# research statistics problem

#### potatoeyed

##### New Member
Hi, I'm helping a friend with the statistical analysis of some research she's supposed to do. Unfortunately she's getting no support from her advisors and there are no other graduate students in the program she can go to. We are trying to figure out how a previous research article found a statistical difference in the number of sightings observed for each of five years. The first year there were 19 sightings, the second 61, third 33, fourth 24, and fifth 20. The article gives the statistics for this difference as (t=4.1, DF=4, P<0.05).

It puzzles us, because this looks like a t-test, but it seems like we're comparing five different groups, which might indicate the use of an ANOVA. But we have no data for any within-group variation to produce a meaningful f value. We're not sure how to look at this in terms of a t-test either, because we are not comparing one or two samples and there doesn't appear to be any population parameters available.

Her advisor said this was supposed to be very easy, very basic statistics, but we've been going through a number of stats books and are really struggling with finding the right strategy for this context. Any help would be appreciated. We are getting desperate.

Thanks

#### sarahc123

##### New Member
Hmm, that is weird. I would've guessed that they had just done a t-test on the smallest difference between two means, but even that doesn't make sense given the degrees of freedom (not to mention the fact that 19 vs. 20 would probably not yield that t-value). But, wait...those aren't means, those are frequencies (i.e., a strict count of number of sightings), which may indicate some kind of non-parametric test. Again, though, that would be a chi-squared test...there's no non-parametric test that uses the t statistic, to my knowledge. If I were you, I'd email the authors of the article the question that you posted here. That is perfectly legitimate if you cannot figure this out on your own. They will probably be helpful. Good luck!

#### jkotlerman

##### New Member
I agree, these are actual values not means, so its can't be an anova. It might be a straight one sample t-test, so what we need to know is whether there is a comparison value that they used. There should be a description of the test in the statistical methods section.

Jenny Kotlerman
www.statisticalconsultingnetwork.com

#### potatoeyed

##### New Member
Thanks for the replies Sarah and Jenny,
It was good to have that feedback to know I wasn't completely overlooking something. My friend managed to solicit help from another researcher who was able to walk her through the process. I didn't ask her exactly how it was done because she's been so busy, but I'll try to the next time I see her. Once again, thanks for the insight.