Discriminating Between Attribute, Item-Position, and Wording Effects by the Congeneric and Tau-Equivalent...

Area of Science:

• Psychometrics
• Statistical Modeling

Background:

• Confirmatory factor analysis (CFA) is widely used to assess measurement models.
• Method effects, such as item position and wording, can influence observed variables.
• Understanding how CFA models handle these effects is crucial for valid interpretation.

Purpose of the Study:

• To investigate the capability of different CFA models to discriminate common systematic variation from attribute variation.
• To evaluate the impact of item-position and wording effects on model fit.
• To compare the performance of congeneric, tau-equivalent, and two-factor tau models in detecting method effects.

Main Methods:

• Simulated data were generated with constant attribute variation and increasing amounts of item-position or wording effects.
• Congeneric, tau-equivalent, and a two-factor tau model were applied to the simulated data.
• Model fit indices were used to assess the discrimination of method effects.

Main Results:

• The congeneric model showed no discrimination, consistently indicating good fit regardless of method effects.
• The tau-equivalent model demonstrated negative discrimination, shifting from good to bad fit as method effects increased.
• The two-factor tau model exhibited positive discrimination, successfully identifying method effects.

Conclusions:

• Different CFA models possess varying sensitivities to method effects.
• The tau-equivalent model may mask or misrepresent the impact of method effects.
• Pre-screening data for potential method effects is essential for robust psychometric analysis.