TPC Journal-Vol 10- Issue 3-FULL ISSUE

354 The Professional Counselor | Volume 10, Issue 3 Data Analysis As part of our preliminary analyses, we first tested all variables for the assumptions of analysis. Specifically, when examining the skew and kurtosis of the composite variables, we used ±2 as our acceptable range of values. Following advice from Tabachnick and Fidell (2019), we transformed variables that did not meet our criteria for normality. To better understand family functioning, we conducted descriptive analyses for all seven predictive variables (problem-solving, communication, roles, affective responsiveness, affective involvement, behavioral control, and conflict) separately for adolescent and caregiver scores. We assessed the degree of healthy family functioning using I. W. Miller et al.’s (1985) suggested cut-off scores, which can be used to distinguish between healthy and unhealthy family environments. We also conducted paired sample t -tests to compare the adolescent and caregiver reports of family functioning. Next, we tested the fit of our theoretical model of family functioning using structural equation modeling (SEM) with maximum likelihood as the method of estimation. We used multiple fit indices to assess the model fit. Specifically, the chi-square statistic assesses absolute model fit, demonstrating good fit when not statistically significant. The chi-square test can also be used to compare the relative fit of two models. Additionally, comparative fit index (CFI), root mean square error of approximation (RMSEA), and standardized root-mean-square residual (SRMR) are all indicators of model fit, with 0.95 or higher, 0.05 or lower, and 0.08 or lower indicating good fit, respectively (for more information on SEM fit indices, see Hooper et al., 2008). Notably, Iacobucci (2010) suggested that researchers can use SEM and establish good model fit even with small samples. We also conducted descriptive analyses of the participants’ self-reported SIB. We left these variables raw (untransformed) to evaluate how participants viewed their own SIB. We examined the specific SIB methods that participants reported using (e.g., cutting, burning) as well as three outcome variables (suicidal SIB, nonsuicidal SIB, and ambivalent SIB; all transformed because of issues with skew and kurtosis). Lastly, we used SEM to predict SIB with the proposed model of family functioning. Given our small sample size, we conducted this analysis separately for suicidal SIB, nonsuicidal SIB, and ambivalent SIB. We set alpha at .05 for each model; given the small sample size, we did not apply corrections to the alpha for the multiple analyses. Results We used SPSS 24.0 and Amos 24 to analyze our data. Because this study was primarily descriptive, we conducted multiple analyses to better understand the family environment of treatment-seeking adolescents, experiences of SIB for adolescents, and the role of family environment in adolescent engagement in SIB. Family Characteristics and Functioning Means, standard deviations, and range of scores for the family functioning variables are shown in Table 1. With the exception of the caregiver reports on affective responsiveness and behavioral control, both adolescent and caregiver reports on every subscale of the FAD fell above the McMaster clinical cut-off (see Table 1) described by I. W. Miller et al.’s (1985) cut-off scores, indicating on average all of the families demonstrated unhealthy functioning. It is worth noting that adolescents and their caregivers reported similar levels in five of the seven indicators of family functioning from the FAD and CBQ