There are better statistical methods to use, but before you worry about those, this might be a good confidence boost). We trust studies more that have smaller standard errors, so take all of the means of the studies and weight their means by the standard error (standard error = standard deviation/sqrt(sample size)), this will give you a preliminary estimate of the overall mean. You have all the information right in front of you to do a basic version of this (although we would probably prefer more information, for instance the data itself so we could use a Bootstrap or similar to calculate standard error), and so it might raise your confidence to try this. What you want to do is get an overarching value for the mean for any population. Your supervisor should be able to provide more direction, but assuming your datasets all do the same thing (not true if you're doing a meta-analysis in ecology or evolution say) and there's no secondary factors that you have to consider (a big assumption).
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |