Descriptive Statistics From Published Research: A Readily Available Alternative to Raw Data to Assess Analytic Reproducibility and Robustness. Issue 4 (November 2019)

• Record Type: Journal Article
• Authors: Nimon, Kim; Conley, David; Bontrager, Marvin; Keiffer, Greggory L.; Hammack-Brown, Bryn
• Other Names: Turner John R. guest-editor.; Baker Rose guest-editor.; Passmore David L. guest-editor.
• Is Part Of: Advances in developing human resources. Volume 21:Issue 4(2019)
• Page Start: 421
• Page End: 437
• Publication Date: 2019-11
• Subjects: descriptive statistics -- meta-science -- analytic reproducibility -- analytic robustness -- general linear model; Personnel management -- Periodicals; Organizational effectiveness -- Periodicals; Human capital -- Periodicals; 658.30105
• DOI: 10.1177/1523422319869853
• Languages: English
• ISSNs: 1523-4223

Abstract

The Problem: Meta-science literature calls for data to be made openly available so that scholars and scholar-practitioners can validate published findings, a foundational step in the reproducibility spectrum. However, access to original research data is an ongoing dilemma in various disciplines, including human resource development. The Solution: Scholars and scholar-practitioners have the opportunity to evaluate the credibility of previous studies without access to the original raw data. The use of descriptive statistics from published research offers an alternative to assess the reproducibility and robustness of selected prior research. The Stakeholders: In addition to validating research before applying implications for practice in the field, practitioners could benefit from working with scholars and scholar-practitioners by assessing analytic robustness and reevaluating data through a new framework to address burgeoning organizational problems, potentially saving resources. Scholars can reimagine conceptual frameworks based on advances to theory and statistical analyses capabilities. For emerging scholars, the ability to validate prior research or apply new models using the information contained in a publication can create a learning opportunity to understand statistical analyses.