Promoting research utilisation and evidence-based decision making amongst healthcare managers. Utilising nonrecursive structural equation modelling to develop the Theory of Planned Behaviour

The purpose of this study was to identify those factors that impede or encourage the use of research-based decision making by healthcare managers. The research drew on the Theory of Planned Behaviour (TPB), self-efficacy theory, expectancy theory and theories of research-based decision making. The interrelationships between individual, role and organisational variables were investigated, utilising nonrecursive structural equation modelling (SEM) to consider the reciprocal relationships between the variables, including the impact of past behaviour on future behaviour and its mechanisms of action. Nonrecursive structural equation modelling will be discussed in greater detail in a later chapter. However, it is worth saying a little about such models at this stage as the term has a precise meaning within structural equation modelling. Nonrecursive models are those which (perhaps counter intuitively) contain feedback loops between variables. Evaluating nonrecursive models is difficult, since it requires an understanding of necessary and sufficient conditions in order to claim that the model is identified (i.e. stable parameter estimates can be obtained). A further aim was to explore the feasibility of employing a cross-sectional methodology to test hypotheses regarding causation, utilising SEM to estimate causal versus non-causal aspects of observed correlations. This approach enabled the testing of causal hypotheses in an area where the conditions for longitudinal research could not have been met. Causal modelling examined the extent to which the data failed to agree with the nonrecursive model that represented the hypotheses to be tested