Latent Transition Analysis (LTA)

Latent transition analysis is a longitudinal extension of latent class analysis that identifies unobserved classes at each time point and estimates the probabilities of individuals moving between classes over time.

잠재전이분석은 잠재계층분석을 종단으로 확장한 방법으로, 각 시점의 관찰되지 않은 계층을 찾고 개인이 시간에 따라 계층 사이를 이동할 확률을 추정합니다.

How do I study how people move between unobserved subgroups over time?

사람들이 시간에 따라 관찰되지 않은 하위집단 사이를 어떻게 이동하는지 분석하려면 어떻게 하나요?

Latent transition analysis identifies latent classes from indicators measured at two or more time points and estimates a transition probability matrix showing how membership changes between occasions. It fits longitudinal panel data when the research question is about stage-like change, such as shifts between behavior profiles.

When to use Latent Transition Analysis (LTA)

• The same set of indicators measured on the same people at two or more waves
• Questions about stability and change in subgroup membership
• Modeling covariates that predict class membership or transitions

When to use something else

• Only one time point: use latent class analysis or latent profile analysis
• Interest in continuous growth trajectories: use a growth mixture model

Assumptions and requirements

• Same indicators measured at each time point
• Local independence of indicators within classes
• Measurement invariance across time, often imposed to keep class meaning stable
• Large sample size for stable estimation

What the analysis reports

• Class prevalences at each time point
• Item-response probabilities or class means
• Transition probability matrix
• AIC, BIC, and entropy for model selection

How to report it

A [k]-class latent transition model was selected (BIC = [value], entropy = [value]), and individuals in [class A] at Time 1 had a transition probability of [value] to [class B] at Time 2.

Related methods

• Latent Class Analysis
• Latent Profile Analysis
• Growth Mixture Model

Run Latent Transition Analysis (LTA) on Quorant