Beyond p-Values: Using Bayesian Data Analysis in Science Education Research

Autor/inn/en	Kubsch, Marcus; Stamer, Insa; Steiner, Mara; Neumann, Knut; Parchmann, Ilka
Titel	Beyond p-Values: Using Bayesian Data Analysis in Science Education Research
Quelle	In: Practical Assessment, Research & Evaluation, 26 (2021), Artikel 4 (20 Seiten)Infoseite zur Zeitschrift PDF als Volltext kostenfreie Datei Verfügbarkeit
Sprache	englisch
Dokumenttyp	gedruckt; online; Zeitschriftenaufsatz
ISSN	1531-7714
Schlagwörter	Data Analysis; Bayesian Statistics; Educational Research; Science Education; Data Interpretation; Replication (Evaluation); Hypothesis Testing; High School Students; Graduate Students; College Faculty; Foreign Countries; Scientific Attitudes; Scientists; Germany + Suchen Sie Ihr Suchwort? Auswertung; Bildungsforschung; Pädagogische Forschung; Naturwissenschaftliche Bildung; Data evaluation; Datenauswertung; Hypothesenprüfung; Hypothesentest; High school; High schools; Student; Students; Oberschule; Schüler; Schülerin; Studentin; Graduate Study; Aufbaustudium; Graduiertenstudium; Hauptstudium; Fakultät; Ausland; Scientist; Wissenschaftler; Deutschland
Abstract	In light of the replication crisis in psychology, null-hypothesis significance testing (NHST) and "p"-values have been heavily criticized and various alternatives have been proposed, ranging from slight modifications of the current paradigm to banning "p"-values from journals. Since the physics education research community often relies on quantitative statistical approaches, the challenges the replication crisis poses to these approaches need to be considered. "p"-values suffer primarily from the fact that they carry little information by themselves and lend themselves to misinterpretations. As one alternative, Bayesian approaches have become increasingly popular as the posterior distributions they provide carry more relevant information than "p"-values. In this paper, we discuss practical issues related to "p-values" with respect to interpreting and communicating results and how these issues can be addressed using a Bayesian approach. Drawing on a science education data set, we demonstrate how Bayesian data analysis methods go beyond p-values and can help to make more valid conclusions and to communicate them more easily in a manner that lends itself to less misinterpretations. (As Provided).
Anmerkungen	Center for Educational Assessment. 813 North Pleasant Street, Amherst, MA 01002. e-mail: pare@umass.edu; Tel: 413-577-2180; Web site: https://scholarworks.umass.edu/pare
Erfasst von	ERIC (Education Resources Information Center), Washington, DC
Update	2024/1/01