Melbourne School of Psychological Sciences, University of Melbourne, Parkville, VIC, Australia.
Computing and Information Systems, University of Melbourne, Parkville, VIC, Australia.
J Clin Exp Neuropsychol. 2022 Feb;44(1):73-84. doi: 10.1080/13803395.2022.2080185. Epub 2022 Jun 4.
Raven's Advanced Progressive Matrices (APM) are frequently utilized in clinical and experimental settings to index intellectual capacity. As the APM is a relatively long assessment, abridged versions of the test have been proposed. The psychometric properties of an untimed 12-item APM have received some consideration in the literature, but validity explorations have been limited. Moreover, both reliability and validity of a timed 12-item APM have not previously been examined.
We considered the psychometric properties of untimed (Study 1; = 608; = 27.89, = 11.68) and timed (Study 2; = 479; = 20.93, = 3.12) versions of a brief online 12-item form of the APM.
Confirmatory factor analyses established both versions of the tests are unidimensional. Item response theory analyses revealed that, in each case, the 12 items are characterized by distinct differences in difficulty, discrimination, and guessing. Differential item functioning showed few male/female or native English/non-native English performance differences. Test-retest reliability was .65 (Study 1) to .69 (Study 2). Both tests had medium-to-large correlations with the Wechsler Abbreviated Scale of Intelligence (2nd ed.) Perceptual Reasoning Index ( = .50, Study 1; = .56, Study 2) and Full-Scale IQ ( = .34, Study 1; = .41, Study 2).
In sum, results suggest both untimed and timed online versions of the brief APM are psychometrically sound. As test duration was found to be highly variable for the untimed version, the timed form might be a more suitable choice when it is likely to form part of a longer battery of tests. Nonetheless, classical test and item response theory analyses, plus validity considerations, suggest the untimed version might be the superior abridged form.
Raven 的高级渐进矩阵（APM）常用于临床和实验环境中，以衡量智力水平。由于 APM 是一项相对较长的评估，因此已经提出了该测试的缩写版本。在文献中，已经对无时间限制的 12 项 APM 的心理测量特性进行了一些考虑，但有效性探索受到限制。此外，以前尚未检查过定时的 12 项 APM 的信度和效度。
我们考虑了无时间限制（研究 1； = 608； = 27.89， = 11.68）和定时（研究 2； = 479； = 20.93， = 3.12）两种简短的在线 12 项 APM 形式的心理测量特性。
验证性因素分析确定了这两种版本的测试都是单维的。项目反应理论分析表明，在每种情况下，12 项测试的难度，区分度和猜测度都有明显的差异。差异项目功能显示出男性/女性或英语母语者/非英语母语者之间的很少表现差异。测试-重测信度为.65（研究 1）至.69（研究 2）。这两种测试与韦氏简化智力量表（第 2 版）的知觉推理指数（ =.50，研究 1； =.56，研究 2）和全量表智商（ =.34，研究 1； =.41，研究 2）具有中等至较大的相关性。
总之，结果表明简短的 APM 的无时间限制和定时在线版本在心理测量上都是合理的。由于无时间限制版本的测试持续时间变化很大，因此当它可能成为更长的测试组合的一部分时，定时形式可能是更合适的选择。尽管如此，经典测试和项目反应理论分析以及有效性考虑表明，无时间限制的版本可能是更好的缩写形式。
