Effektstärke

Diese SVG-Grafik fensterfüllend anzeigen

Effektstärke, effect size

Die Effektgrösse (oder «Effektstärke») beschreibt die relative Grösse eines Unterschieds zwischen zwei Gruppenmittelwerten. Sie steht in Ergänzung zur Signifikanzangabe. Eine Effektgrösse von d = 0.2 verweist auf schwache Effekte, d = 0.5 auf mittlere und d = 0.8 auf starke Effekte (Cohen, 1988, S. 25ff.).

Von Claudia Zahner Rossier im Buch PISA 2003: Kompetenzen für die Zukunft I (2004) auf Seite 75

A statistic indicating how many standard deviations difference there is between the mean score of the experimental group and the mean score of the control group. A useful metric to determine the practical significance of research results. Effect sizes greater than .5 indicate an outcome of practical significance worthy of implementation.

Von Ruth Colvin Clark, Richard E. Mayer im Buch E-Learning and the Science of Instruction (2002) auf Seite 459

Effect sizes estimate the magnitude of effect or association between two or more variables (Ferguson, in press; Snyder & Lawson, 1993). As with all statistical tools, effect size estimates are just that, estimates. Mostly, effect sizes are resistant to sample size influence, and thus provide a truer measure of the magnitude of effect between variables.

Von Christopher J. Ferguson im Text An Effect Size Primer (2009)

It is meaningless to report an effect size without specifying the index or measure used. An effect of size = 0.5 will mean something quite different depending on whether it belongs to the d or r family of effects. (An r of 0.5 is about twice as large as a d of 0.5.)

Von Paul D. Ellis im Buch The Essential Guide to Effect Sizes (2010) im Text Introduction to effect sizes auf Seite 16

The estimation of an effect size is distinct from assessments of statistical significance. Although they are related, statistical significance is also affected by the size of the sample. The bigger the sample, the more likely an effect will be judged statistically significant.

Von Paul D. Ellis im Buch The Essential Guide to Effect Sizes (2010) im Text Interpreting effects auf Seite 32

We do need to be careful about ascribing adjectives such as small, medium, and large to these effect sizes. Cohen (1988), for example, suggested that d = 0.2 was small, d = 0.5 medium, and d = 0.8 large, whereas the results in this book could suggest d = 0.2 for small, d = 0.4 for medium, and d — 0.6 for large when judging educational outcomes.

Von John Hattie im Buch Visible Learning (2009) im Text The nature of evidence auf Seite 9

Zum Problem der Messung hinzu kommen fragwürdige statistische Praktiken, etwa bei der Ermittlung von Eff ektstärken (Simpson, 2017): Um eine besonders hohe Eff ektstärke zu erzielen, muss man eine möglichst homogene Gruppe untersuchen. Beim Berechnen der Eff ektstärke kann man einen Vorher- Nachher-Vergleich ziehen, aber auch einen Vergleich Intervention versus normalen Unterricht oder versus Nichtstun ziehen. Publikumswirksame Zahlenwerte liefert sicher die erste Art der Rechnung.

Von Jörn Loviscach im Buch Vom E-Learning zur Digitalisierung (2020) im Text Digitalisierung der Hochschullehre: Was wissen wir wirklich?

There are many examples that show small effects may be important. A vivid example comes from medicine. Rosenthal and DiMatteo (2001) demonstrated that the effect size of taking low dose aspirin in preventing a heart attack was d - 0.07, indicating that less than one-eighth of one percent of the variance in heart attacks was accounted for by using aspirin. Although the effect size is small, this translates into the conclusion that 34 out of every 1,000 people would be saved from a heart attack if they used low dose aspirin on a regular basis.This sounds worth it to me.

Von John Hattie im Buch Visible Learning (2009) im Text The nature of evidence auf Seite 9

Verwandte Begriffe
(co-word occurance)

Metaanalyse meta-analysis(0.25), Signifikanz(0.03)

John
Hattie

Paul D.
Ellis

John
Hattie