Use of the jackknife statistic to evaluate result replicability.
The writer points out that result replicability has often been confused with significance testing as a result of a misinformed view that statistical significance evaluates result importance and result replicability. She states that the jackknife statistic is one of several alternatives that provide...
| Publicado en: | Journal of General Psychology Vol. 125; no. 3; pp. 218 - 229 |
|---|---|
| Autor principal: | |
| Formato: | Artículo |
| Publicado: |
Taylor & Francis Ltd
July 1998
|
| Materias: | |
| Acceso en línea: | Ver este registro en EBSCOhost |
| Sumario: | The writer points out that result replicability has often been confused with significance testing as a result of a misinformed view that statistical significance evaluates result importance and result replicability. She states that the jackknife statistic is one of several alternatives that provide researchers with tools for estimating result replicability. She points out that the jackknife procedure entails repetitively dropping cases or subsets of cases from an analysis in order to determine how stable results are across various configurations of cases. She utilizes an available data set from A.L. Edwards (1985) to demonstrate how this jackknife statistic can be used to evaluate the replicability of the results of multiple regression in a concrete fashion. She computes the jackknifed coefficients to evaluate the stability of beta weights and the multiple R2 value. In addition, she calculates confidence intervals and t statistics in order to facilitate the interpretation of these jackknifed coefficients. |
|---|