Akademik Veri Yönetim Sistemi
TURK PSIKOLOJI DERGISI, cilt.19, sa.54, ss.77-88, 2004 (SSCI)
In this study two criteria and three techniques related with item selection procedures for construction of Likert Type scales were investigated. One of these criteria was item selection via fixed significance level of item statistics (p <= 0.01), and the other criterion was selection via fixed number of item (K = 20 items). Three techniques for item selection procedures were examined under these criteria. The first technique depends on calculation of correlation coefficients between item scores and scale scores. The second technique depends on calculation of correlation coefficients between item scores and scale scores obtained with the related item score excluded. The third technique depends on significance of difference between averages of upper and lower group item scores. Five separate attitude scales were prepared for investigative purposes. Results indicated that using fixed number of item criterion provided equivalent reliability with less amounth of items. Reliability coefficients of obtained scales via three item selection techniques were not significantly different.