Description
Writes the theory of change as a rule for each occasion-to-occasion change, solves it, and fits it to repeated measures to test competing change processes.
A latent change score model expresses each score as an initial level plus the sum of latent changes between occasions. It is equivalent to a standard growth model but makes the change process explicit, so each change can depend on the current level, on other variables or on external inputs. Ram and Grimm recommend starting from a differential equation that states the theory: constant change gives linear growth, change proportional to level gives exponential growth, change towards a carrying capacity gives a negative exponential (a learning curve with a ceiling), and time-varying rate or capacity can represent an intervention. The solved function is fitted with person-level random coefficients in SEM or multilevel software, and parameters are read as named process quantities such as rate, capacity and initial condition. Competing change rules are compared formally.
Attributes
- Discipline
- Statistics & Modelling
- Loss type
- Unmigrated
- Stage
- Evaluation, Post-launch
- Data
- Quantitative
- Effort
- Weeks
Steps
- State the hypothesised change rule as a differential equation (dy/dt) or difference equation, for example change proportional to the distance from a ceiling.
- Solve it over the observation window to obtain the expected trajectory function.
- Specify the model in SEM as latent changes between occasions, or as a nonlinear mixed model, adding person-level random coefficients.
- Fit competing change rules (constant, proportional, ceiling-seeking, coupled with another variable) and compare them with information criteria or likelihood-ratio tests.
- Interpret the winning model's parameters as process quantities and test whether an intervention changes rate or capacity.
Sources
- Handbook of Child Psychology Vol 1 (Lerner et al., 2015)
Also known as
- LCS model
- Latent difference score model
- Dual change score model
- Differential equation modelling
- Latent change score and differential-equation models of change
Categories
- Analytics & Behaviour
- Statistics & Modelling