Linux Users Drop to Under 1% In Latest Steam Survey

[Lightknight]

Ok, so, is no one else thinking critically about this article? This just gives % which probably means that number of linux users isn't down, there just isn't as many joining steam. Which would show the % of them is lower, despite the fact the number isn't.

It's a survey of existing users who took a survey, not new users. For example, I was asked to take the survey last month but declined since I was at my work computer and my specs wouldn't be reflective of me as a demographic.

The percentage being lower shows a decline in market share even if the actual number is the same. However, in my experience these numbers change every month. It's nearly pointless to discuss them individually as far as I can tell.

Actually, the article doesn't tell us if it is existing users over new users, then how does it define the difference between new users and existing ones? All we know is that these are percentages and it doesn't show us user numbers, so its likely that more and more people are doing the survey, which just means the number of linux users isn't growing as fast as windows ones. Which makes sense, because users are much more familiar with windows.

The steam survey is a known entity that recurs regularly. So the article doesn't need to explain the methodology for the same reason an article on home foreclosure results doesn't.

What you've got to understand is that we're dealing with proportions here. If more people take the survey then the proportions should still remain the same. So if your Linux users are 1% of your population, then if you get 100 people to take the survey then one should be a Linux user and if you get 1 million users to take it then 10,000 should be Linux users.

The number of the sample size should not impact the proportion unless you had a sample size that was too small to be representative. In this case, the numbers are well above the minimum threshold for legitimacy.

It doesn't give us numbers so we don't know, there could have been less people than the last time they issued it, we just don't know. Percentages are almost worthless by themselves and the difference if % is so small it could easily be explained by just not as many linux users doing the survey.

They went to .94%. In order to get results that break up into less than a percentage you need to have a significant number of results. This wasn't 100 respondents.

 

[Worgen]

Ok, so, is no one else thinking critically about this article? This just gives % which probably means that number of linux users isn't down, there just isn't as many joining steam. Which would show the % of them is lower, despite the fact the number isn't.

It's a survey of existing users who took a survey, not new users. For example, I was asked to take the survey last month but declined since I was at my work computer and my specs wouldn't be reflective of me as a demographic.

The percentage being lower shows a decline in market share even if the actual number is the same. However, in my experience these numbers change every month. It's nearly pointless to discuss them individually as far as I can tell.

Actually, the article doesn't tell us if it is existing users over new users, then how does it define the difference between new users and existing ones? All we know is that these are percentages and it doesn't show us user numbers, so its likely that more and more people are doing the survey, which just means the number of linux users isn't growing as fast as windows ones. Which makes sense, because users are much more familiar with windows.

The steam survey is a known entity that recurs regularly. So the article doesn't need to explain the methodology for the same reason an article on home foreclosure results doesn't.

What you've got to understand is that we're dealing with proportions here. If more people take the survey then the proportions should still remain the same. So if your Linux users are 1% of your population, then if you get 100 people to take the survey then one should be a Linux user and if you get 1 million users to take it then 10,000 should be Linux users.

The number of the sample size should not impact the proportion unless you had a sample size that was too small to be representative. In this case, the numbers are well above the minimum threshold for legitimacy.

It doesn't give us numbers so we don't know, there could have been less people than the last time they issued it, we just don't know. Percentages are almost worthless by themselves and the difference if % is so small it could easily be explained by just not as many linux users doing the survey.

They went to .94%. In order to get results that break up into less than a percentage you need to have a significant number of results. This wasn't 100 respondents.

Its a small enough change to be meaningless, it could easily just be the result of fewer linux users taking the survey. Since we don't have hard numbers we don't know.

 

[Lightknight]

Ok, so, is no one else thinking critically about this article? This just gives % which probably means that number of linux users isn't down, there just isn't as many joining steam. Which would show the % of them is lower, despite the fact the number isn't.

It's a survey of existing users who took a survey, not new users. For example, I was asked to take the survey last month but declined since I was at my work computer and my specs wouldn't be reflective of me as a demographic.

The percentage being lower shows a decline in market share even if the actual number is the same. However, in my experience these numbers change every month. It's nearly pointless to discuss them individually as far as I can tell.

Actually, the article doesn't tell us if it is existing users over new users, then how does it define the difference between new users and existing ones? All we know is that these are percentages and it doesn't show us user numbers, so its likely that more and more people are doing the survey, which just means the number of linux users isn't growing as fast as windows ones. Which makes sense, because users are much more familiar with windows.

The steam survey is a known entity that recurs regularly. So the article doesn't need to explain the methodology for the same reason an article on home foreclosure results doesn't.

What you've got to understand is that we're dealing with proportions here. If more people take the survey then the proportions should still remain the same. So if your Linux users are 1% of your population, then if you get 100 people to take the survey then one should be a Linux user and if you get 1 million users to take it then 10,000 should be Linux users.

The number of the sample size should not impact the proportion unless you had a sample size that was too small to be representative. In this case, the numbers are well above the minimum threshold for legitimacy.

It doesn't give us numbers so we don't know, there could have been less people than the last time they issued it, we just don't know. Percentages are almost worthless by themselves and the difference if % is so small it could easily be explained by just not as many linux users doing the survey.

They went to .94%. In order to get results that break up into less than a percentage you need to have a significant number of results. This wasn't 100 respondents.

Its a small enough change to be meaningless, it could easily just be the result of fewer linux users taking the survey. Since we don't have hard numbers we don't know.

That's not what I mean. I'm saying that since the number they dropped to was down to the hundredths place, we know that the number of respondents was likely a larger number of people. It is hard to get a demographic down to .94%. For example, if you have 100 respondents then you can only have whole numbers. If you have 200 respondents then you start to have half-percentages. How many respondents do you think they need to have before you can get data that is broken up into the hundredths place? For example, I've seen 1.01%, what is the lowest number of respondents where you can get that percentage out of it? I don't know and I'm not willing to do the math because they are likely rounding up or down from whatever is in the thousandths decimal place.

So they are getting a legitimate sampling of Steam gamers. That's about all we know. But seeing as this month the Linux users are down to .88% I think we can agree that these numbers fluxuate enough to indicate some sort of problem in the testing procedure. For example, two of the three times I have been surveyed have been when I'm on my work machine where I never play games. So I had to not respond those two times so as to not skew the data.

 

[Worgen]

Ok, so, is no one else thinking critically about this article? This just gives % which probably means that number of linux users isn't down, there just isn't as many joining steam. Which would show the % of them is lower, despite the fact the number isn't.

It's a survey of existing users who took a survey, not new users. For example, I was asked to take the survey last month but declined since I was at my work computer and my specs wouldn't be reflective of me as a demographic.

The percentage being lower shows a decline in market share even if the actual number is the same. However, in my experience these numbers change every month. It's nearly pointless to discuss them individually as far as I can tell.

Actually, the article doesn't tell us if it is existing users over new users, then how does it define the difference between new users and existing ones? All we know is that these are percentages and it doesn't show us user numbers, so its likely that more and more people are doing the survey, which just means the number of linux users isn't growing as fast as windows ones. Which makes sense, because users are much more familiar with windows.

The steam survey is a known entity that recurs regularly. So the article doesn't need to explain the methodology for the same reason an article on home foreclosure results doesn't.

What you've got to understand is that we're dealing with proportions here. If more people take the survey then the proportions should still remain the same. So if your Linux users are 1% of your population, then if you get 100 people to take the survey then one should be a Linux user and if you get 1 million users to take it then 10,000 should be Linux users.

The number of the sample size should not impact the proportion unless you had a sample size that was too small to be representative. In this case, the numbers are well above the minimum threshold for legitimacy.

It doesn't give us numbers so we don't know, there could have been less people than the last time they issued it, we just don't know. Percentages are almost worthless by themselves and the difference if % is so small it could easily be explained by just not as many linux users doing the survey.

They went to .94%. In order to get results that break up into less than a percentage you need to have a significant number of results. This wasn't 100 respondents.

Its a small enough change to be meaningless, it could easily just be the result of fewer linux users taking the survey. Since we don't have hard numbers we don't know.

That's not what I mean. I'm saying that since the number they dropped to was down to the hundredths place, we know that the number of respondents was likely a larger number of people. It is hard to get a demographic down to .94%. For example, if you have 100 respondents then you can only have whole numbers. If you have 200 respondents then you start to have half-percentages. How many respondents do you think they need to have before you can get data that is broken up into the hundredths place? For example, I've seen 1.01%, what is the lowest number of respondents where you can get that percentage out of it? I don't know and I'm not willing to do the math because they are likely rounding up or down from whatever is in the thousandths decimal place.

So they are getting a legitimate sampling of Steam gamers. That's about all we know. But seeing as this month the Linux users are down to .88% I think we can agree that these numbers fluxuate enough to indicate some sort of problem in the testing procedure. For example, two of the three times I have been surveyed have been when I'm on my work machine where I never play games. So I had to not respond those two times so as to not skew the data.

Well, there are well over 20 million steam accounts, but the survey could have as few as 10000 respondents since thats how many would be needed to change a % two decimal places down. From my own rather bad grasp of mathmatics. Still doesn't change the fact that percentages are pretty worthless.

 

[Lightknight]

Ok, so, is no one else thinking critically about this article? This just gives % which probably means that number of linux users isn't down, there just isn't as many joining steam. Which would show the % of them is lower, despite the fact the number isn't.

It's a survey of existing users who took a survey, not new users. For example, I was asked to take the survey last month but declined since I was at my work computer and my specs wouldn't be reflective of me as a demographic.

The percentage being lower shows a decline in market share even if the actual number is the same. However, in my experience these numbers change every month. It's nearly pointless to discuss them individually as far as I can tell.

Actually, the article doesn't tell us if it is existing users over new users, then how does it define the difference between new users and existing ones? All we know is that these are percentages and it doesn't show us user numbers, so its likely that more and more people are doing the survey, which just means the number of linux users isn't growing as fast as windows ones. Which makes sense, because users are much more familiar with windows.

The steam survey is a known entity that recurs regularly. So the article doesn't need to explain the methodology for the same reason an article on home foreclosure results doesn't.

What you've got to understand is that we're dealing with proportions here. If more people take the survey then the proportions should still remain the same. So if your Linux users are 1% of your population, then if you get 100 people to take the survey then one should be a Linux user and if you get 1 million users to take it then 10,000 should be Linux users.

The number of the sample size should not impact the proportion unless you had a sample size that was too small to be representative. In this case, the numbers are well above the minimum threshold for legitimacy.

It doesn't give us numbers so we don't know, there could have been less people than the last time they issued it, we just don't know. Percentages are almost worthless by themselves and the difference if % is so small it could easily be explained by just not as many linux users doing the survey.

They went to .94%. In order to get results that break up into less than a percentage you need to have a significant number of results. This wasn't 100 respondents.

Its a small enough change to be meaningless, it could easily just be the result of fewer linux users taking the survey. Since we don't have hard numbers we don't know.

That's not what I mean. I'm saying that since the number they dropped to was down to the hundredths place, we know that the number of respondents was likely a larger number of people. It is hard to get a demographic down to .94%. For example, if you have 100 respondents then you can only have whole numbers. If you have 200 respondents then you start to have half-percentages. How many respondents do you think they need to have before you can get data that is broken up into the hundredths place? For example, I've seen 1.01%, what is the lowest number of respondents where you can get that percentage out of it? I don't know and I'm not willing to do the math because they are likely rounding up or down from whatever is in the thousandths decimal place.

So they are getting a legitimate sampling of Steam gamers. That's about all we know. But seeing as this month the Linux users are down to .88% I think we can agree that these numbers fluxuate enough to indicate some sort of problem in the testing procedure. For example, two of the three times I have been surveyed have been when I'm on my work machine where I never play games. So I had to not respond those two times so as to not skew the data.

Well, there are well over 20 million steam accounts, but the survey could have as few as 10000 respondents since thats how many would be needed to change a % two decimal places down. From my own rather bad grasp of mathmatics. Still doesn't change the fact that percentages are pretty worthless.

They provide aggregate information about a population as long as the sample size is large enough (according to you it certainly is) and properly randomized (it is randomized).

This should mean that Linux users are under 1% of the steam gaming population depending on what the margin of error is.

Now, would you consider this fact relevant if you were developing a pc game that was going to be sold primarily on steam and were considering pumping in a significant amount of money into porting to the Linux platform? I would.