IT SUPPORTED STATISTICAL INFERENCE: A COMBINED META- ANALYSIS OF TERRITORIALLY STRATIFIED DATA
There are various designs for inference about stratified data, based on stratified sampling schemes. The approach in this paper proposes to deploy the methodology of meta-analysis, in particular the combined frequentist and Bayesian model for territorially stratified data. The research presented in this paper uses meta-regression of statistical data collected in Federation of Bosnia and Herzegovina, territorially stratified in administrative units – 10 cantons. In order to show methodologically richer example, we use odds ratio of new versus old passenger versus transport vehicle purchase, so infringing on one aspect of the consumer power in BiH. The proposed estimate is the REM Bayesian regression (log odds), which is impossible to derive without the use of IT. We use the results of FEM and REM frequentist meta-analysis to frame and test the results.