I wish I still remember enough specifics about the various statistics classes I took. I helped out with numerous graduate projects (Anthropology and Sociology) in college. Sadly, those classes and projects are 40 years and 2 (maybe more) strokes ago. I do remember being surprised at exactly how small a sample size could be considered statistically valid. In some cases, a sample size of 128 could be useful.
With an indefinite population size (all gamers), I think the usual minimal sample size is 1024, with a 1 standard deviation error. I suspect the survey that generated these results tried to query 1024 gamers, got 300 responses and proceeded from there with statistical analysis on that 300 respondents. It may be difficult to do the math to consider this a valid survey.
My suspicion. This is an implementation of Mark Twain's view on statistics ('Liars, Damned Liars, and Statisticians) applied to some marketing survey generating an overly optimistic picture.
Logic, my dear, merely enables one to be wrong with great authority.
scanning this article is appears for 95% confidence we need almost 400 per 1 million. what is the sliding scale of that I dont know (85% confidence?) etc. also does it need to 74% of repondence for it to be 'majority'? well no it doesnt etc
For large populations such as this, a reasonable approximation for a 95% confidence interval is 1/sqrt(N).
For 300 people, if the sample was truly representative than there would be a 95% chance that the true portion would be 74% +/- 5.8%
what should the population size be for 95%? its been a long time since college
Population size doesn't matter. It only matters if you get an unbiased sample.
Which, for a population as broad as 'gamers', is pretty damn tough to do.
Political polls are typically in the 500-1000 sample size range
yeah that is not right sorry.
I know that much, sample size of the population size does matter.
It matters to the extent that a 10 size sample might be useful for a population of 30 but not useful for a population of 100,000. But the confidence interval I posted above is for large population sizes.
You are going to have a fairly high margin of error at 300, but it would certainly be a ballpark figure if it was an unbiased sample with well worded questions. You would be able to at least draw the conclusion that more than 60% of gamers plan to purchase a VR set with very high confidence.
For large populations such as this, a reasonable approximation for a 95% confidence interval is 1/sqrt(N).
For 300 people, if the sample was truly representative than there would be a 95% chance that the true portion would be 74% +/- 5.8%
what should the population size be for 95%? its been a long time since college
Population size doesn't matter. It only matters if you get an unbiased sample.
Which, for a population as broad as 'gamers', is pretty damn tough to do.
Political polls are typically in the 500-1000 sample size range
yeah that is not right sorry.
I know that much, sample size of the population size does matter.
It matters to the extent that a 10 size sample might be useful for a population of 30 but not useful for a population of 100,000. But the confidence interval I posted above is for large population sizes.
You are going to have a fairly high margin of error at 300, but it would certainly be a ballpark figure if it was an unbiased sample with well worded questions. You would be able to at least draw the conclusion that more than 60% of gamers plan to purchase a VR set with very high confidence.
two entirely different subjects.
before you even THINK about crafting the questions you have to decide on the sample size. The sample size is separate from the question of bias.
In a matter of a few unlazy mins on my part I discovered you need almost 400 per 1 million for a 95% confidence level. that is what I was asking
Please do not respond to me, even if I ask you a question, its rhetorical.
scanning this article is appears for 95% confidence we need almost 400 per 1 million. what is the sliding scale of that I dont know (85% confidence?) etc. also does it need to 74% of repondence for it to be 'majority'? well no it doesnt etc
It's all about how sure you want to be. for a 95% confidence level +/- 6% for a population of 10,000,000 you'd need a sample size of 267.
So in this case, if they say 74% of people want vr you can extrapolate that for a population of 10 million minimally 68% want vr.
Honestly depending on what they asked this could well be true. If you asked me "are you going to buy a rift/vive/etc" I'd say no. If you asked me if I like the idea of VR and want to get it I'd say yes. I wouldn't take this to mean 68% of people want to buy today's vr solution, more that 68% of people want the experience.
For large populations such as this, a reasonable approximation for a 95% confidence interval is 1/sqrt(N).
For 300 people, if the sample was truly representative than there would be a 95% chance that the true portion would be 74% +/- 5.8%
what should the population size be for 95%? its been a long time since college
Population size doesn't matter. It only matters if you get an unbiased sample.
Which, for a population as broad as 'gamers', is pretty damn tough to do.
Political polls are typically in the 500-1000 sample size range
yeah that is not right sorry.
I know that much, sample size of the population size does matter.
It matters to the extent that a 10 size sample might be useful for a population of 30 but not useful for a population of 100,000. But the confidence interval I posted above is for large population sizes.
You are going to have a fairly high margin of error at 300, but it would certainly be a ballpark figure if it was an unbiased sample with well worded questions. You would be able to at least draw the conclusion that more than 60% of gamers plan to purchase a VR set with very high confidence.
two entirely different subjects.
before you even THINK about crafting the questions you have to decide on the sample size. The sample size is separate from the question of bias.
In a matter of a few unlazy mins on my part I discovered you need almost 400 per 1 million for a 95% confidence level. that is what I was asking
Look at the chart on that page. notice that the surveys needed levels off at high populations?
you may have heard of Rasmussen. They are one of the leaders in political polls. they are using a sample size of 1,000 with a 3% margin of error and 95% confidence. Population size of voting americans is ~125M.
scanning this article is appears for 95% confidence we need almost 400 per 1 million. what is the sliding scale of that I dont know (85% confidence?) etc. also does it need to 74% of repondence for it to be 'majority'? well no it doesnt etc
It's all about how sure you want to be. for a 95% confidence level +/- 6% for a population of 10,000,000 you'd need a sample size of 267.
So in this case, if they say 74% of people want vr you can extrapolate that for a population of 10 million minimally 68% want vr.
Honestly depending on what they asked this could well be true. If you asked me "are you going to buy a rift/vive/etc" I'd say no. If you asked me if I like the idea of VR and want to get it I'd say yes. I wouldn't take this to mean 68% of people want to buy today's vr solution, more that 68% of people want the experience.
from the math side that is the answer I was looking for.
So basically possible question bias and sample size bias aside 300 count for a deviation of 6% is reasonable. ok cool.
the indentended population is 'serious gamers' (not what they said exactly but yeah gamers) so I think the selected sample is good. leaving only the questions I think those questions would have to be wildly obviously wrong in order to sway the result set.
which incidentally means on gaming forums like this one there are more laggards then not.
Please do not respond to me, even if I ask you a question, its rhetorical.
before you even THINK about crafting the questions you have to decide on the sample size. The sample size is separate from the question of bias.
Yes, they are two entirely different subjects.
My point is the sample size isnt the big issue with this poll. Its almost certainly incredibly biased, although no methodology is available for it that i can see.
before you even THINK about crafting the questions you have to decide on the sample size. The sample size is separate from the question of bias.
Yes, they are two entirely different subjects.
My point is the sample size isnt the big issue with this poll. Its almost certainly incredibly biased, although no methodology is available for it that i can see.
1. I think sample size is the most important part someone broke down the math for me in a way that I asked and it seems 300 is actually not to bad of a sample size 2. the intented sample was of 'gamers' people who game often. that requirement seems to have been met. 3. leaving only the questions. which leaves this question, how does one ask 'are you planning on buying a VR headset' if you want to know how many people are 'planning to buy a headset'. I highly suggest to read the lead of the article and try to think about what they are saying the study shows is somehow different. They are not saying '74% of people will buy' they are not saying '74% of the people have bought' they are pretty clear, they are saying 74% of the people want to buy. seems straightforward to me
Please do not respond to me, even if I ask you a question, its rhetorical.
I'm looking to buy a flying car but won't for a long time because of current cost and viability.
What matters is what people actually buy.
yes! and 'want to buy a porche' as part of a survey question and 'plan to buy a porche' on a survey question are not the same quesiton
I think his point is that entrance polls are much less accurate than exit polls. Someone can say they're going to do something, but it's less relevant than what they actually do.
but the article is not suggesting anyone is going to buy a VR headset is my point
Please do not respond to me, even if I ask you a question, its rhetorical.
2. the intented sample was of 'gamers' people who game often. that requirement seems to have been met.
So you think that a sample where 56% of people would consider giving a VR set as a gift is a reasonable snapshot of gamers?
I would not consider the questions ask to mean that.
1. I would not suggest that 100% of the people answered the question as 'gift' 2. I would not suggest the questions even implied what you are implying
I would however suggest based on the article that the minority (laggards in this case) post on these forums or they are just trolling but I think its former
Please do not respond to me, even if I ask you a question, its rhetorical.
What did they do review games that could ONLY be played with Vr?
Like where the heck would they get a number like that ,,geesus murphy,such a retarded number to toss around with absolutely no way of knowing that what so ever.
Never forget 3 mile Island and never trust a government official or company spokesman.
OP linked an article referencing a poll that is an obvious attempt at posting intentionally misleading 'data'.
He then tries to divert this by calling out that is, in fact, a potentially bullshit data set, but for a reason he knows is incorrect.
I point out that the sample size is fine, OP plays dumb and misinterprets (most likely on purpose) some random survey selling site's information. then another poster posts something fairly similar to what I posted earlier, and now OP is "well, looks like the survey is probably fine after all."
If you really feel VR is going to be that great down the road and want to argue it, fine. But if I wanted to see this kind of deceitful manipulation I would turn on coverage of the US presidential race.
I thought the survey was very good. I admit the sample size was a bit small, but it suggests people are aware of VR and intend to buy VR products in the future. Am I missing something? I don't see the problem with this survey. Maybe the sample size should be bigger, but you can get a good indication what people think surveying 300 people. I guess if you think the survey is hogwash and doesn't come from a reliable source then I could see a problem.
Comments
Please do not respond to me, even if I ask you a question, its rhetorical.
Please do not respond to me
Which, for a population as broad as 'gamers', is pretty damn tough to do.
Political polls are typically in the 500-1000 sample size range
¯\_(ツ)_/¯
I know that much, sample size of the population size does matter.
Please do not respond to me, even if I ask you a question, its rhetorical.
Please do not respond to me
With an indefinite population size (all gamers), I think the usual minimal sample size is 1024, with a 1 standard deviation error. I suspect the survey that generated these results tried to query 1024 gamers, got 300 responses and proceeded from there with statistical analysis on that 300 respondents. It may be difficult to do the math to consider this a valid survey.
My suspicion. This is an implementation of Mark Twain's view on statistics ('Liars, Damned Liars, and Statisticians) applied to some marketing survey generating an overly optimistic picture.
Logic, my dear, merely enables one to be wrong with great authority.
https://www.checkmarket.com/blog/how-to-estimate-your-population-and-survey-sample-size/
Please do not respond to me, even if I ask you a question, its rhetorical.
Please do not respond to me
You are going to have a fairly high margin of error at 300, but it would certainly be a ballpark figure if it was an unbiased sample with well worded questions. You would be able to at least draw the conclusion that more than 60% of gamers plan to purchase a VR set with very high confidence.
before you even THINK about crafting the questions you have to decide on the sample size. The sample size is separate from the question of bias.
In a matter of a few unlazy mins on my part I discovered you need almost 400 per 1 million for a 95% confidence level. that is what I was asking
Please do not respond to me, even if I ask you a question, its rhetorical.
Please do not respond to me
So in this case, if they say 74% of people want vr you can extrapolate that for a population of 10 million minimally 68% want vr.
Honestly depending on what they asked this could well be true. If you asked me "are you going to buy a rift/vive/etc" I'd say no. If you asked me if I like the idea of VR and want to get it I'd say yes. I wouldn't take this to mean 68% of people want to buy today's vr solution, more that 68% of people want the experience.
Take a look here
http://www.rasmussenreports.com/public_content/politics
you may have heard of Rasmussen. They are one of the leaders in political polls. they are using a sample size of 1,000 with a 3% margin of error and 95% confidence. Population size of voting americans is ~125M.
So basically possible question bias and sample size bias aside 300 count for a deviation of 6% is reasonable. ok cool.
the indentended population is 'serious gamers' (not what they said exactly but yeah gamers) so I think the selected sample is good. leaving only the questions I think those questions would have to be wildly obviously wrong in order to sway the result set.
which incidentally means on gaming forums like this one there are more laggards then not.
Please do not respond to me, even if I ask you a question, its rhetorical.
Please do not respond to me
My point is the sample size isnt the big issue with this poll. Its almost certainly incredibly biased, although no methodology is available for it that i can see.
2. the intented sample was of 'gamers' people who game often. that requirement seems to have been met.
3. leaving only the questions. which leaves this question, how does one ask 'are you planning on buying a VR headset' if you want to know how many people are 'planning to buy a headset'. I highly suggest to read the lead of the article and try to think about what they are saying the study shows is somehow different. They are not saying '74% of people will buy' they are not saying '74% of the people have bought' they are pretty clear, they are saying 74% of the people want to buy. seems straightforward to me
Please do not respond to me, even if I ask you a question, its rhetorical.
Please do not respond to me
Please do not respond to me, even if I ask you a question, its rhetorical.
Please do not respond to me
1. I would not suggest that 100% of the people answered the question as 'gift'
2. I would not suggest the questions even implied what you are implying
I would however suggest based on the article that the minority (laggards in this case) post on these forums or they are just trolling but I think its former
Please do not respond to me, even if I ask you a question, its rhetorical.
Please do not respond to me
Read more at http://gamingbolt.com/74-console-and-pc-players-looking-to-buy-a-vr-product-according-to-study#R2xe5oKsHWHKeuC0.99
Please do not respond to me, even if I ask you a question, its rhetorical.
Please do not respond to me
What did they do review games that could ONLY be played with Vr?
Like where the heck would they get a number like that ,,geesus murphy,such a retarded number to toss around with absolutely no way of knowing that what so ever.
Never forget 3 mile Island and never trust a government official or company spokesman.
OP linked an article referencing a poll that is an obvious attempt at posting intentionally misleading 'data'.
He then tries to divert this by calling out that is, in fact, a potentially bullshit data set, but for a reason he knows is incorrect.
I point out that the sample size is fine, OP plays dumb and misinterprets (most likely on purpose) some random survey selling site's information. then another poster posts something fairly similar to what I posted earlier, and now OP is "well, looks like the survey is probably fine after all."
If you really feel VR is going to be that great down the road and want to argue it, fine. But if I wanted to see this kind of deceitful manipulation I would turn on coverage of the US presidential race.
However, I posted this earlier in the thread:
"For large populations such as this, a reasonable approximation for a 95% confidence interval is 1/sqrt(N).
For 300 people, if the sample was truly representative than there would be a 95% chance that the true portion would be 74% +/- 5.8%"
That wasn't good enough for him though. Pretty strange.
And as its a cheap purchase they will happily give it as a give or hope to get it as a gift.