Enhancing the Use of Attitudinal Marker Variables in Travel Behavior Models: Evaluation of Latent Class Modeling Approaches Using a Nationwide Travel Survey

Term Start:

July 1, 2025

Term End:

May 31, 2026

Budget:

$100,773

Keywords:

Attitudinal Marker Variables, Latent Class Modeling, Travel Behavior Models

Thrust Area(s):

Data Modeling and Analytic Tools, Understanding User Needs

University Lead:

Georgia Institute of Technology

Researcher(s):

Patricia Mokhtarian

Recent studies have shown that an abbreviated set of attitudinal marker variables (MVs) can serve as proxies for attitudinal factor scores from the full set of attitudinal variables, either directly or through machine learning-based imputation. Compared to models without attitudes, these MVs enhance model fit, identify additional significant explanatory variables, and improve prediction of less-often chosen alternatives. Building on these findings, this study examines the potential improvement that using latent class (LC) modeling could bring to travel behavior (TB) models with attitudinal MVs incorporated. Using data from the 2024 Transportation Heartbeat of America (THA) Survey (N ≈ 8,200), the project compares four models for each of several TB variables: non-LC models with and without MVs, and LC models with MVs in either the membership or outcome component. The results will provide empirical evidence on (1) which MVs are most useful for specific TB variables, (2) how much incremental value MVs offer in the non-LC and LC models, and (3) how to specify LC models to best leverage attitudinal information. Overall, this study is expected to guide travel demand modeling practitioners and researchers in evaluating the benefits of including attitudinal variables in travel surveys and help them make well-informed decisions regarding measurement strategies and model development. These anticipated contributions are both essential and timely given that a few recent government-funded household travel surveys have started including attitudinal MVs.

Scroll to Top