Table 1.
Types of questions each specification of the LGCM can address.
| Question | Basic LGCM | LGMM | PLGCM | LGCM for 2 parallel processes |
|---|---|---|---|---|
| 1. What is the continuous rate of change in cortisol? | Provides information on the rate of cortisol change throughout the study. When nonlinear slopes are specified, gives insight into the rate of the nonlinearity present. | Provides information on how the rate of change in cortisol differs across unobserved groups. | Provides information on the rate of cortisol change during the baseline phase (slope 1) and the recovery phase (slope 2) | Provides information on the rate of change for two processes (i.e., cortisol and alpha-amylase), whether linear or nonlinear |
| 2. What does the change in cortisol look like over time | Can gain insight into the stress reaction and the recovery period through the specification of nonlinear growth factors. | Provides information on how the trend differs across unobserved groups (e.g., linear in one group and quadratic in another). | Can evaluate the change in cortisol over time for the baseline and recovery periods separately. | Can address in the same way as the basic LGCM, LGMM, and PLGCM depending on specification. Addresses these question for each process and provides insight into how they are related across processes. |
| 3. How do cortisol and alpha-amylase relate over time? | Can evaluate growth of cortisol and alpha-amylase through a multivariate LGCM or can control for the effect of alpha-amylase at each measurement of cortisol. | Provides information into how these relationships differ across unobserved groups. | Can evaluate growth of cortisol and alpha-amylase through a multivariate PLGCM or can control for the effect of alpha-amylase at each measurement of cortisol. | Evaluates how activation of each system is related through relationships specified between growth factors of each system. |
| 4. Are there (observed or unobserved) group differences in the rate of change in cortisol? | Can control for the effects of a grouping variable (i.e., time-invariant covariate) or can compare the trajectories and rates of changes of each group. | Can evaluate differences in the trajectories and rate of changes for unobserved groups (e.g., extreme responders vs. normal responders) | Can control for the effects of a grouping variable (e.g., gender) or can compare the trajectories and rates of changes of each group. | Can control for the effects of a grouping variable (e.g., gender) or can compare the trajectories and rates of changes of each group. |
| 5. Does the rate of change in cortisol predict health outcomes? | Can answer whether the rate of change in cortisol affects a health outcome (e.g., the number of medical office visits) | Provides insight into how cortisol predicts health outcomes may differ across unobserved groups | Can answer whether the rate of change in cortisol at baseline or recovery affects a health outcome (e.g., the number of medical office visits). | Same as the basic LGCM and the PLGCM, except can now include how the rate of change in alpha-amylase also affects health outcomes (e.g., the number of medical office visits). |
LGCM, Latent growth curve model; PLGCM, piecewise LGCM; LGMM, latent growth mixture model.