Longitudinal Study into the Relationships Between Psychological Safety, Burnout and Percieved Quality of Care

The study will test four longitudinal structural equation models: Model 1: a RI-CLPM investigating the links between psychological safety and perceived quality of care across the four study timepoints, and whether higher than personal average psychological safety at one timepoint can predict higher than personal average quality of care at a subsequent timepoint. The reverse path (that higher QoC predicting higher subsequent psychological safety) will be estimated and reported but it is not hypothesised due to the limited theoretical backing. Model 2: A RI-CLPM same as model 1, but with burnout as the outcome of interest as opposed to quality of care. The reverse path will be estimated, but it is hypothesised unlike model one. In this model, the path that higher burnout on average for a participant at any timepoint will predict subsequently lower psychological safety will be tested. Model 3: A final RI-CLPM of the same structure as model 2, but with work engagement as the outcome. Like model 2, its reverse path will be predicted and hypothesised; in this case the idea that higher average work engagement for a participant at one timepoint, will predict subsequently higher psychological safety. Model 4: Will use a latent change score framework as opposed to the earlier RI-CLPMs. The model will test if baseline familiarity is associated with psychological safety, whether familiarity increases across the four study timepoints, whether familiarity at each study timepoint predicts subsequent increases in psychological safety, and finally whether changes in familiarity are associated with changes in psychological safety. The theoretical basis for this model (and the need for a different analysis) is that familiarity is expected to grow systematically over the study period based on past research. For each model, a range of individual level covariates (such as age and job role) and work related variables (Areas of Work-Life, Time pressure, Inclusive leadership) will be measured at each of the study timepoints. The former will be modelled as time invariant, while the latter will be accounted for as time variant. The full range of predictors and covariates will be explained in the "Predictors and Covariates" document. These four models test the core study hypotheses, these will each be fit twice – once with the Edmondson (1999) measure, one with the SCORE (Adair et al, 2022), giving 8 confirmatory model fits in total. Other models will be used, substituting the overall scores for subscales of each construct (burnout, work engagement, familiarity, quality of care) resulting in a further 6 or 7 models depending on the outcome of the AHRQ factor analysis described in the statistical models’ section. These will be fit using the Edmondson measure only. As these models are exploratory, they are not hypothesized for.