Scholarlink

<p>An attempt to replicate Kinder (Presidential traits: pilot study report to the 1984 NES planning committee and NES board. American National Election Studies, 1983 <a href="https://electionstudies.org/wp-content/uploads/2018/07/nes002244.pdf">https://electionstudies.org/wp-content/uploads/2018/07/nes002244.pdf</a>, In: Lau, R.R., Sears, D.O. (eds) Political cognition, Erlbaum, Hillsdale, 1986) original set of 30 candidate trait questions used in the to develop the subset used American National Election Study (ANES) since 1980 in the Czech Republic in July 2021 resulted in a ‘replication contradiction’. Using confirmatory factor analysis (CFA), replication was successful; however, with exploratory factor analysis (EFA) it was a failure. This paper explores the hypothesis that this replication contradiction arose because the factor analysis method has a limited ability to recover true response structures. Using a Monte Carlo simulation strategy, CFA, EFA, and bifactor models were evaluated on their abilities to recover an a priori 5 factor structure from 30 nominal indicators when four conditions were varied: factor loadings, factor correlations, sample size, and presence/absence of a general factor. The simulation results reveal that with higher factor loadings and sample sizes, EFA, CFA and bifactor models can recover the true data structure. Consequently, the replication contradiction has data origins. If there are strong item intercorrelations and factor loads, all methods give the same substantive answer; contradictions arise when data are insufficiently structured for EFA to identify a trait pattern.</p>

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

Use of simulation to explore a ‘replication contradiction’: the case of leader trait measurement

  • 期刊论文

  • 发表日期：

    2025年10月17日

查看全文
  • Pat Lyons

摘要

An attempt to replicate Kinder (Presidential traits: pilot study report to the 1984 NES planning committee and NES board. American National Election Studies, 1983 https://electionstudies.org/wp-content/uploads/2018/07/nes002244.pdf, In: Lau, R.R., Sears, D.O. (eds) Political cognition, Erlbaum, Hillsdale, 1986) original set of 30 candidate trait questions used in the to develop the subset used American National Election Study (ANES) since 1980 in the Czech Republic in July 2021 resulted in a ‘replication contradiction’. Using confirmatory factor analysis (CFA), replication was successful; however, with exploratory factor analysis (EFA) it was a failure. This paper explores the hypothesis that this replication contradiction arose because the factor analysis method has a limited ability to recover true response structures. Using a Monte Carlo simulation strategy, CFA, EFA, and bifactor models were evaluated on their abilities to recover an a priori 5 factor structure from 30 nominal indicators when four conditions were varied: factor loadings, factor correlations, sample size, and presence/absence of a general factor. The simulation results reveal that with higher factor loadings and sample sizes, EFA, CFA and bifactor models can recover the true data structure. Consequently, the replication contradiction has data origins. If there are strong item intercorrelations and factor loads, all methods give the same substantive answer; contradictions arise when data are insufficiently structured for EFA to identify a trait pattern.