An increase in attitude of one point was associated with an increase in the likelihood of a parent immunising by a factor of 13.56 and an increase in number of children by one increased intention by a factor of 5.76. Thus, for dTaP/IPV, stronger intentions to immunise were associated with having more positive attitudes towards vaccination and having more children in the family. These findings suggest that whilst behavioural beliefs and control beliefs were mediated by their respective TPB components (attitude and perceived ABT199 control, respectively), there was an unmediated effect of number of children (the TPB would predict that background variables, such as number of children,

would be mediated by the TPB components). Subjective norm exerted no influence on

intentions to immunise. It has been argued that stepwise methods are not appropriate for theory testing because they are influenced by random variation in the data and so often do not give replicable results if the model is retested within the same sample [24]. However, some studies have used stepwise regression methods to predict immunisation intentions or a child’s immunisation status [9] and [13]. Thus, in order to check the above analyses, a stepwise regression was run with the direct predictors of intention entered in the first step and all other variables entered in the second step (MMR and dTaP/IPV separately). These analyses identified the same predictors of intentions as the sequential regression analyses indicating that, regardless of the regression technique used, the significant predictors MG-132 mouse were the same. In addition, as non-significant variables

were included in the regression analyses to determine the effect of additional variables when all existing TPB components were taken into account [23], the logistic regressions were re-run not without the non-significant predictors included. Although not reported here for reasons of space, this too identified the same significant predictors as the regression analyses presented. Each of the belief composites (behavioural beliefs; normative beliefs; control beliefs) was found to correlate significantly with their director predictor of intention (attitude; subjective norm; perceived behavioural control, respectively). Thus, as attitude and perceived control were significant reliable predictors of intention for MMR and attitude for dTaP/IPV, the separate beliefs included within these two proximal determinants were examined. Mann–Whitney U-tests were used to compare parents with maximum immunisation intentions (MI) and parents with less than maximum intentions (LMI) in terms of their scores on the individual behavioural belief and control belief items for each vaccination separately. By identifying the specific beliefs that underlie parents’ attitudes and perceptions of control, the most salient beliefs can then be targeted in future interventions to improve vaccine coverage.