Review
Economics
Sung Hoo Kim, Patricia L. Mokhtarian
Summary: This study examines the usage landscape and insights of finite mixture modeling in transportation research. Through a comprehensive review and database search, the study identifies the importance of accounting for heterogeneity and provides directions for future research.
TRANSPORTATION RESEARCH PART B-METHODOLOGICAL
(2023)
Article
Economics
Zhengying Liu, Astrid Kemperman, Harry Timmermans, Dongfeng Yang
Summary: This study investigates the physical activity patterns of older adults in Dalian, China, identifying two main types: leisure-oriented and both transport-related and leisure-oriented. Factors such as age, physical limitations, and having grandchildren in the household were found to significantly influence the physical activity patterns of older adults. Environmental characteristics, except for accessibility of local shops, condition of footpaths, and safety from crime, did not show significant relationships with physical activity patterns. The findings have implications for tailored interventions to maintain or promote physical activity among older adults.
JOURNAL OF TRANSPORT GEOGRAPHY
(2021)
Article
Psychology, Clinical
Kate Walsh, Anuj H. P. Mehta, Zach Buehler, Melanie Wall, John Santelli, Jennifer S. Hirsch, Claude A. Mellins
Summary: This study used latent class analysis to identify four distinct drinking typologies among college students based on quantity, frequency, and social/environmental contexts. These typologies differed in demographics, age, campus activities, protective behaviors, and alcohol problems.
ADDICTIVE BEHAVIORS
(2023)
Article
Public, Environmental & Occupational Health
Isabella Toledo Caetano, Valter Paulo Neves Miranda, Fernanda Karina dos Santos, Paulo Roberto dos Santos Amorim
Summary: This study used latent class analysis to identify three different perceived environmental classes and found associations between light physical activity, total sitting time, and socioeconomic status with the prevalence of these classes. The results indicate a significant influence of the neighborhood environment on adolescents' light physical activity and total sitting time.
Article
Health Care Sciences & Services
Daniel McNeish, Jeffrey R. Harring
Summary: Growth mixture models are a popular method for uncovering heterogeneity in growth trajectories, but applying them to empirical data often encounters issues such as nonconvergence. While researchers may resort to adding constraints to the random effect covariance structure for estimation, this approach can negatively impact parameter estimation and class assignment. Therefore, advocating for specifying models using a marginal approach to prevent sacrificing class-specific covariance structures is crucial in handling these challenges.
STATISTICAL METHODS IN MEDICAL RESEARCH
(2021)
Article
Statistics & Probability
Caleb Kwon, Eric Mbakop
Summary: In this study, a novel estimator for the number of mixture components in a nonparametric finite mixture model is proposed. The estimator is based on the rank of an integral operator identified from the data and the stability of singular values under perturbations. It is shown to be consistent and has finite sample performance guarantees. Monte Carlo simulations demonstrate good performance for moderate sample sizes.
ANNALS OF STATISTICS
(2021)
Review
Public, Environmental & Occupational Health
Sung Hoo Kim
Summary: This study examines how heterogeneity is considered in transportation safety analyses, focusing on the use of latent class modeling. The study identifies relevant papers in the field and analyzes the key elements of the latent class model. Various class-specific outcome models have been used in the literature, depending on the type of outcome variable and the analysis context. The number of classes selected in empirical applications is often based on the analyst's judgement rather than quantitative measures.
ANALYTIC METHODS IN ACCIDENT RESEARCH
(2023)
Article
Health Care Sciences & Services
Min Zhang, Chong Wang, Annette M. O'Connor
Summary: This research developed a method to quantify the correlation in antimicrobial resistance across populations and detected an unnoticed trend of susceptible bacteria. The application of this method revealed positive correlation in antibiotic resistance of susceptible bacteria across different populations, providing insights for potential clinical interventions.
BMC MEDICAL RESEARCH METHODOLOGY
(2021)
Article
Public, Environmental & Occupational Health
Kimberly B. Roth, Dana K. Goplerud, Leslie B. Adams, Mikalah E. Maury, Rashelle J. Musci
Summary: This study used latent class analysis to examine the impact of different neighborhood types on residents' mental and physical health in Maryland. The results showed that suburban resourced neighborhoods had the lowest prevalence of poor health, while urban underserved neighborhoods had the highest prevalence. The study highlights the complexity of defining healthy neighborhoods and the need to focus on improving community health.
Article
Psychology, Multidisciplinary
Daniel McNeish, Jeffrey R. Harring, Daniel J. Bauer
Summary: Clustering growth trajectories into classes can simplify interpretation of longitudinal data, but estimation is complicated and convergence issues are common. The typical approach of constraining covariance across classes performs poorly, while the covariance pattern GMM approach has the highest convergence rates and performs best in identifying the correct number of classes and assigning people correctly.
PSYCHOLOGICAL METHODS
(2023)
Article
Social Sciences, Interdisciplinary
Seuk Yen Phoong, Shi Ling Khek, Seuk Wai Phoong
Summary: This study reviews the literature on finite mixture models through bibliometric analysis, revealing an increasing trend in annual publications and identifying key journals and highly cited articles. Network analysis is also conducted to explore scientific cooperation in the field of finite mixture model studies. The study proposes a research agenda based on the current state and population trends of finite mixture models.
Review
Criminology & Penology
Valdemir Ferreira, Juliana Y. Valente, Zila M. Sanchez
Summary: This study utilized latent class analysis to identify patterns of bullying victimization and perpetration, finding associations with alcohol use and poor school performance. Fifth-grade black/brown students were more likely to be in the high bullying victimization class. Race and gender were found to influence students' experiences of violence, highlighting the importance of considering these factors in prevention programs targeting bullying and alcohol use.
JOURNAL OF INTERPERSONAL VIOLENCE
(2022)
Article
Green & Sustainable Science & Technology
Yulin Long, Debin Zheng, Yuehua Wei, Zhenyu Cai, Yitao Ye, Biqi Mao, Changlin Ao
Summary: Public preferences and needs are crucial for air pollution treatment policies. Behavioral economics has discovered that risk perception affects the heterogeneous choices behavior of the public. Those with higher risk controllability tend to make rational decisions, while those more susceptible to risk prefer bounded rational decisions.
JOURNAL OF CLEANER PRODUCTION
(2023)
Article
Computer Science, Interdisciplinary Applications
Spyridon J. Hatjispyros, Christos Merkatas, Stephen G. Walker
Summary: Decreasing weight prior distributions are important for nonparametric Bayesian inference, offering an efficient alternative to traditional Dirichlet process mixture models and addressing the label switching problem through weight ordering. A general procedure is provided for defining models with decreasing weights based on a characterization of a discrete random variable, allowing for easy sampling and accurate estimation of the model. Numerical investigations on simulated data sets evaluate the performance of the mixture model.
COMPUTATIONAL STATISTICS & DATA ANALYSIS
(2023)
Review
Business
Marko Sarstedt, Lacramioara Radomir, Ovidiu Ioan Moisescu, Christian M. Ringle
Summary: With the increasing use of partial least squares structural equation modeling (PLS-SEM) in business research, latent class analyses for identifying and addressing unobserved heterogeneity have also gained attention. Finite mixture PLS (FIMIX-PLS) plays a central role in this area, but researchers need to make careful choices to avoid incorrect results.
JOURNAL OF BUSINESS RESEARCH
(2022)