Steam Hardware Survey updated with GPUs

Algrim said:
That was a classic mistake of polling method. The poll was conducted by telephone but not everyone who was able to vote had a telephone.

EDIT: Polls are not perfect; the way questions are asked can bias a poll; the questions asked can bias a poll. Polls with large sample selections can be pretty damn accurate. Not polling you and your friends are not valid reasons to discount polling, though.
Click to expand...

I argue that having the tools to ask more of the populace for information, and then flat out refusing to, is wrong as it has painted the wrong picture in the past. Am I saying that the Steam numbers are completely fabricated and wrong? No. I'm saying that they should take the time and ask their entire user base that opts in at least once a year. If they did that and made a baseline, I would be more inclined to pay attention to it. Right now I see it as marketing 101 to keep people interested in Steam.
 
I probably am missing something, but I thought the survey was all hardware that is on Steam? I mean I used to check that I OK'd the collection of hardware info every year, but isn't it everyone these day?
 
I've clicked yes to the Steam hardware survey at least 3-4 times in my lifetime.

Anyway, I still remember from my operations management class that in statistical analysis, sampling is used in process control and quality control because it would be unfeasible to test every single product that comes out of the production line. Similarly with surveys, it would be so costly and time consuming to survey an entire population hence the determining a sample that's representative of a population with a margin of error. There are statistical methods/models in determining sample sizes and sample sources. It is done because it is practical. One simply cannot dismiss statistical data just because the entire population was not sampled or does not conform with one's own anecdotal experiences.
 
jologskyblues said:
I've clicked yes to the Steam hardware survey at least 3-4 times in my lifetime.

Anyway, I still remember from my operations management class that in statistical analysis, sampling is used in process control and quality control because it would be unfeasible to test every single product that comes out of the production line. Similarly with surveys, it would be so costly and time consuming to survey an entire population hence the determining a sample that's representative of a population with a margin of error. There are statistical methods/models in determining sample sizes and sample sources. It is done because it is practical. One simply cannot dismiss statistical data just because the entire population was not sampled or does not conform with one's own anecdotal experiences.
Click to expand...
difference here and in most polls is that it has an easy medium by which to attain said information. Steam could add a "send hardware info" selection with a 28-30 day cool down to ensure single occurrences of hardware ownership. Will not likely have 100% participation but higher accuracy than what they have now. Steams current hardware survey is good for GENERAL ownership %s but not accurate enough to say with certainty sales figures.
 
JustReason said:
difference here and in most polls is that it has an easy medium by which to attain said information. Steam could add a "send hardware info" selection with a 28-30 day cool down to ensure single occurrences of hardware ownership. Will not likely have 100% participation but higher accuracy than what they have now. Steams current hardware survey is good for GENERAL ownership %s but not accurate enough to say with certainty sales figures.
Click to expand...

Still doesn't mean you should dismiss a statistical analysis based on a sample representative of a population as BS since it is unfavorable to a certain company you like.
 
jologskyblues said:
Still doesn't mean you should dismiss a statistical analysis based on a sample representative of a population as BS since it is unfavorable to a certain company you like.
Click to expand...

No one has mentioned shit about being "unfavorable".

Every night I open steam, every night I get a pop-up for the days/weekends sales.

For the first week of the month put a slide in the pop-up asking if you want to take the survey. Increases reach, accuracy, etc. and I don't see how this would be "costly or time consuming"
 
You only have to look on the month after month statistics to know its accurate.
 
This thread is a sad example of fallicies and ignorance based "argumentation".

Even if Steam's hardware survey NEVER polls you, your famile, your friends and EVERY single person you have EVER encountered...it would not alter the validity or accurracy of the poll.

Simple as that.
If you do not understand this simple fact:
Please shut the fuck up and...seriously...do some fucking reading up!

Enough with the stupidity and ignorance.
If you are to thick to understand SIMPLE math...then be quiet when the adult talk...ffs the amount of stupid in this thread..,"I have never been polled..,"...no wonder 98% of the world are unable to join Mensa...ffs!!!
 
Isn't it hard to say anything about the quality of their survey without knowing the details?
How many people do they poll? Are they chosen at random? How does it choose people and how often? So on and so forth.
 
Can't we just compare the hardware survey to other data on gpu sales and shipments and see if they correlate?
 
You'd have to figure out how to remove the non-gaming (rendering farms, mining, etc) uses...
 
DWolvin said:
You'd have to figure out how to remove the non-gaming (rendering farms, mining, etc) uses...
Click to expand...

or in that case, add the non-gaming cards.. even if they are not gaming they are part of the total sales of a product.
 
True, I guess defining the desired answer you want is step one...





42
 
normal 1q woes, seasonal drop. q1, q2 are always weak, q3 and q4, for back to school and holidays are strong.
 
razor1 said:
normal 1q woes, seasonal drop. q1, q2 are always weak, q3 and q4, for back to school and holidays are strong.
Click to expand...

No, the numbers decreased almost 20% YEAR ON YEAR!

Quarter-to-quarter graphics board shipments decreased 29.8%, and decreased 19.2% year-to-year.
Click to expand...

The numbers we're seeing now are usually the ones we seen in Q2! That's a huge drop from last year, 3 months early!

aib-pr3.PNG


Notice the seasonal dip at Q2 15, Q2 16, but it's moved forward to Q1 17.

I suppose it's okay, since that's the highest Q4 we've seen in years. But man the graphics industry must be brutal this year.
 
well we have to see, its a bigger drop than usual, but Q1 last year didn't drop which is unusual.
 
That is a pretty large drop, will be interesting to see next quarters results.
 
Seems like release dates needs to be factored in too. Eg with Maxwell high end came out in Q3/Q4-ish but the 960 dropped in Q1... less of a peak and less of a drop off. With Pascal it was all compressed, bigger peak but also bigger drop off.
 
Lots of people bought 10XXs and 4XXs (or used 980s and Furys) late last year, so not as many looking to buy this early in the year, would be my take.
 
Steam May

DX12 systems. This requires Windows 10 with a DX12 GPU.
GTX 1080TI unlisted or below threshold.
GTX 1080 1.19% (+0.02%)
GTX 1070 2.30% (-0.01%)
GTX 1060 2.84% (+0.11%)
GTX 1050TI 0.91% (+0.13%)
GTX 1050 0.57%(+0.12%)
GTX 1030 unlisted or below threshold.
RX 580 unlisted or below threshold.
RX 570 unlisted or below threshold.
RX 560 unlisted or below threshold.
RX 550 unlisted or below threshold.
RX 480 0.73% (-0.01%)
RX 470 0.23% (-0.01%)
RX 460 0.24% (+0.01%)

DX12 system share:
Base: 86.7% Nvidia / 13.3% AMD
New cards: 102.8% Nvidia / -2.8% AMD

Installed base.
AMD loses 0.4% in graphics.
Nvidia gains 0.9% in graphics.
Intel loses 0.5% in graphics.
Intel gains 0.5% in CPU and is now above 80% installed base.
AMD loses 0.5% in CPU and is now below 20% installed base.

-0.01% change in VR headsets down to 0.22%.

DX12 GPUs alone without OS requirement.
GTX 1080TI unlisted or below threshold.
GTX 1080 1.68% (+0.02%)
GTX 1070 3.34% (-0.05%)
GTX 1060 5.19% (+0.17%)
GTX 1050TI 2.04% (+0.30%)
GTX 1050 1.54%(+0.46%)
GTX 1030 unlisted or below threshold.
RX 580 unlisted or below threshold.
RX 570 unlisted or below threshold.
RX 560 unlisted or below threshold.
RX 550 unlisted or below threshold.
RX 480 1.06% (-0.04%)
RX 470 0.38% (-0.02%)
RX 460 0.51% (+0.06%)

DX12 GPU share:
Base: 91.2% Nvidia / 8.8% AMD
New cards: 100% Nvidia / 0% AMD

Interesting to see 1070, 480 and 470 in decline in both cases. Perhaps in reality due to stagnation while total volume for the segment grows via other cards. Also resulting in the new cards distribution being very odd this time around.

8 core CPUs grew 0.03%. 6 cores grew 0.01%.
 
Last edited:
Algrim said:
Do you?
Click to expand...
If you don 't know you can't defend it.

Factum said:
This thread is a sad example of fallicies and ignorance based "argumentation".

Even if Steam's hardware survey NEVER polls you, your famile, your friends and EVERY single person you have EVER encountered...it would not alter the validity or accurracy of the poll.
Click to expand...
It sure raises some questions about their methods, and you can't answer so stop assuming they're perfect.
 
Factum said:
I like that you have no facts...nor it seems any clues to polling.
Try again.
Click to expand...
Where are yours? Like many have posted, Steam could be very accurate up to every user they have if they so deemed it. We here in this US know full well how accurate polling is (/s).
 
JustReason said:
Where are yours? Like many have posted, Steam could be very accurate up to every user they have if they so deemed it. We here in this US know full well how accurate polling is (/s).
Click to expand...

Are you trying to compare political polling with an undecided factor vs fixed hardware polling?

There is no lack of excuses are there.
 
Shintai said:
Are you trying to compare political polling with an undecided factor vs fixed hardware polling?

There is no lack of excuses are there.
Click to expand...
Since they are polling randomly then it isn't much different. But again you have any factual basis or information as to accuracy then by all means share.
 
JustReason said:
Where are yours? Like many have posted, Steam could be very accurate up to every user they have if they so deemed it. We here in this US know full well how accurate polling is (/s).
Click to expand...


Different type of polling man, there is no selective demographics going on with Steam.

And Steam "trend" is pretty much what is going on in the JPR and Mercury numbers, so either way you look at it, AMD is not doing any good with marketshare. Hence why they lost a couple marketshare points last quarter, a quarter which is seasonly week, and should have hurt nV more because nV has more to loss.
 
Why would the AMD fanboys want to skew the survey? What could they possibly gain by providing fake numbers?
 
You must log in or register to reply here.
Back
Top