Thread: Help!! need help with Statistics for a meta analysis

Originally Posted by flic

I'm writting a short review on how i propose to do a meta analysis on the effect of alkalising agents on sports performance. It needs to be handed on tomorrow! (eek!) There are a number of variables namely: age, **** doseage, alkalising agent used, then the studies have used many different ways of recording data such as blood pH levels, blood lactate levels etc. I am really confused and dont know where to start!. Which statisics would I use and why?,
Thank you to any one who may help!

Use standardised scores (effectivly use the z-score for each studies results - though as you will be using sample based estimates for the SD's this technicaly will be a t-score I suppose).

If the sample sizes are large-ish the sum of the scores can be treated as normal, othewise I would think about generating a empirical-bootstrapped distribution for the sum of scores.

RonL