TPC Journal-V6, Issue 4- FULL ISSUE

The Professional Counselor /Volume 6, Issue 4 351 subscale of devaluating client given: (a) the low factor loading produced after modification of the initial model, and (b) the lack of normality in the composite score. Next, we examined the new modified measurement model that included the removal of the subscale devaluing client. The resulting model (see Figure 1) produced a model in which all fit indices indicated a good fitting model: χ 2 ( df = 2, N = 926) = 8.25, p = .02; GFI = .99; CFI = .99; RMSEA = .06; SRMR = .02; NFI = .99; TLI = .98. The modified measurement model for burnout produced acceptable standardized factor loadings ranging from .53 (incompetence) to .63 (negative work environment), .77 (deterioration in personal life), and .85 (exhaustion). In review of the model fit indices and standardized factor loadings, we deemed the measurement model acceptable for use in the structural model. Structural model. We developed the structural model (see Figure 1) based on a review of the literature, and it was theorized in this model that school counselors’ perceived stress correlates to school counselors’ burnout and contributes to the frequency with which they provide direct student services. In addition, this model tested the hypothesized model that school counselors’ burnout contributes to their frequency of direct student services. The structural model includes the measurement model previously tested that consisted of the latent variable of burnout. School counselors’ perceived stress and burnout were defined as exogenous or independent variables. Perceived stress was a manifest variable consisting of participants’ composite scores on the PSS (Cohen et al., 1983). Additionally, we defined the manifest variables of percentage of time at work providing direct services to students, direct curriculum activities, and direct counseling activities as the endogenous or dependent variables that measure participants’ facilitation of direct student services. The variable of percentage of time at work providing direct services to students was a single demographic item reported by participants, while direct curriculum activities and direct counseling activities were the participants’ composite scores derived from subscales on the SCARS (Scarborough, 2005). In addition, the error terms of the direct student services variables—percentage of time at work providing direct services to students, direct curriculum activities and direct counseling activities—were correlated given that they measure similar constructs. An examination of the structural model indicated a strong goodness of fit for all fit indices except for chi-square: χ 2 ( df = 14, N = 926) = 108.37, p < .001; GFI = .97; CFI = .96; RMSEA = .07; SRMR = .04; NFI = .95; TLI = .91. The researchers deemed the structural model as suitable with these data despite the significant chi-square (Henson, 2006; Kline, 2011; Weston & Gore, 2006). A closer examination of the standardized regression weights identified that school counselors’ burnout scores contributed to 12% (β = -.35, p < .001) of the variance in their direct counseling activities and 5% (β = -.22, p < .001) of the variance in their direct curriculum activities. Furthermore, school counselors’ burnout scores contributed to 6% (β = -.24, p < .001) of the variance in percentage of time at work providing direct services to students. Perceived stress did not contribute to direct counseling activities (β = .11, p = .04), direct curriculum activities (β = .06, p = .31), and percentage of time at work providing direct services to students (β = .04, p = .51). In addition, perceived stress and burnout produced a statistically significant correlation (β = .75, p < .001; 56% of the variance explained). The structural model (Figure 1) indicates that school counselors’ level of counselor burnout had a negative contribution to the frequency of their direct counseling activities, direct curriculum activities and percentage of time at work providing direct services to students. However, it should be noted that the effect sizes of these findings were small to medium (Sink & Stroh, 2006). An additional