TAN, Şeref Gazi Üniversitesi - Gazi Eğitim Fakültesi - Eğitimde Ölçme ve Değerlendirme Anabilim Dalı, Turkey

Accuracy of Beta4 Presmoothing Polynomial Loglineer Presmoothing and Cubic Spline Postsmoothing Methods for Small Samples

19247

In this study, a comparison of beta4 and polynomial loglinear presmoothing methods and the cubic spline postsmoothing method of equipercentile equating with balanced single group design data set for small samples are presented. When the beta4 presmoothing method was applied to Form X and Form Y data, Form X maintained for all the four sample data moments and Form Y maintained for the first 3 sample data moments. According to the chi square statistics beta4 presmoothing method showed good fit with sample data. In the polynomial loglinear presmoothing, presmoothing with second degree polynomial model provided the best data fit for both Form X and FormY. Cubic spline postsmoothing method has provided the most accurate fit for S = .30. As a conclusion, all of the three smoothing methods, used in the study, were effective and relatively beta4 presmoothing method was found the most appropriate method among the three smoothing methods by using both bootstrap standart errors of equating and moment preservation criterion for small samples such as 200-250.

123

Beta4 presmoothing , Polynomial loglineer presmoothing , Cubic postsmoothing , Equipercentile equating

Gazi University Journal Of Gazi Educational Faculty

151