The four dimensions of resource loss, as well as total resource loss, were analyzed separately to determine whether or not they had differential effects on the proposed mediational relationship. Thus, the five primary hypotheses of this study were as follows:
- Object resource loss will serve as a mediator in the relationship between exposure and total distress.
- Condition resource loss will serve as a mediator in the relationship between exposure and total distress.
- Personal characteristics resource loss will serve as a mediator in the relationship between exposure and total distress.
- Energy resource loss will serve as a mediator in the relationship between exposure and total distress.
- Total resource loss will serve as a mediator in the relationship between exposure and total distress.
More specifically, there is greater evidence to hypothesize a mediating effect for object resource loss and condition resource loss. For instance, Vernberg et al. (1996) found that factors related to object resource loss are significantly correlated to children’s levels of PTSD. Likewise, factors related to condition loss have also shown significant correlations to children’s levels of PTSD (Garrison et al., 1995; Vernberg et al., 1996). Thus, there was a greater likelihood that the mediational hypotheses for object resource loss and condition resource loss would be supported than the other three hypotheses. A mediating effect for resource loss would be established if the statistical relationship between exposure and distress was found to drop in significance once resource loss was entered into the regression equation (Baron & Kenny, 1986, Holmbeck, 1997).
Additionally, several demographic variables were tested for. As girls have consistently been shown to exhibit more distress following traumatic events (Lonigan, Shannon, Finch, Daugherty, & Taylor, 1991; Shannon, Lonigan, Finch, & Taylor, 1994; Vernberg et al., 1996), the effects of gender were examined. Additionally, African Americans have been shown to exhibit greater levels of PTSD following trauma (Lonigan et al., 1991). Thus, this portion of the study examined race as well. Finally, as children’s ages have been shown to be associated with differential levels of PTSD (Lonigan et al., 1991; Shannon et al., 1994), this demographic variable was also investigated. Given the often contradictory nature of several of these studies, no specific hypotheses will be made with regard to demographic variables.
The data examined in this study were collected by Jones and Ollendick (2002) in the context of a National Institute of Mental Health (NIMH) grant designed to assess the psychological impact of residential fire on children. The sample in this study consisted of 140 children who had experienced a residential fire. The median age of this sample of children was 12.11 (SD = 2.08). Fifty-four percent of the children in the study were girls and 46% were boys. With regard to race, 74 of the children were identified as African American while 66 were identified as Caucasian. Additionally, 54% of the children in the sample were at home during the time of the fire. Approximately 30% of the children thought that they were going to die during the fire, 10% observed some family member get hurt or burned, and 10% lost a family pet. In preparation for the current study’s analyses, a listwise deletion of subjects was conducted to address missing data. Following this listwise deletion, the remaining sample size was 120 children (mean age = 12.31, SD = 2.83). Sixty-eight of the remaining children in the study were girls and 52 were boys. Finally, with regard to the racial makeup of the remaining 120 children, 62 were identified as African American, while 58 were identified as Caucasian.
A modified version of Freedy’s Resource Loss Scale (FRTE; Freedy, Shaw, Jarrell, & Masters, 1992, Jones & Ollendick, 1994) was used to obtain resource loss ratings from children. This Resource Loss Scale for Children (RLSC; Jones & Ollendick, 1994) provides a measure of children’s total resource loss. In addition to a total resource loss scale, it contains subscales measuring Hobfoll’s (1998) four resource loss subtypes (object loss, condition loss, personal characteristics loss, and energy loss). The RLSC is a self-report measure containing 22 items assessing loss along the four subtypes. Children responded “yes” or “no” to whether or not they experienced the loss of a particular item. With respect to items where they responded “yes”, children were asked to further rate the extent of their loss (1 = a little, 2 = some, 3 = a lot). Total loss scores were obtained by adding the impact of loss across the 22 items. Children were reported to have understood the concept of loss and were able to provide information about the way in which the loss affected their lives. A Cronbach’s alpha analysis was performed to obtain internal consistency ratings for all four resource loss types, as well as total resource loss in the current study’s sample (see Table 2). Using standardized values for the RLSC, internal consistency for the current study was as follows for each particular type of resource loss: object resource loss was 0.82, condition resource loss was 0.58, personal characteristics resource loss was 0.67, energy resource loss was 0.65, and total resource loss was 0.81. Of the five variables examined, only object resource loss and total resource loss met the 0.70 reliability coefficient threshold established by Nunnaly (1978). A summary of the RLSC’s results is presented in Table 1.
The Fire Questionnaire (Jones & Ollendick, 2002) was used to obtain a measure of children’s exposure to the fire. It uses a combination of multiple response items and open-ended questions to assess children’s level of exposure to residential fires. During their original NIMH study, Jones and Ollendick (2002) implemented a grouping of Fire Questionnaire items that could be used to obtain a measure of exposure. However, using this grouping proved to be inappropriate for the current study. Some of the items used by Jones and Ollendick (2002) involved content related to resource loss. For example, the personal injury item suggested in this grouping could also be conceptualized as an energy resource loss item. Thus, using the exposure measure suggested by Jones and Ollendick (2002) for the current study would have been inappropriate, in that it would adversely affect the independence between the predictor variable and the potential mediators (resource loss types). As such, an appraisal item from the Fire Questionnaire was used as a measure of exposure. This appraisal item asked children to rate how serious they thought the fire had been after it was over (1 = not at all, 2 = a little, 3 = some, 4 = a lot). It was believed that this item allowed for an assessment of the scope and seriousness of the fire, while still staying relatively independent of the resource loss items used to measure the potential mediator. Thus, this appraisal item was used as a measure of exposure for the current study. Currently, psychometrics for the Fire Questionnaire are still being determined.
As such, there is no psychometric information concerning this instrument available at this point.
A summary of the Fire Questionnaire’s results is presented in Table 1.
The Children’s Reaction to Traumatic Events Scale (CRTES; Jones, 1994) was used to obtain a measure of children’s distress. This 15-item self-report assesses PTSD symptomatology over the past week. Specifically, this version of the CRTES measures intrusion and avoidance, and then combines these scores to obtain a measure of total distress. Children respond to each item using a four-point likert scale (0 = not at all, 1 = rarely, 3 = sometimes, 5 = often). As previously discussed, total CRTES scores were obtained by adding the scores across the 15 items. Jones, Fletcher, and Ribbe (2002) have suggested that CRTES scores falling between 0 and 14 be used as indicators of low distress, scores falling between 15 and 27 be used as indicators of moderate distress, and scores of 28 and above be used as indicators of high distress. A Cronbach’s alpha analysis was performed to obtain internal consistency ratings for both symptom clusters, as well as total distress (see Table 2). Using standardized values for the CRTES, internal consistency for the current study was 0.86 for intrusion and 0.77 for avoidance. The internal consistency for total distress was 0.87. All three of the variables examined met the 0.70 reliability coefficient threshold established by Nunnaly (1978). A summary of the CRTES’ results is presented in Table 1.
Table of Contents
List of Tables and Figures
Tables and Figures
GET THE COMPLETE PROJECT
The Mediational Role of Resource Loss between Residential Fire Exposure and Psychological Distress