Scholarlink

<p>The measurement efficiency of a grid multidimensional computerized classification test (grid MCCT), which makes a classification decision per dimension, can be improved by taking the correlations between the dimensions into account in the termination criterion. The higher the correlations, the better the improvement in measurement efficiency. However, a termination criterion utilizing inter-dimensional information (i.e., SPRT-C; Liu et al., <CitationRef CitationID="CR8">2022</CitationRef>) was found to yield lower levels of correct classification rates than not utilizing it (i.e., SPRT-SF; Seitz &amp; Frey, <CitationRef CitationID="CR14">2013</CitationRef>) under the between-item grid MCCT when the cutoff was set at the mean of the latent trait distribution. This study proposes a pre-rule to determine whether the SPRT-SF or SPRT-C should be used during the process of classification test administration. Through a series of simulation studies, the results showed that our proposed method (called P-SPRT) can substantially improve upon the SPRT-C in terms of correct classification rates, while maintaining its high measurement efficiency in terms of test length. This paper concludes with a discussion of the findings and further applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A pre-rule for the sequential probability ratio test in a between-item grid multidimensional computerized classification test

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年1月29日

查看全文
  • Po-Hsien Hu,
  • Ching-Lin Shih,
  • Cheng-Te Chen

摘要

The measurement efficiency of a grid multidimensional computerized classification test (grid MCCT), which makes a classification decision per dimension, can be improved by taking the correlations between the dimensions into account in the termination criterion. The higher the correlations, the better the improvement in measurement efficiency. However, a termination criterion utilizing inter-dimensional information (i.e., SPRT-C; Liu et al., 2022) was found to yield lower levels of correct classification rates than not utilizing it (i.e., SPRT-SF; Seitz & Frey, 2013) under the between-item grid MCCT when the cutoff was set at the mean of the latent trait distribution. This study proposes a pre-rule to determine whether the SPRT-SF or SPRT-C should be used during the process of classification test administration. Through a series of simulation studies, the results showed that our proposed method (called P-SPRT) can substantially improve upon the SPRT-C in terms of correct classification rates, while maintaining its high measurement efficiency in terms of test length. This paper concludes with a discussion of the findings and further applications.