poll: camera freezing?

[Gary02468]

(Though even the best sampling strategies cannot overcome too small [] a sample, Gallup be damned).

Gallup has math on his side. If the sample here were random (which, as I acknowledge, it is not), then with 36% of 58 respondents reporting camera-freezing, the math shows that we can be 95% confident that 23-49% of all Note9 s have camera-freezing, and 99% confident that 19-52% of all Note9s have camera-freezing. That's not precise enough to predict the outcome of an election, but it's precise enough to show that there's a widespread problem, which is my only goal here.

Our sample is self-selected, though, and we can only guess at the direction and magnitude of the bias. But even if camera-freeze owners are three or four times more likely to participate in the poll than others, the poll still demonstrates a widespread problem. I doubt they're even twice as likely to participate, but that's a subjective impression that would be hard to support or refute.

 

[lcs101]

Gallup has math on his side. If the sample here were random (which, as I acknowledge, it is not), then with 36% of 58 respondents reporting camera-freezing, the math shows that we can be 95% confident that 23-49% of all Note9 s have camera-freezing, and 99% confident that 19-52% of all Note9s have camera-freezing. That's not precise enough to predict the outcome of an election, but it's precise enough to show that there's a widespread problem, which is my only goal here.

Our sample is self-selected, though, and we can only guess at the direction and magnitude of the bias. But even if camera-freeze owners are three or four times more likely to participate in the poll than others, the poll still demonstrates a widespread problem. I doubt they're even twice as likely to participate, but that's a subjective impression that would be hard to support or refute.

From a research perspective, that's pretty crummy precision though. And as you point out, because of the sample, we cannot even say it's anywhere close to that precise, which was my primary point. And keep in mind, it's not just about who is more likely to participate in terms of having problems or not, it's also who is likely to be on Android Central. There is nothing representative about those on this site/forum in terms of the Note 9 owners, of course.

As I said previously, I never claimed that it was not a real issue. I just took (statistical) issue with the statement that around a third of people were having the problem, which is lacking in statistical evidence. Now of course it doesn't have to be anywhere close to that to be a real issue. Even 5% for example would indicate a significant problem.

My apologies to all of the people we have bored with this statistical debate. Kind of kicking a dead horse at this point.

 

[Gary02468]

From a research perspective, that's pretty crummy precision though. And as you point out, because of the sample, we cannot even say it's anywhere close to that precise, which was my primary point.

I don't think we disagree, though we may be talking at cross-purposes. I agree that the poll doesn't pass muster from a research perspective; I just think it's better than nothing as a basis for a ballpark guess. I've seen forum posts suggesting that fewer than 1% of Note9s may have the problem, and I think the poll gives evidence to reject that claim as highly implausible.