139 Case study challenging behavior 6 care (e.g., anger after imposed security measure). Challenging behavior always came with some form of emotional tension. To better interpret the thematic map, the researcher then asked the participant’s clinician whether the participant knows better and worse times and what typically indicates to staff whether her overall well-being is high or low. Before having seen the results, she confirmed that the frequency of self-injurious and physically aggressive incidents is most telling about her overall well-being. This indicates challenging behaviors summarize her overall state. From the (sub)themes generated in the thematic analysis, the clinician then identified 11 staff-hypothesized risk- and protective factors for her challenging behaviors. These factors were either specific codes or broader (sub)themes: reliving past trauma, hallucinating, negative affect, receiving medical care, receiving compliments, the imposing of freedom restricting measures, experiences of physical pain and sickness, receiving psychological therapy, tensions with her family, and positive social interactions. These variables were used for subsequent analyses. 3.2 Describing change trajectory The participant completed the daily survey 494 times during the 560 days (88%). Physical aggressive incidents were self-reported on 65 days (13%), while self-injury was self-reported on 247 days (50%). Staff reported aggressive and self-injurious incidents on respectively 75 days (16%) and 164 days (33%). A χ2 test indicated agreement between self- and informant ratings. That is, counts of observed matches between self- and informant ratings of these challenging behaviors (i.e., both reporting daily presence or absence of behavior) was significantly higher than the expected count for self-injury, χ2 (1, N =494)=91.56, p <0.001, and for aggression, χ2 (1, N = 494) = 12.76, p < 0.001. As both challenging behaviors can occur without being noticed by staff (e.g., when on leave), we analyze self-reported challenging behavioral dynamics. Figure 4A illustrates the raw binary timeseries of self-reported physical aggression and self-injury for 560 days. The recursive partitioning algorithm (Hasselman, 2023) first detected mean-frequency changes in raw diary timelines (4A) – the outcome of which is visualized with dashed and solid lines in Figure 4B. After visual inspection of the binary timeseries (4A) and their mean-levels (lines in 4B), we found 10 transitions that mark that end of an old- and start of a new attractor (colors in 4B). When the mean-level changes detected by recursive partitioning (up or downward trend in lines 4B) of the two challenging behaviors occurred in the same direction within close proximity to one another (i.e., within 14 days), we marked it as transition that starts or
RkJQdWJsaXNoZXIy MTk4NDMw