If anyone could help, I would appreciate it very much! Ta.

It sounds like you want to do a chi-square test, anonymology. A stats package should be able to do this. In SPSS and Stata this would involve creating a cross-tabulation between the variables gender and dataset source, and asking the program to also generate a chi-square statistic.

Manually calculating it isn't terribly difficult once you know what you are doing, but it is tricky to explain over the Internet. I could send you my PowerPoint presentation which I use to teach it in my stats class, which should help you manually calculate it.

(just by coincidence, it turns I am covering chi-square tomorrow).

AH! Now, Chi-square is the one and only statistic I'm familiar with!! I think I know what you're talking about, and I think I know how to do it.

Let me think. Right, if, say, the Chi-square is significant (?), I could conclude that the smaller set is not very representative of the larger set (because there's a significant difference between the totals)? And vice versa?

That is right anonymology. A small p-value/high chi-square statistic means there is a significant difference between the two datasets--the samples were most likely drawn from different populations.

YAY! Thank you so much Josh!

And isn't it amazing how something you learned 10 years ago can still come back to you! I'm relieved because for a moment there I thought I would have to do some massive computation that would just be beyond me!

You're welcome anon. I take you found a non-significant difference?

Well, for some yes and for some no. Yes for gender and age. No for ethnic background. That's all I done so far.

This exchange represents the awesomeness of Clipmarks very well.

You know, that's very true mousie!