Background:
In mixed studies reviewing, one possibility to match findings from qualitative and quantitative studies is to incorporate qualitative evidence in an informative prior distribution in a Bayesian meta-analysis. There are few examples of Bayesian meta-analysis incorporating qualitative evidence, and there is a need for instruction on how to use this method.
Objectives:
This study discusses a worked example of a four-step method to conduct a Bayesian meta-regression with an informative prior distribution constructed by qualitative evidence.
Methods:
The first step of the method discusses how variables from quantitative and qualitative datasets are similar with regard to their function in bivariate relationships. The second step describes the meta-regression model required for our estimates and explains where we use which information in the model. The third step describes how we adjust the meta-regression model in order to account for the differences between the two data sources. The fourth step presents the posterior distribution for the results of the worked example as well as the results of the sensitivity analyses in which we explore the influence of the precision of the prior distribution.
Results:
The results for the worked example show that the posterior distribution is quite different from the prior distribution, indicating that the quantitative and qualitative studies yielded different findings, and indicating that the prior had a considerable influence.
Conclusions:
We conclude that Bayesian meta-analysis is appropriate for matching quantitative and qualitative evidence when similarities and differences are carefully studied and accounted for.
