Latent Markov Latent Trait Analysis for Exploring Measurement Model Changes in Intensive Longitudinal Data.

Area of Science:

• Psychological science
• Quantitative psychology
• Longitudinal data analysis

Background:

• Drawing accurate inferences from intensive longitudinal data requires measurement model (MM) invariance.
• Measurement invariance can be violated due to individual differences in item interpretation or changing response styles over time.
• Existing methods like latent Markov factor analysis (LMFA) are limited to continuous data.

Purpose of the Study:

• To introduce latent Markov latent trait analysis (LMLTA) for evaluating measurement invariance with ordinal data.
• To address limitations of LMFA for non-normally distributed and ordinal response data.
• To demonstrate the application of LMLTA in analyzing psychological constructs.

Main Methods:

• Latent Markov latent trait analysis (LMLTA) was developed, building upon LMFA.
• LMLTA treats responses as ordinal, accommodating data like Likert scales.
• The method classifies observations into latent states with distinct MMs, invariant within each state.

Main Results:

• LMLTA was applied to adolescent affective well-being data across different social contexts.
• Differences in MMs were identified across these social contexts.
• The study highlights the importance of assessing measurement invariance for valid psychological research.

Conclusions:

• LMLTA provides a valuable tool for researchers working with ordinal intensive longitudinal data.
• Understanding measurement invariance is critical for accurate psychological science and practice.
• The method enhances the ability to study the dynamics of psychological constructs.