Subsequently a series of multivariate random coefficient models (MRM) are conducted to examine the between-person correlations among craving, withdrawal/NA and PA growth parameters (i.e. intercept and instantaneous rate of change) at particular time points during the nicotine deprivation. These models tested, for example, whether level of withdrawal or affect at a given time point was significantly correlated with level of craving at the same time (i.e., correlation of intercept terms) and whether the instantaneous linear rate of change in the various domains were correlated at particular time points (i.e., correlations of linear slope terms). Thus these analyses addressed the question of whether subjects’ rate of change of withdrawal at a particular point in time was correlated with rate of change for craving. This sort of analytical approach is more commonly applied to long-term longitudinal data; however given that this particular EMA dataset has a very clear time origin (i.e., finishing the last cigarette) and a focus on how individual experience changes over time relative to that point, such an approach is well-suited to address the questions of interest.
