Study review question for all you fellow biostat nerds. :)

[bionerdette]

I'm thinking a welch's t-test is appropriate, but unsure of that? Can anyone lead me in the right direction?

A geographer studying “hammock islands” (small patches of tropical hardwood forest) in the Florida Everglades was interested in determining the effect of invasion by the shrub Brazilian Pepper (Schinus sp.) on the species richness of native plants in hammocks. She sampled 12 un-invaded and 8 invaded hammock islands and determined that the average species richness (mean±standard error) was 27.4±3.65 and 19.2±4.39 on un-invaded and invaded islands, respectively. Would she be safe in concluding that the invaded islands have lower species richness than un-invaded islands? Perform a statistical test and report the results of that test to support your interpretation of the data.

 

[marshmugger]

You don't have the complete dataset, only summary statistics?

You haven't converted the standard errors to variances. Are the variances nearly equal?

Both the regular t-test (easy to compute by hand) and Welch's t-test can be computed from summary statistics. All of this is should be in any biostats text.

I suggest you also report the difference and confidence interval to indicate effect size.

Stephen.