This table shows the results of the analysis where HONOS scores for the social problem were summarised for clients by those caregivers who received or did not receive intervention (KPP intervention; the variable: Interventioni Group). In this case Intervention group was assigned a value of 1 and the control group was assigned a value of 0. The term wave indicated the impact of time. As these variables were correlated observations over time, hence a GEE was used to assess the relative impacts of time and intervention. We also tested whether the intervention varied over time.

As can be seen from the table, the intervention targeted at the caregivers in terms of noting the summative assessments for the clients had minimal impacts (beta = 0.324, p = 0.348). This indicated that on an average, for those who received the KPP intervention, on an average they registered 0.32 higher score for the social problem summary score for their clients over the entire 12 time points. However, time itself was a borderline statistically significant predictor and showed that with each two monthly increment in time (that is, with each subsequent time point or wave), on an average the social problem summmary score dropped by 0.03. While this figure itself is not substantively significant, when time points and interventions were considered together in an interactive model, the impact of the time persisted (beta coefficient = -0.032, p = 0.357) although this change itself was not statistically significant.

We further examined the pattern of variability of the time response (see Figure 1).