"Somer’s d effect sizes“ in Odds Ratios umwandeln
Verfasst: Do 15. Sep 2022, 15:14Guten Tag,
kann mir netterwiese jemand sagen, wie man Somer’s d Effekt-Größen in Odds Ratios umwandeln kann? Bei dem folgenden Link kann man d-Werte in Odds Ratios umwandeln, mir ist aber nicht klar, ob dieser Rechner auch für Somer’s d Effekt-Größen gilt:
https://www.psychometrica.de/effektstaerke.html
Vielen Dank!
Die Effektgröße stammt aus diesem frei zugänglichen Artikel:
Screening for childhood adversity: the what and when of identifying individuals at risk for lifespan health disparities
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6031456/
Da steht:
„We then conducted v2 analyses to determine whether a positive indication on an adversity item was associated with each health outcome, and computed directional Somer’s d effect sizes to estimate the magnitude of each association; 95% confidence intervals were used to determine statistical reliability of these effects. Confidence intervals that do not encompass 0 are considered significant. Effect sizes \0.20 are considered very small, between 0.20 and 0.49 are considered small, between 0.50 and 0.79 are considered medium, and effect sizes C 0.80 are considered large (Cohen, 1988).“
kann mir netterwiese jemand sagen, wie man Somer’s d Effekt-Größen in Odds Ratios umwandeln kann? Bei dem folgenden Link kann man d-Werte in Odds Ratios umwandeln, mir ist aber nicht klar, ob dieser Rechner auch für Somer’s d Effekt-Größen gilt:
https://www.psychometrica.de/effektstaerke.html
Vielen Dank!
Die Effektgröße stammt aus diesem frei zugänglichen Artikel:
Screening for childhood adversity: the what and when of identifying individuals at risk for lifespan health disparities
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6031456/
Da steht:
„We then conducted v2 analyses to determine whether a positive indication on an adversity item was associated with each health outcome, and computed directional Somer’s d effect sizes to estimate the magnitude of each association; 95% confidence intervals were used to determine statistical reliability of these effects. Confidence intervals that do not encompass 0 are considered significant. Effect sizes \0.20 are considered very small, between 0.20 and 0.49 are considered small, between 0.50 and 0.79 are considered medium, and effect sizes C 0.80 are considered large (Cohen, 1988).“