Lange Rael T, Schoenberg Mike R, Woodward Todd S, Brickell Tracey A
Riverview Hospital, Coquitlam, British Columbia, Canada.
Arch Clin Neuropsychol. 2005 Dec;20(8):1009-24. doi: 10.1016/j.acn.2005.06.004. Epub 2005 Aug 8.
This study developed regression algorithms for estimating IQ scores using the Canadian WAIS-III norms. Participants were the Canadian WAIS-III standardization sample (n = 1,105). The sample was randomly divided into two groups (Development and Validation groups). The Development group was used to generate 12 regression algorithms for FSIQ and three algorithms each for VIQ and PIQ. Algorithms combined demographic variables with WAIS-III subtest raw scores. The algorithms accounted for 48-78% of the variance in FSIQ, 70-71% in VIQ, and 45-55% in PIQ. In the Validation group, the majority of the sample had predicted IQs that fell within a 95% CI band (FSIQ=92-94%; VIQ=93-95%; PIQ=94-94%). These algorithms yielded reasonably accurate estimates of FSIQ, VIQ, and PIQ in this healthy adult population. It is anticipated that these algorithms will be useful as a means for estimating premorbid IQ scores in a clinical population. However, prior to clinical use, these algorithms must be validated for this purpose.
本研究利用加拿大韦氏成人智力量表第三版（WAIS - III）常模开发了用于估计智商分数的回归算法。参与者为加拿大WAIS - III标准化样本（n = 1,105）。该样本被随机分为两组（开发组和验证组）。开发组用于生成12种全量表智商（FSIQ）回归算法以及各3种言语智商（VIQ）和操作智商（PIQ）回归算法。这些算法将人口统计学变量与WAIS - III分测验原始分数相结合。这些算法解释了FSIQ中48 - 78%的方差、VIQ中70 - 71%的方差以及PIQ中45 - 55%的方差。在验证组中，大多数样本的预测智商落在95%置信区间范围内（FSIQ = 92 - 94%；VIQ = 93 - 95%；PIQ = 94 - 94%）。这些算法在该健康成年人群中对FSIQ、VIQ和PIQ产生了较为准确的估计。预计这些算法将作为一种估计临床人群病前智商分数的方法而有用。然而，在临床使用之前，必须针对此目的对这些算法进行验证。
