OT: Stats T-Test Question

[Hungry Jack]

Sorry to do this, but I am stuck and getting dumber. I need Wildcat assistance.

I have two populations that voted on a 5-point scale. I have two separate distributions (N=115, 243 respectively), one with a mean of 3.53 and another a mean score of 4.43. Standard deviations are .69 and .81.

My hypothesis is that the results suggest that each population applied the scale (1-5) differently, and that the difference in scoring is significant (not random) and that I should re-vote after re-orienting each team.

Should I run a student's T-Test on this to be sure? I did in Excel, and the t values are very low (.00058, 00000137). Am i seeing this right?

 

[Gladeskat]

Different sample sizes doesn't mean distributions are different. Without testing whether distributions are normal and homoscedastic (no difference in scatter), it sounds like you should run a Mann-Whitney U test.

If you do test for normality and homoscedasticity and find both samples are normally distributed with equal variance, then run the Student's T. From the data you present, I don't think it will matter much that you apply a non-parametric test like a Mann-Whitney U Test (less sensitive if data is normally distributed) to your data set. Your sampled groups are quite large and the difference seems quite large as well.

 

[FloridAlum]

Different sample sizes doesn't mean distributions are different. Without testing whether distributions are normal and homoscedastic (no difference in scatter), it sounds like you should run a Mann-Whitney U test.

If you do test for normality and homoscedasticity and find both samples are normally distributed with equal variance, then run the Student's T. From the data you present, I don't think it will matter much that you apply a non-parametric test like a Mann-Whitney U Test (less sensitive if data is normally distributed) to your data set. Your sampled groups are quite large and the difference seems quite large as well.

Excellent advice. I was going to say that myself but you beat me to the punch . If your idea does not work , I would then try tossing a coin .

 

[DaCat]

Excellent advice. I was going to say that myself but you beat me to the punch . If your idea does not work , I would then try tossing a coin .

LOL - you made me splurt out my drink. You owe me an expensive Scotch now.

 

[Hungry Jack]

Thanks Glades. I ran the Manfred Mann test and got a significant result.