Psyc930 Meta-Analysis Project
The purpose of the project is to identify articles that examine the same basic bivariate relationship and use these to perform four common meta-analytic analyses.
Your project database must include at least:
· 12 appropriate effect sizes
· 5 coding variables appropriate for a meta-regression analysis
o You must have complete coding for all 12 effect sizes
o At least one of the coded variables must be a 2-category variable
Warning: Finding 12 appropriate effect sizes that you can code with all 5 coding variables may require you to examine way more than 12 articles!!!!
Remember: You can often get multiple effect sizes from a single publication that has multiple studies, factorial designs,, etc.
Project analyses include:
· Overall effect size (transformed & inverse variance weighted)
o Compose a xls spread sheet for a fixed effect model & write up the interpretations
o Compute the transformed variance weighted ES and convert it to r/d/OR
o Compute the significance test of variance weighted ES & interpret it
o Compute the 95% confidence interval for ES, convert it to r-values & interpret it
· Overall effect size (transformed inverse variance weight)
o Use Wilson’s MetaES macro to compute fixed and random effect ES
o Convert both to d/r/OR
· Homogeneity test of the effect sizes
o Compose a xls spread sheet for a fixed effect model & write up the interpretations
o Be sure to compute and show all 3 significance tests & interpret each
· Group comparison analysis
o Identify some a 2-category variable you can use to divide and compare the studies
o Use Wilson’s MetaF macro to compute fixed and random effect Q-test
· Meta-Regression Analysis
o Must include all 5 coding variables (which have complete data)
o Use Wilson’s MetaReg macro to compute fixed and random effect models
o Write up proper interpretations of…
§ Model significance test & effect size
§ Each coding variable’s significance test and regression weight