OCR MEI S3 20th January 2012

Did everyone find the CV for the confidence interval in 2.(v) by using the t-distribution or the Normal distribution tables? Because n=20 (which is small, so i thought t-distribution would be correct), but there isn't a value for v=19 in the t-distribution table...

I think I used the normal distribution tables there, in the question I think it said that the standard deviation was assumed to be 15 (?) so as it wasn't estimated from sample data I used the normal tables

The normal distribution question where you had to find the probablity that two small bags differed by less than 25 grams, were you meant to find P(-25<S1-S2<25)? I don't remember seeing a question along these lines in any of the past papers so it confused me a little Generally though I thought the paper was ok, anyone remember any of the test statistics they got for any of the questions?

I think that's right yeah. I'm kicking myself though - I believe I wrote P(S1-S2<25) or something like that...

Can't really remember... For Wilcoxon I think I got W=6, which was larger than critical value, which was 5.

Does anyone remember how many estimated parameters you took away for the chi-squared one?

I think I ended up with v=2, because you had 5 clases, and you had to merge two of them as one of the expected frequences was less than 5, giving 4 classes; and the mean was used from the observed frequencies to calculate the expected frquencies, but nothing was mentioned about standard deviations or variance, so I ended up with v=4-1-1=2.
I remember the critical value being 5, but I think I got something like W=11
I can't remember the test statistics for other tests, but I think the t distribution question rejected the null hypothesis and the chi-squared one accepted the null hypothesis, though marginally.

Ah good, I ended up with v=2 too

Not so good with the W value... I got W- as my smaller one, did you?

I think I ended up with v=2, because you had 5 clases, and you had to merge two of them as one of the expected frequences was less than 5, giving 4 classes; and the mean was used from the observed frequencies to calculate the expected frquencies, but nothing was mentioned about standard deviations or variance, so I ended up with v=4-1-1=2.
I remember the critical value being 5, but I think I got something like W=11
I can't remember the test statistics for other tests, but I think the t distribution question rejected the null hypothesis and the chi-squared one accepted the null hypothesis, though marginally.

Yes i agree with you here and i felt quite confident about these questions

You mean you got W=11 as well?

Well, here goes a lot of marks...

It was something like that... because n = 8, so 0.5n(n+1) = 36 and i remember both W's were double digits. The other was around 25 so W was about 11. Either way it was above the critical value, so H1 was accepted

I randomly just remembered in the last two parts of question 4 that the variance i got with the central limit theorem was 0.28125 (9/32) does this ring any bells?