JCC Help me with a Stat problem

Guys, this is a review test preparing for the real one and I have no idea what I'm doing. Short preface: It's a community college class, the teacher is from the Cameroon and he is good at explaining things but he takes way too long to do it and doesn't properly show us how to plug the information into our calculators essentially.

Here's the problem: A controversial bill is being debated in the state legislature. Representative Williams wants to estimate within 3 percentage points and with 90% confidence the difference in the proportion of her male and female constituents who favor the bill. What sample size should she obtain?

The problem I'm coming up against is that I'm not given the standard deviation. I've searched online for about 45 minutes and I'm not finding anything so I'd hoped that one of you would know how to do it.

 

Ramza

 

How can they specify both the confidence level and say "within 3 percentage points?" Shouldn't your level of confidence determine how wide or narrow your error bars are?

 

I really have no idea. According to the teacher, he just tells it to generate problems and the program does it. He says he didn't really look at many of the problems and that the one problem we didn't even learn how to do.

I'm almost out of this hellhole of a college, I just have to pass this and microeconomics.

Also, I know the answer. Just don't know how to get it.

 

Calculation of sample size:
We already know that the margin of error is 1.96 times the
standard error and that the standard error is
r
^p(1^p)
n . In general
the formula is
ME = z
s
^p(1 ^p)
n
(y)
where
 ME is the desired margin of error
 z is the z-score, e.g. 1.645 for a 90% condence interval,
1.96 for a 90% condence interval, 2.58 for a 99% condence
interval (see Table 8.2, page 369)
 ^p is our prior judgment of the correct value of p.
 n is the sample size (to be found)

http://www.unc.edu/~rls/s151-2010/class23.pdf

Or: http://www.macorr.com/sample-size-calculator.htm

 

Hello Ross

 

Well, if it's a 90% confidence, wouldn't you just set .03 = .82/sqrt(Sample Size) and solve?

Edit: I mean, AFAIK we have no p value, so that's about all you can do.

 

I don't see how that could possibly be true, if only because the answer barely breaches 7.

 

That's because I typed .082 when I meant .82. I've had a few.

 

Ah... nothing like drunken math. Fond mammeries.

 

Since I understand just about half of what you guys posted, what would the answer be then? The answer I'm provided with is 1504.

 

Yeah, that's not the answer Ramza's approach gives.

 

Yeah because I get 6.lots of numbers E

 

Pics plz.

 

lol and going into the example dp4m provided they essentially surmise ^p=0.5 and end up with an answer of 1537 which is much closer, but still not right.

 

No, man. That's even more wrong. Ramza has edited to correct the decimal point. But even when you do the math right and use the right numbers the approach still doesn't yield 1504.

 

Yeah I'm just going to conclude this problem is **** and not worry about it. I'll probably ask him about it before the test Monday. Thanks guys.

The answer I'm provided with says exactly: n1=n2=1504. Does that mean anything?

 

NO YOU WON'T.

This is our Malaysian airline story. We are finding the answer here, and no one is logging off until we do.

 

That'd be because of something both dp4m and myself missed on the first pass, which is

In other words, there are two groups of respondents that we're differentiating.

That means you need to take either mine (Or dp4m's) formula and stick a factor of 2 on sample size (IE 1.645*sqrt(.25/(2n)) where n is sample size). That gives 1503.347, or 1504 since you probably have to round up.

I apologize for that inconvenience, darthcaedus1138

 

Anyway thank you very much, I very much doubt we'll actually be tested on that but it's nice to know how to answer it.

Thanks Ramza, dp4m, Jabba-wocky

EDIT: Also you're right, one thing I learned tonight is that you have to round up.

 

For everyone keeping records, this counts as a namedrop equivalent to an anniversary thread.

 

No problem. Your thread has served as a poignant reminder for why I do silly nonsense math and not stats.