混合模型
计算机科学
人工智能
鉴定(生物学)
统计模型
数据挖掘
组分(热力学)
计量经济学
机器学习
期限(时间)
作者
Youxiang Jiang,Chongqin Xi,Hongbo Wen
摘要
With the rise of computerized testing platforms, social science surveys increasingly use continuous rating scales (CRS), such as visual analog scales and sliders, to capture nuanced respondent data. However, careless and insufficient effort responding (C/IER), driven by low respondent motivation, introduces significant response bias, compromising measurement accuracy and the validity of statistical analyses. Traditional methods for detecting C/IER, developed for Likert-type responses, are incompatible with the continuous bounded data from CRS. This study develops the continuous bounded item response model for C/IER (CBIRM-C/IER), which integrates continuous bounded item response theory modeling with response-time-based mixture modeling to identify and model C/IER at the person-by-item level. Through two simulation studies, we evaluated the model's parameter recovery under various conditions, demonstrating consistent accuracy and robust performance with C/IER-contaminated data compared with conventional item response theory models. An empirical application evaluated the CBIRM-C/IER model's fit and parameter estimation against traditional models. By detecting and modeling C/IER at the person-by-item level, CBIRM-C/IER provides more reliable estimates of the trait of interest. In summary, CBIRM-C/IER can help practitioners identify and model C/IER within CRS data to obtain more accurate estimates of the trait of interest. (PsycInfo Database Record (c) 2026 APA, all rights reserved).
科研通智能强力驱动
Strongly Powered by AbleSci AI