ESMARConf2023: MetaPipeX: Data analysis & harmonization for multi-lab replications of exptl designs скачать в хорошем качестве

ESMARConf2023: MetaPipeX: Data analysis & harmonization for multi-lab replications of exptl designs

Presenter: Jens Fünderich Authors: Fuenderich, Jens; Frank, Maximilian; Beinhauer, Lukas Session: Quantitative synthesis 2 Title: MetaPipeX: Data analysis & harmonization for multi-lab replications of experimental designs Abstract: The number of multi-lab replication studies (e.g. ManyLabs, Registered Replication Reports) in psychology is gradually increasing, with few uniform standards in data preparation or provision. This leads to challenges in both access and re-use of multi-lab replication data. The MetaPipeX framework takes on these challenges, serving as a novel proposal to standardize data structure, analysis code and reporting for experimental data of between groups comparisons in replication projects. It provides users with both structure and tools to synthesize and analyse mulit-lab replication data, select relevant subsets or create helpful graphics such as violin-, forest- and funnel-plots. MetaPipeX consists of three components: A descriptive pipeline for data transformations and analyses, analysis functions that implement the pipeline and a Shiny App utilizing the standardized structure for insights into the data at different aggregation levels. The analysis functions are largely built around meta-analysis of effect sizes (components), utilizing the metafor::rma.mv function (Viechtbauer, 2010). The analysis results consist of replication statistics, meta-analytical model- and heterogeneity-estimates. Additionally the functions provide documented data exports of various agreggation levels. All kinds of data subsets or graphics may be exported for further use. In this tutorial at ESMARConf we will present the framework and show personas ("prototypical users") with different use cases ranging from data analytical tasks to educational purposes. In order to contextualize the framework and its features we will provide a brief summary of the current state of repositories from multi-lab replication projects and discuss potential benefits and limitations of standardization. Using the MetaPipeX framework, we aim to save other researcher countless hours of data manipulation and harmonization, building a foundation for future reproducible multi-lab replication studies. GitHub repository: https://github.com/JensFuenderich/Met...