Fixed-Effect Model
Fixed-Effect Model
As many researchers engage in meta-analysis in our day today, they are guided by two
available statistical models for meta-analysis – the fixed-effect and random-effect models. Both
models are used to mathematically combine the results from multiple studies; however, they
make different assumptions about the nature of the studies. At the same time, their respective
assumptions lead to diverse descriptions for the true or combined effect and different methods
for assigning weights. Hence, researchers should be knowledgeable of each model. This paper
Borenstein, Hedges, Higgins, and Rothstein (2009) contends that when using the fixed-
effect model in the meta-analysis, an assumption that all included studies share a common effect
size is made. Moreover, all studies are making an approximation of the same effect size in this
model, hence, assigning weights to the studies is based solely on the amount of information
captured by every study, which, most of the time, is dependent on its sample size. Lastly, the
mere source of error in estimating the common true effect considered in this model is the
arbitrary error within studies, where for studies with large sample sizes, the error will have a
propensity towards zero. Simply put, this model makes the researcher assume that there is one
true effect size that lies beneath all studies in the analysis and that the dissimilarities in observed
All said, two statistical models for use in a meta-analysis are available and deciding
which model to use will depend on the researcher's expectations about the common effect size, as
well as on the purpose of the analysis. If researchers deem that all studies in the analysis are
functionally indistinguishable, and they aim to compute effect sizes for the identified population
and not to generalize to other populations, then they are correct in using the fixed-effect model.
FIXED-EFFECT MODEL 2
References
Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2009). Introduction to Meta-Analysis.