prev next front |1 |2 |3 |4 |5 |6 |7 |8 |9 |10 |11 |12 |13 |14 |15 |16 |17 |18 |19 |20 |21 |review

Search for most updated materials ↑

 
The purpose of conducting a statistical analysis is to determine a summary estimate of effect. The effect measures could be rate difference, relative risk estimate (e.g. an odds’ ratio or relative risk) or a rate ratio. Selecting an effects model is important. Usually the choice is between fixed effects model, which indicates that the conclusions derived in the meta-analysis are valid for the studies included in the analysis, and random effects model, which assume that the studies included in the meta-analysis belong to a random sample of a universe of such studies. When the studies are found to be homogeneous, random and fixed effects models are indistinguishable. Finally, conduct a sensitivity analysis to test that the model you propose is a good model.