Background There is increasing recognition of job insecurity as an emerging issue in public health. We sought to examine whether job security improvements were associated with improvements in mental health in a large, working population-representative repeated-measures panel study.

Methods We used both within-person fixed effects (FE) and random effects (RE) regression to analyse data from 14 annual waves of a national Australian survey (19 169 persons, 1 06 942 observations). Mental Health Inventory-5 scores (outcome) were modeled in relation to self-reported job security (categorical, quintiles), adjusting for age, year, education, and job change in the past year.

Results Both FE and RE models showed stepwise improvements in MHI-5 scores with improving job security, with stronger exposure-outcome relationships in the RE models, and for men compared to women. All models showed roughly monotonic improvements in MHI-5 score by quintile of improvement in job security. The strongest relationship was observed in the RE model for males: for a one-quintile improvement in job security, beta=2.06 [1.67, 2.46], and the following for two- (3.94 [3.54, 4.34]), three- (5.82 [5.40, 6.24]), and four- (7.18 [6.71, 7.64]) quintile improvements. The FE model for males produced slightly smaller coefficients, reaching a maximum of 5.55 [5.06, 6.05] for a four-quintile improvement.

Conclusions This Australian national panel study showed a strong dose-response relationship between job security and depression and anxiety symptoms. Stronger causal inference over previous observational research is supported by the dose-response finding and the relative consistency of the FE and RE results. Policy and practice intervention to improve job security could benefit population health.