A Systematic Evaluation of Wording Effects Modeling Under the Exploratory Structural Equation Modeling Framework.

Area of Science:

• Psychometrics
• Behavioral Science
• Health Science

Background:

• Wording effects, a systematic variance from inconsistent responses to positively and negatively worded items, are common in behavioral and health research.
• Existing methods to address wording effects have limited evaluation within exploratory structural equation models (ESEM).

Purpose of the Study:

• To assess the performance of different response bias modeling strategies against wording effects in ESEM.
• To compare the effectiveness of the correlated traits-correlated methods minus one (CTC[M-1]) model, random intercept item factor analysis (RIIFA), and a no-correction approach.

Main Methods:

• Monte Carlo simulations were used to manipulate response bias types and magnitudes, factor loadings, factor correlations, and sample size.
• The study examined the impact of these manipulations on ESEM models using CTC[M-1], RIIFA, and a baseline approach.

Main Results:

• Ignoring wording effects significantly degrades ESEM model fit and distorts estimates.
• RIIFA demonstrated superior performance in mitigating bias and recovering accurate factor structures.
• CTC[M-1] models showed limitations when biases influenced both positively and negatively worded items.

Conclusions:

• Wording effects necessitate careful handling in ESEM to ensure valid results.
• RIIFA is recommended for addressing wording effects in ESEM, offering a robust solution.
• Researchers should be cautious about method factor interpretation, as they can absorb substantive variance.