The net response rate was 41% (N= 82). As can be inferred from Table 1, presenting the descriptive demographic features of participants, our sample had a relatively equal distribution of age, gender and training institute. Additionally, 78% of participants reported that they had performed more than three DOs over the past three-month period.
Table 1. Demographic variables of GP supervisors (N= 82).
Assessing the structure of the measures in the TPB-questionnaire
Based on reliability analyses of the direct measures of the TPB questionnaire, the Cronbach’s α values were 0.92 for intention, 0.75 for subjective norms, 0.73 for attitude, and 0.65 for perceived behavioural control, indicating that the internal consistency values ranged from satisfactory to very good across our sample (37). Scores for the control belief (CB) items were normally distributed among the questionnaire scores, a precondition for using the maximum likelihood extraction method in the exploratory factor analysis (48). A scree plot of the extracted factors pointed to the presence of two separate factors in the data, with Eigenvalues of 4.0 and 1.8 respectively. CB factor 1 contained four suitable loadings and CB factor 2 contained five suitable loadings (Table 2). These two factors were iteratively interpreted by four team members (AT, MG, IM, LJ). The best fitting descriptive label for CB factor 1 (4 items) was found to be ‘self-confidence in performing the task of DO’ and for CB factor 2 (5 items) ‘feelings of stress induced by practical conditions that limit the performance of DOs’. We treated these two factors as distinct measures in the analysis of the TPB questionnaire.
Table 2. Exploratory factor analysis of control beliefs in the TPB model.
Determinants of supervisors’ intention to perform DOs
The means, standard deviations and correlations between the extended TPB measures are presented in Table 3. On average, participants reported a strong intention to perform DOs, and a positive attitude, positively perceived subjective norms and a relatively high perceived behavioural control towards performing DOs (all mean scores above five on a seven-point scale, Table 3). A significant and positive correlation between intention and all indirect and direct measures in the TPB model was revealed. Furthermore, the indirect, belief-based measures were all positively and significantly correlated with their corresponding direct measurements. Likewise there was a significant correlation between control beliefs (total) and its direct measure ‘perceived behavioural control’(r=.33, p<0.01). But, unlike CB factor 2 (‘feelings of stress induced by practical conditions that limit the performance of DOs’, r=.56, p<0.01), CB factor 1 (‘self-confidence in performing the task of DO’ ) had no significant correlation with perceived behavioural control (r=.06, ns). From the additional measures only past behaviour had a significant correlation with intention; there was no significant correlation between intention and the measures ‘social pressure’ and ‘modelling’. However, there was a positive correlation between modelling and normative beliefs and a negative correlation between social pressure and supervisors attitude towards performing DOs.
Table 3: Correlation matrix and descriptive statistics for measures of the extended TPB model .
Predicting supervisors’ intention to perform DOs.
We conducted a four-step hierarchical regression analysis to test the impact of the measures of the extended TPB model on the prediction of supervisors’ intention to perform DOs (Table 4). In the first step of the analysis, we entered the demographic variables into the predictive equation. The demographic variables did not account for a statistically significant proportion of the variance in intention (R2 change=0.02, F(5,76)= 0.37, p= 0.87). Entry of the indirect measures of the TPB in step 2 resulted in a statistically significant increase in the variance explained (R2change = 0.29, Fchange (4,72) =3.56, p <0.01). Normative beliefs (i.e. beliefs about the normative expectations of other people) and CB factor 2 (i.e. feelings of stress induced by practical conditions that limit the performance of DOs) had a significant positive beta weight of .30 and .25 respectively, reflecting their contribution to the predictive equation. Behavioural beliefs (β =.18) and CB factor 1 (β=.03) were not significantly related to intention. At step 3, the addition of the direct measures resulted in a further increase in the variance explained (R2change = 0.8, Fchange (5,67)=2.76, p= .31); the weighted betas of all single measures were no longer significant, indicating no single measure had a significant contribution in the predictive equation of the intention to perform DOs. At step 4, entry of the additional variable ‘past behaviour’ resulted in a further statistically significant increase of 8% in the variance explained to 45% (R2change = 0.8, Fchange (1,66) =3.57, p=.003). Moreover, both normative beliefs (β=.27, P<0.05) and past behaviour (β=.33, p<0.01) had a statistically significant beta weight.
Table 4. Determinants of the intention to perform direct observations resulting from a hierarchical regression analysis (N=82).