In an attempt to promote the transformation of information into knowledge, this article compares meta-analyses with two of its variants, mega-analyses and retro-analyses, techniques developed to summarize and compare results that are not very accurate or contradictory. While meta-analyses refer to the synthesis of results of different primary researches. mega-analyses refer to the synthesis of results collected by primary researches and retroanalyses, to the recreation and posterior synthesis of that data. Using a simulation, Study 1 pairs the average trends from meta, mega and retroanalyses regarding nine different configurations that vary according to the number of studies used and the size of the individual sample of each of those studies, in the full data, light data and severe lack of data modes. In Study 2, the same data used in the simulation of severe lack of data were used to test the accuracy of mega and retroanalyses when performing data imputation. The results show a slight general superiority of mega-analyses, a difference that fades away in ideal situations.
Meta-analysis; Mega-analyses; Retroanalyses; Knowledge; Statistical analysis
