×

Effect of measurement errors on the efficiency assessment of a treatment. (Effet des erreurs de mesure sur l’évaluation de l’efficacité d’un traitement.) (French) Zbl 0793.62068

Summary: The effect of a treatment \(T\) on a skill can be conveyed by the partial regression coefficient \(\beta_{YT}\) with controlling the initial skill \(X\). Where \(Y\) and \(X\) are measured with errors by \(y\) and \(\vec x\), this effect is revealed by the coefficient \(\beta_{yT}\) of the regression of \(y\) onto \(T\) and \(\vec x\). Bias between \(\beta_{YT}\) and \(\beta_{yT}\) is made explicit, discussed and shown to be one-sided if \(X\) and \(T\) correlate. Importance of good specification of the model, for which a necessary and sufficient identification condition is given, is emphasized.

MSC:

62P15 Applications of statistics to psychology
62H20 Measures of association (correlation, canonical correlation, etc.)
62J05 Linear regression; mixed models
PDFBibTeX XMLCite
Full Text: Numdam EuDML