問項反應理論 模型에 의한 「運動選手에 대한 態度」측정 = Partial Credit Analysis of Attitudes toward Athletes

다국어 초록 (Multilingual Abstract)

The Rasch Partial Credit model in calibrating a scale assessing the attitude toward athletes. has several advantages. First, the appropriateness of testing can be evaluated based upon their difficulty level, because task difficulties and examinee abilities share the same metric. Second, goodness-of-fit statistics generated from the Rasch modeling provide a quantitative index to determine whether or not testing items have met the requirements of the calibration. Third, the Rasch Partial Credit modeling provided a more appropriate interpretation of examinees' scores(Zhu, 1995). The purpose of this study was to apply the Rasch Partial Credit model for scaling the Attitudes toward Athletes Scale into a mealsure of attitudes.
A total of 421 students(196 males and 225 females), junior high to college in Seoul, were involved in this study. Attitudes toward Athletes Scale, consisting of 7 bipolar adjectives which form a unidimensionality of the scale, was developed on the basis of factor analysis. The selected 7 bipolar adjectives are beautiful-ugly, clean-dirty, wisefoolish, good-bad, like-dislike, happy-sad, and valuable-invaluable which represent 'evaluative' dimension.
The raw scores were analyzed by PC-CREDIT, a computer program for partial credit analysis(Masters & Wilson, 1988). The unconditional maximum likelihood procedure was used for parameter estimation. Because 7 levels were defined for each bipolar adjective items, 6 step-difficulty parameters were estimated for each item. The step difficulty was defined as the intersection between two adjacent levels. The model-data fit was evaluated by t statistics(Wright & Masters, 1982). A t-statistic value between -2 and +2 indicated an acceptable model-data fit ; a value between -2 and -3, or between +2 and +3, indicated a marginal model-data fit ; and a value either less than -3 or larger than +3 indicated a model-data misfit.
Step difficulties estimated for 7 bipolar adjective items ranged from -1.62 to 1.98. The range of difficulty increases dramatically because items have multi-category outcomes. The easiest(the most negative) step was the fourth step at item 5 "like-dislike"(??) and the hardest(the most positive) step was the seventh step at item 3 "wise-foolish"(??). Satisfactory model-data fit was found for item 1, 2, 4, 5, and 6 and marginal model-data fit found for item 7, but not for item 3(t=4.48). This indicated that the 7 bipolar adjective items except item 3(wise-foolish) were appropriately defined and were measuring a similar trait, i.e., attitude toward athletes. The average level of positive attitude toward athletes was .47, with a standard deviation of .58. The positive attitude
level toward athletes of males(??) was higher than that of females(??), and the level of middle and high school students were higher than those of college students.