Premium
A Comparison of Strategies for Smoothing Parameter Selection for Mixed‐Format Tests Under the Random Groups Design
Author(s) -
Liu Chunyan,
Kolen Michael J.
Publication year - 2018
Publication title -
journal of educational measurement
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.917
H-Index - 47
eISSN - 1745-3984
pISSN - 0022-0655
DOI - 10.1111/jedm.12192
Subject(s) - equating , akaike information criterion , log linear model , statistics , mathematics , selection (genetic algorithm) , smoothing , model selection , smoothing spline , statistic , polynomial , linear model , computer science , spline interpolation , artificial intelligence , mathematical analysis , rasch model , bilinear interpolation
Smoothing techniques are designed to improve the accuracy of equating functions. The main purpose of this study is to compare seven model selection strategies for choosing the smoothing parameter ( C ) for polynomial loglinear presmoothing and one procedure for model selection in cubic spline postsmoothing for mixed‐format pseudo tests under the random groups design. These model selection strategies were compared for four sample sizes (500, 1,000, 2,000, and 3,000) and two content areas (Advanced Placement [AP] Biology and AP Environmental Science). For polynomial loglinear presmoothing, the Akaike information criterion (AIC) was the only statistic that reduced both random equating error and total equating error in all investigated conditions. Cubic spline postsmoothing tended to produce more accurate results than any of the model selection strategies in polynomial loglinear smoothing.