Scholarlink

This study examines the relationship between extreme response style (ERS) and rapid responding (RR) behavior in non-cognitive measures. Both phenomena can decrease the comparability, validity, and reliability of large-scale assessments (LSA), and different methods have been established and investigated to reduce the bias introduced by them. For example, RR can be identified using item-level response time thresholds, and ERS can be identified using binary pseudo items modeled in IRTree models. Previous studies could show that both response styles (acquiescence and ERS) and RR behavior can be successfully related to shorter response times and lower response time effort (RTE), respectively. Therefore, the question arises whether these two phenomena are similar types of behavior that potentially could be identified more effectively with a joint modeling approach, or whether they are behaviors with a different quality that must be treated separately. Analyzing the PISA 2015 data from Switzerland, we differentiate between groups based on their RR behavior shown as RTE and test them for ERS, and vice versa. We examine how the person variables for ERS and RTE are related and if the two phenomena can be empirically identified simultaneously. Results are compared regarding the validity of the interpretation of ERS and RTE.

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

Extreme Response Style and Rapid Responding: Two Sides of the Same Coin? A Joint Modeling Approach Using IRTree and Mixture IRT Models

  • 章节

  • 发表日期：

    2025年

查看全文
  • Ulf Kroehne,
  • Lale Khorramdel,
  • Frank Goldhammer,
  • Matthias von Davier

摘要

This study examines the relationship between extreme response style (ERS) and rapid responding (RR) behavior in non-cognitive measures. Both phenomena can decrease the comparability, validity, and reliability of large-scale assessments (LSA), and different methods have been established and investigated to reduce the bias introduced by them. For example, RR can be identified using item-level response time thresholds, and ERS can be identified using binary pseudo items modeled in IRTree models. Previous studies could show that both response styles (acquiescence and ERS) and RR behavior can be successfully related to shorter response times and lower response time effort (RTE), respectively. Therefore, the question arises whether these two phenomena are similar types of behavior that potentially could be identified more effectively with a joint modeling approach, or whether they are behaviors with a different quality that must be treated separately. Analyzing the PISA 2015 data from Switzerland, we differentiate between groups based on their RR behavior shown as RTE and test them for ERS, and vice versa. We examine how the person variables for ERS and RTE are related and if the two phenomena can be empirically identified simultaneously. Results are compared regarding the validity of the interpretation of ERS and RTE.