Latent Growth Modeling (LGM)

Latent growth modeling (LGM) is a structural equation modeling approach that estimates individual trajectories over repeated measurements as latent intercept and slope factors and their variation across people.

잠재성장모형(LGM)은 반복 측정 자료에서 개인의 변화 궤적을 잠재 절편과 기울기 요인으로 추정하고 그 개인 간 차이를 분석하는 구조방정식모형 접근이다.

How do I model individual change over time across several measurement waves?

여러 시점에 걸친 개인의 시간에 따른 변화를 어떻게 모형화하나요?

Latent growth modeling treats each person's starting level and rate of change as latent factors estimated from three or more repeated measurements. It fits longitudinal data where you want the average trajectory, individual differences in change and predictors of those differences.

When to use Latent Growth Modeling (LGM)

• The same variable measured on the same people across three or more waves
• Estimating average change and between-person variability in change
• Testing whether covariates predict initial levels or growth rates

When to use something else

• Interest is in reciprocal influences between two variables over time: use a cross-lagged panel model
• Unequally spaced or individually varying measurement times are central: use a multilevel growth model

Assumptions and requirements

• At least three measurement occasions for a linear trajectory
• Correctly specified functional form of change (linear, quadratic, piecewise)
• Multivariate normality or a robust estimator
• Missing data at random when using full information maximum likelihood

What the analysis reports

• Means and variances of intercept and slope factors
• Intercept–slope covariance
• Fit indices (CFI, TLI, RMSEA, SRMR)
• Effects of time-invariant covariates
• Individual trajectory plots

How to report it

The linear growth model fit the data well, χ²([df]) = [value], CFI = [value], RMSEA = [value]; the mean slope was [value], p = [value], with significant variance in slopes, [value], p = [value].

Related methods

• Hierarchical Linear Modeling (HLM)
• Cross Lagged Panel Model (CLPM)
• Growth Mixture Modeling

Run Latent Growth Modeling (LGM) on Quorant