Modeling insufficient effort responses in mixed-worded scales.

Area of Science:

• Psychometrics
• Survey Methodology
• Statistical Modeling

Background:

• Surveys commonly use reverse-coded items to detect insufficient effort responses (IERs).
• A common assumption is that respondents consistently exert full effort, which is often incorrect.
• Ignoring IERs can lead to flawed survey data and inaccurate conclusions.

Purpose of the Study:

• To expand mixture models for identifying IERs.
• To demonstrate the detrimental effects of ignoring IERs on survey data quality.
• To illustrate the practical application of these methods using real-world datasets.

Main Methods:

• Expanded mixture models for insufficient effort responses (IERs).
• Simulation study using LatentGOLD software.
• Application to two public datasets: Machiavellianism and self-reported depression scales.

Main Results:

• Ignoring IERs negatively impacts test reliability.
• Failure to account for IERs introduces bias into survey data.
• Reduced accuracy in slope and intercept parameters was observed when IERs were ignored.

Conclusions:

• Expanded mixture models provide a robust approach to handling IERs.
• Ignoring IERs in survey data analysis leads to significant psychometric and statistical problems.
• Accurate analysis requires accounting for respondent effort, regardless of question wording.