Data_Sheet_3_Speed Effect Analysis Using the CFA Framework.ZIP

The paper outlines a method for investigating the speed effect due to a time limit in testing. It is assumed that the time limit enables latent processing speed to influence responses by causing omissions in the case of insufficient speed. Because of processing speed as additional latent source, the customary confirmatory factor model is enlarged by a second latent variable representing latent processing speed. For distinguishing this effect from other method effects, the factor loadings are fixed according to the cumulative normal distribution. With the second latent variable added, confirmatory factor analysis of reasoning data (N=518) including omissions because of a time limit yielded good model fit and discriminated the speed effect from other possible effects due to the item difficulty, the homogeneity of an item subset and the item positions. Because of the crucial role of the cumulative normal distribution for fixing the factor loadings a check of the normality assumption is also reported.本论文阐述了一种用于探究测试中时限诱发速度效应的研究方法。研究假设，当加工速度不足时，时限会通过引发作答遗漏，使得潜在加工速度对作答产生影响。由于加工速度作为额外的潜在来源，常规验证性因子模型（confirmatory factor model）被扩充为包含第二个代表潜在加工速度的潜变量。为将该速度效应与其他方法效应相区分，因子载荷需依据累积正态分布（cumulative normal distribution）进行固定。在加入第二个潜变量后，针对包含时限导致作答遗漏的推理任务数据（样本量N=518）开展验证性因子分析（confirmatory factor analysis），结果显示模型拟合效果优异，且能够将速度效应与其他可能由题目难度、题目子集同质性以及题目位置所引发的效应区分开来。鉴于累积正态分布在固定因子载荷过程中起到关键作用，本研究同时报告了针对正态性假设的检验结果。