Design
The current study is based on data collected in June and July of 2016 as part of a larger investigation into life satisfaction among public sector social workers in the OPT. The investigation used a cross-sectional design and employed convenience sampling techniques. The project received support from the Ministry of Social Development (MOSD; formerly Ministry of Social Affairs) of the Palestinian National Authority and human subjects approval from the Institutional Review Board at a major research university in the Northeast of the United States of America. As such, it was deemed to be in compliance with ethical standards for research, including the Declaration of Helsinki.
Data source
The target population consisted of MOSD social workers who are organized into 12 directorates and local offices in West Bank cities and towns such as Ramallah, Jericho, Salfit, Nablus, and Hebron. These public employees provide a wide range of direct services (e.g., economic assistance, health prevention/treatment, educational and social programming) to various constituents: abused children, disabled individuals with chronic conditions, older adults, battered women, and families and individuals living in poverty.
Researchers worked with MOSD administrators to develop a schedule for data collection and sent an announcement of the voluntary, unpaid research opportunity to each local office. The second author then visited directorates and local offices, holding small group meetings to introduce the purpose and procedure of the study and distribute and review consent forms. Interested participants signed consent forms prior to completing the survey. The researcher remained on-site to answer questions, collect surveys, and debrief participants.
The survey consisted of 100 closed-ended items based on adapted versions of standardized measures of concepts such as life satisfaction, organizational support, job stress, and mental and physical well-being. Measures of mental and somatic health were situated within the first one-third of questions in the survey; demographic and background questions were contained in the last one-third of the survey. Measures were translated from English into Modern Standard Arabic by a nationally-certified Arabic language instructor with nearly three decades of teaching experience at the high school and college levels in the United States and the Middle East. She is a leader in the design of Arabic language curriculum for both traditional and online courses at U.S. colleges and high schools and has nearly 20 years of professional translation experience, including standardized state educational assessments such as the Michigan Educational Assessment Program.
To promote accuracy, standard translation protocol and techniques (e.g., adaption, transposition, multiple sourcing; [23]) were employed. Additionally, two faculty members at Al-Quds University, Jerusalem, completed quality checks for the entire translated survey. Both professors have appointments in the Department of English Language and Literature, hold doctorate degrees, and have research expertise in translator training, translation technology, and discourse analysis. Quality checks resulted in numerous clarifications and modifications to ensure items were comprehensive and acceptable to the target group. Researchers kept extensive records of translation efforts as part of a thorough audit trail.
Measures
Psychological distress
This concept was assessed using the Kessler Psychological Distress Scale, a measure of non-specific psychological distress based on a framework that includes behavioral, emotional, cognitive, and psychophysiological manifestations [2]. The scale was created using highly sensitive items that identify extreme psychological distress in the general population. The ten-item version (K10) measures frequency with which respondents experienced symptoms in the past month, including nervousness, hopelessness, sadness, worthlessness, and fatigue. Response choices are based on 5-point Likert-type scale ranging from 1 (none of the time) to 5 (all of the time). Responses are summed to create a total score (range = 10–50) with higher scores signifying more psychological distress. Research has suggested that the optimal cut-point for a psychological disorder is 24 [24]. In previous studies, K10 had strong scale reliability with Cronbach’s α greater than 0.88 [15, 25].
K6 is a shortened, six-item version of the K10 that assesses frequency of the following mental health symptoms in the past month: feeling nervous, hopeless, restless or fidgety, so sad that nothing could cheer them up, that everything was an effort, and worthless. In the current study, items were extracted from the K10 and used the same response set. [9, 26]. Responses were summed to produce a total score (range = 6–30), with higher scores signifying more distress. Based on a previous study [13], the K6 cutoff point for psychological disorders for our study was 16.25. K6 has been found to be reliable with Cronbach’s α ranging from 0.89 to 0.92 [1].
Both scales are easy to understand and publicly available; interviewer-administration and self-administration versions are online [1]. English versions of K10 [16] and K6 [3, 27] have been validated by past research.
Generalized anxiety
Generalized Anxiety Disorder (GAD-7) is a seven-item measure of respondents’ level of recent anxiety [26]. Respondents were asked how often they were bothered by problems (e.g., “not being able to stop or control worrying” or “worrying too much about different things”) in the past two weeks. Response choices were based on a 4-point Likert-type scale ranging from 0 (not at all) to 3 (nearly every day). Item responses were summed to produce a total score ranging from 0 to 21; higher scores signified more anxiety. Previous research among patients in primary care clinics suggested a cut point score of 10 for identifying anxiety disorders [26].
Somatic symptoms
Somatic Symptoms Scale (SSS-8) was used to assess the level of recent somatic symptoms burden. Previous research has found that the SSS-8 is a reliable and valid self-report measure of somatic symptom burden [28]. Respondents were asked how often in the past week they were bothered by common problems such as headaches, pain (arm/leg/joint), stomach or bowel problems, and sleep problems. Response choices were based on a 5-point, Likert-type scale ranging from 0 (not at all) to 4 (very much). Total scores ranged from 0 to 32, with higher scores signifying more burdens. Suggested cut points for SSS-8 are as follows: 0–3 points (minimum to no burden), 4–7 points (low), 8–11 points (medium), 12–15 points (high), over 16 points (very high burden) [28].
Background characteristics
Demographic and background characteristics were assessed, including age (years), gender (male/female), marital status (married, never married, other), educational level (secondary diploma, college diploma, bachelor’s degree, master’s degree or higher) refugee status (yes/no), full-time employment (yes/no), and monthly income (U.S. dollars).
Data analysis
Descriptive statistics and correlation tests were performed using SPSS, version 24.0 [29]. Confirmatory factor analysis (CFA) was performed using LISREL, version 9.1 student edition [30]. Consistent with recommended practice when a dataset has minimal levels of missing data (i.e., < 5%), listwise deletion was used [31]. Cases with missing data on variables of interest in our analysis were removed, resulting in a final sample size of 234. Before reporting univariate statistics for demographic background and mental health variables, multivariate normality was examined and confirmed for both K6 and K10 versions.
Next, a variance-covariance matrix with maximum likelihood (ML) estimation was used as input matrix. We reported and compared model fit indices for three models: one-factor K10 model, one-factor K6 model, and two-factor K6 model. χ2 statistics and significance levels were reported. A large and significant χ2 indicates poor model fit [32]. As suggested by Schmitt [33], we went beyond a global model evaluation and conducted additional analysis using several fit indices: root mean square error of approximation (RMSEA; 34), comparative fit index (CFI; [34]), Akaike information criteria (AIC), Bayesian information criteria (BIC), and standardized root mean square residual (SRMR; [34]). We applied Byrne’s [32] suggestion that fit indices should serve as guidelines that provide information on a model’s lack of fit and should be used along with “theoretical, statistical, and practical considerations” (p. 77). Current guidelines suggest that CFI values greater than or equal to 0.90 indicate acceptable fit; values greater than or equal to 0.95 imply very good fit [35]. RMSEA values less than 0.05 indicate close model fit, and values exceeding 0.10 indicate poor fit [36]. SRMR values less than 0.08 also indicate good fit [35].
We also examined standardized residuals and individual parameter estimates for three models. In studies that screen for mental illness among the general population, Kessler et al. [27] suggested that the unidimensional K6 model performs the best. In another study that examined populations for non-specific psychological distress, Kessler and colleagues [1] found support for a single factor model of K10. Bessaha [3] suggested a two-factor K6 has better model fit than one-factor K6 in screening for psychological distress within young adult populations. To compare the three competing models among our sample, we examined each question on the scale, evaluating signs and magnitude of each parameter.
Last, we performed the Pearson’s correlation test to examine relationships between the K6 total score and its subscores. We also evaluated convergent validity by examining correlations between K6 and two other scales measuring mental health: GAD-7 and SSS-8.