Educational Growth Curve Modeling
Also known as: Latent Growth Curve Modeling in Education, Multilevel Growth Models for Achievement, Individual Growth Trajectory Analysis, Learning Trajectory Modeling
Educational growth curve modeling is a longitudinal multilevel technique for describing and explaining how individual students change over time on an outcome such as reading or mathematics achievement. Building on the hierarchical linear models framework formalized by Bryk and Raudenbush (1987) and the applied longitudinal treatment of Singer and Willett (2003), it fits each student a personal trajectory — an intercept and one or more slopes — and then models how those personal growth parameters vary across students and relate to learner characteristics, classrooms, and schools.
Key highlights
- Respects the nesting of repeated measures within students, yielding correct standard errors where repeated-measures ANOVA or pooled regression would understate them.
- Estimates both average growth and the variance of growth, directly addressing whether students are diverging or converging over time.
- Handles unbalanced and unequally spaced data and partially missing waves under a missing-at-random assumption without listwise deletion.
- Links each student's trajectory to learner-, classroom-, and school-level predictors, supporting policy-relevant questions about who grows and why.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use educational growth curve modeling when you have repeated measures of an outcome on the same students over time — ideally three or more waves — and the research question concerns the shape of change, individual differences in change, or predictors of change. It is the standard tool for evaluating whether an intervention accelerates learning, for studying achievement gaps that widen or narrow across grades, and for separating true growth from measurement occasion noise. It is less appropriate with only two waves (which cannot distinguish a curve from a line and confound true change with regression to the mean), when the outcome is better treated as discrete-time survival, or when growth is strongly nonlinear in a way a low-order polynomial cannot capture.
Strengths & limitations
- Respects the nesting of repeated measures within students, yielding correct standard errors where repeated-measures ANOVA or pooled regression would understate them.
- Estimates both average growth and the variance of growth, directly addressing whether students are diverging or converging over time.
- Handles unbalanced and unequally spaced data and partially missing waves under a missing-at-random assumption without listwise deletion.
- Links each student's trajectory to learner-, classroom-, and school-level predictors, supporting policy-relevant questions about who grows and why.
- Requires at least three measurement occasions to separate the functional form of growth from within-person error; two waves are insufficient.
- Inference relies on the normality of random effects and of residuals, and on a correctly specified functional form for time (linear, quadratic, spline).
- Time-structured data with strong ceiling or floor effects on the test scale distort estimated slopes unless the metric is vertically equated across grades.
- Cannot by itself identify unobserved subpopulations with qualitatively different trajectories — that requires the growth mixture extension.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
How is educational growth curve modeling different from generic hierarchical linear modeling?
It is a specific application of the multilevel framework in which the lowest level is the repeated measurement occasion nested within the student, and TIME is the central predictor. Generic HLM nests students in classrooms or schools; growth curve modeling nests occasions in students and centers the substantive question on rate of change. See the related Educational Hierarchical Linear Modeling entry for the cross-sectional, students-in-schools formulation.
Why are at least three waves needed?
With two waves you can compute only a single change score per student, which cannot separate the systematic growth trajectory from occasion-specific measurement error, and the design cannot reveal whether growth is linear or curved. Three or more waves let the model estimate a within-student error variance distinct from the variance of true growth, which is what makes the trajectory interpretable.
Is growth curve modeling the same as a latent growth curve in SEM?
They are statistically equivalent formulations of the same model. The multilevel form treats time as a Level-1 predictor; the structural-equation latent-growth form treats the intercept and slope as latent factors with fixed loadings on the repeated measures. With balanced data they yield identical estimates; the multilevel form handles unequal spacing more naturally, while the SEM form integrates measurement models and multiple outcomes more easily.
Sources
- 1.Singer, J. D., & Willett, J. B. (2003). Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. Oxford University Press.ISBN 9780195152968
- 2.Bryk, A. S., & Raudenbush, S. W. (1987). Application of hierarchical linear models to assessing change. Psychological Bulletin, 101(1), 147–158.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 22). Educational Growth Curve Modeling. ScholarGate. https://scholargate.app/education/growth-curve-education