WaccoBB Polling

[Valley Oak]

As far as polls go, I took a statistics class at SRJC in 2001 and one of the many lessons I learned is that any study with less than 30 samples (or votes in this case) has very little or no validity.

There are some situations where the availability of 30 samples or more cannot exist or is impossible to garner, and there are specific statistical formulas for those situations. But as a general rule, until we get at least 30 votes on our poll, it has very little representative value.

Another important weakness is that this is an informal poll in a website and these are notoriously unreliable, even with 30 samples or more. Not everyone in the Wacco list is participating, obviously. And I don't believe that the Wacco list accurately represents the entirety of the Sonoma County population and its diverse demography.

So if we want some semblance of a "litmus test" then we need to get to 30 votes. The more, the better. And then we can look at the results of this poll with a grain of salt.

Edward

[jbox]

As far as polls go, I took a statistics class at SRJC in 2001 and one of the many lessons I learned is that any study with less than 30 samples (or votes in this case) has very little or no validity.

I see, Edward, if we 30 votes on a wacco poll that will have validity, but less than 30 might not be entirely valid so we need to be scientific about it. If we got 50 votes that would be a pretty good representative sample of the population, maybe? Uh huh, oh ya sure, you betcha....pass the pepper, please

[Dixon]

I see, Edward, if we 30 votes on a wacco poll that will have validity, but less than 30 might not be entirely valid so we need to be scientific about it. If we got 50 votes that would be a pretty good representative sample of the population, maybe? Uh huh, oh ya sure, you betcha....

Well of course there's not much difference between, say, a sample size of 29 and one of 31. 30 is just the more or less arbitrary ballpark figure scientists commonly cite as the minimum sample necessary to get valid conclusions (since, all other things being equal, the bigger the sample size the more likely the results actually reflect the truth about the larger population)--and that's assuming that the sampling was done properly, the questions were worded properly, etc. Of course, none of that is particularly relevant here, as Barry has not made any claim that this is a scientific poll or anything more than a little fun thing. And that's good, because this sample of respondents, regardless of sample size, will be "self-selected". They could only be considered representative of Waccoids who are interested enough to have looked at this thread, and only then if the sample size is decent--say, oh, about 30 or so.

[Sara S]

About 25 years ago, there was a list of books recommended by some UC professors as "books they wish all incoming freshmen had read," and one was a classic: "How to Lie With Statistics." That book changed my view of statistics forever.....

As far as polls go, I took a statistics class at SRJC in 2001 and one of the many lessons I learned is that any study with less than 30 samples (or votes in this case) has very little or no validity.

[Dixon]

About 25 years ago, there was a list of books recommended by some UC professors as "books they wish all incoming freshmen had read," and one was a classic: "How to Lie With Statistics." That book changed my view of statistics forever.....

Yes, of course it's possible to lie with statistics, just as it's possible to lie with words or pictures. Unfortunately, some people use that fact as an excuse to reject any statistics they don't want to acknowledge. A better response would be to educate ourselves about how statistics are commonly misused or misinterpreted so we can distinguish between proper and fallacious or dishonest uses. I'll add that book to my reading list.