140 Chapter 6 ends a challenging behavioral phase. For example, on day 86 for self-injury a 30% drop was detected by recursive partitioning and on day 91 aggression dropped by 28%. Here we marked day 91 as the transition, as it marked the end of a phase with frequent challenging behavior. Similarly, when self-injury and aggression increased on respectively day 446 and 452, we marked 446 as the transition for a start with frequent challenging behaviors. One exception was made, based on a clear difference in absolute change: on day 46 the proportion of aggressive incidents increased with 25%, while 11 days later the proportion in self-injurious incidents increased by 60%. Hence, only day 57 was marked as a transition. Two detected mean-changes were not marked as transitions: the increase of self-injury on day 122 and the decrease in aggressive incidents on day 257. The latter (day 257) was not marked as an attractor change, because of the large number of missing values that followed this transition (see gray band in Figure 4A). Day 122 was not marked after visual inspection of the self-injurious incidents timeseries (Figure 4A) we noted that (1) the upward trend may have started sooner (possibly day 110) and (2) this upward trend did not seem significant as the frequency of self-injuries– relative to the entire timeline – was already high between day 57 and day 146.
RkJQdWJsaXNoZXIy MTk4NDMw