Validity of a Job-Exposure Matrix for Psychosocial Job Stressors: Results from the Household Income and Labour Dynamics in Australia Survey.

introductionA Job Exposure Matrix (JEM) for psychosocial job stressors allows assessment of these exposures at a population level. JEMs are particularly useful in situations when information on psychosocial job stressors were not collected individually and can help eliminate the biases that may be present in individual self-report accounts. This research paper describes the development of a JEM in the Australian context.

methodsThe Household Income Labour Dynamics in Australia (HILDA) survey was used to construct a JEM for job control, job demands and complexity, job insecurity, and fairness of pay. Population median values of these variables for all employed people (n = 20,428) were used to define individual exposures across the period 2001 to 2012. The JEM was calculated for the Australian and New Zealand Standard Classification of Occupations (ANZSCO) at the four-digit level, which represents 358 occupations. Both continuous and binary exposures to job stressors were calculated at the 4-digit level. We assessed concordance between the JEM-assigned and individually-reported exposures using the Kappa statistic, sensitivity and specificity assessments. We conducted regression analysis using mental health as an outcome measure.

resultsKappa statistics indicate good agreement between individually-reported and JEM-assigned dichotomous measures for job demands and control, and moderate agreement for job insecurity and fairness of pay. Job control, job demands and security had the highest sensitivity, while specificity was relatively high for the four exposures. Regression analysis shows that most individually reported and JEM measures were significantly associated with mental health, and individually-reported exposures produced much stronger effects on mental health than the JEM-assigned exposures. DISCUSSION: These JEM-based estimates of stressors exposure provide a conservative proxy for individual-level data, and can be applied to a range of health and organisational outcomes.