the data may be wrong because it doesn't measure up for that is more user error than data error.

Ultimately, Valve's silence seems to indicate to me that they are either unaware of the "issue", or don't care and don't consider it an issue as it's not throwing off the data much for its intended purpose in their eyes (or they don't consider it worth addressing even if they do).

[76561199540158642]

less then 2% of vr is normal because

1. it failed

2. bots don't use Vr helms

[Bad 💀 Motha]

Postat inițial de Golden Unicorn:

less then 2% of vr is normal because

1. it failed

2. bots don't use Vr helms

Yet they just developed a VR Headset small enough for Lab Mice to wear and use.
Yes I'm not kidding, it's a real thing.