Effects of respondent training on self-report personality assessment: an item response theory approach

Zhang, Luyao

Permalink

https://hdl.handle.net/2142/102913 Copy

Description

Title

Effects of respondent training on self-report personality assessment: an item response theory approach

Author(s)

Zhang, Luyao

Issue Date

2018-11-26

Director of Research (if dissertation) or Advisor (if thesis)

Drasgow, Fritz

Doctoral Committee Chair(s)

Drasgow, Fritz

Committee Member(s)

• Chang, Hua-Hua
• Fraley, R. Chris
• Newman, Daniel
• Roberts, Brent
• Rounds, James

Department of Study

Psychology

Discipline

Psychology

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Date of Ingest

2019-02-08T18:39:46Z

Keyword(s)

• Rater training
• item response theory (IRT)
• self-report personality assessment
• the ideal-point model
• intermediate items

Abstract

Within the item response theory (IRT) framework and inspired by the rater training literature, this study explored the effects of short online respondent training on personality item interpretation and responding and the number of response categories (i.e. polytomous vs. dichotomous) on item performance, model-data fit, and criterion-related validity. Participants recruited from MTurk (n = 1977) were randomly assigned to 1 of the 4 groups differing in training (i.e. training vs. no training) and response scale (i.e. 4-point Likert scale vs. dichotomous), and their responses to dominance and ideal-point personality measures were analyzed with GGUM, SGR, and 2PL. Results indicated that training was associated with more well-performing and more discriminating and informative intermediate items on the ideal-point scales when a dichotomous response scale was used. The dichotomous scale in general was related to better fit, while criterion-related validity stayed unaffected by both training and the response scale. Participants reported that they had been confused about personality items before, and were positive about the online training, which was consistent with the finding that trained participants on average spent 32 seconds less finishing the ideal-point surveys. Implications for future research and practice are discussed.

Graduation Semester

2018-12

Type of Resource

text

Permalink

http://hdl.handle.net/2142/102913

Copyright and License Information

Copyright 2018 Luyao Zhang

Owning Collections