Response rate was 92.8% in the medical students (MS) and 63.0% in the physiotherapy students (PS). Mean age: MS 21.7 years; PS 19.9 years . Male: MS 58.1%; PS 36.0%. Single’ marital status: MS 97.0%; PS 97.5%
Data from the two samples were analysed using the same procedures. The correlation matrices for the ADLIB items were subjected to principal components analyses. Scree tests in each analysis suggested three main components could be extracted. The three-dimensional solution was rotated using the varimax (orthogonal) method to help label the factors, and because it was the method used in the study of the original 39-item ADLI [7, 8]. The final solutions are shown in Table 1 which contains abbreviations of the items (numbered as shown in Additional File Table 2) along with item communality statistics and item variance explained by the dimensions. Loadings of 0.30 and above are shown in bold type. Overall the three-dimensional solutions account for 51.8% of the item variance in Sample 1 and 48.1% in Sample 2. Only two items did not load uniquely, item 1 in Sample 2 and item 15 in Sample 1. For both samples factor 1 corresponds to Distracted Learning in previous reports [8] consisting of high-loading items reflecting limited time spent studying, deferring work to the end of term, cramming for exams and preference for social activities rather than work. Factor 2, identical with the Unsuccessful Learning scale in previous reports, consists of items reflecting a high level of effort in studying, difficulty with organising the content of learning and dissatisfaction with or concern about performance outcomes. Factor 3 corresponds to the Successful Learning dimension in the original ADLI analyses, consisting of high-loading items that reflect efforts to integrate ideas and information into a broader framework, expenditure of cognitive effort on ideas and information, and testing interpretations of results.
In this report the Factor 1 scale title Distracted Learning (DSL) has been retained. The titles of Factors 2 and 3 have been changed from Unsuccessful Learning to Anxious / Inefficient Learning (AIL) for Factor 2, and from Successful Learning to Independent / Deep Learning (IDL) for Factor 3, because the original titles imply learning outcomes rather than describing learning styles.
The similarity in factor structure across the two samples was explored using Harman’s [13] congruence coefficients, which indicate the degree of similarity in item loadings among the factor matrices. Table 2 shows a summary of congruence coefficients (shown in plain text). Underlined in the box are the congruence coefficients for corresponding factors in the two samples - these values were all 0.96. Coefficients for non-corresponding factors were low in value, as would be expected if the overall solutions were similar.
Internal consistency among the items composing each dimension of the ADLIB was calculated using the Cronbach’s α statistic. Table 2 shows the α coefficients along the diagonal of the matrix (bold text). All values were between 0.75 and 0.87, and were comparable to the long version of the corresponding scales (Klimidis, et al. 1997). In Sample 1, which completed the 39-item ADLI, correlations between scales DSL, AIL and IDL from the ADLIB and corresponding scales from the ADLI were 0.95, 0.96 and 0.89 respectively, demonstrating that abbreviating the scales result in very little loss of information.
Table 2 also shows correlations between the ADLIB subscale scores across the two samples. Examination of the inter-correlation matrices (shown in italic typescript in Table 2) indicates a reasonably coherent picture suggesting construct validity of the instrument. Across the two samples there was a significant negative correlation between DSL and IDL scores. That is, compared with low DSL scorers, high DSL scorers (i.e., students who endorse items reflecting poor time management, preference for socialisation over study, etc.) tend not to use learning strategies that involve effort to cognitively organise and contextualise information. Additionally, in the medical student sample, there was a significant positive correlation between DSL and AIL scores. High DSL scorers, compared to low scorers, tend to report more difficulties with managing course information and to report more worry about academic performance. In the medical student sample there was also a significant negative correlation between IDL and AIL scores. That is, those tending to report use of IDL strategies, relative to low scorers on the IDL scale, were less likely to endorse items from the AIL scale, such as worry about performance, expend effort in unproductive study and have trouble organising information.
Table 3 shows the correlations of the ADLIB sub-scales with relevant external measures. DSL scores are not correlated with scores any of the four external measures. AIL is correlated significantly and in the expected directions with all four external measures in accord with expectations. High scorers on the AIL scale were more likely to have lower scores on the Patient Interaction Questionnaire (indicating lower confidence in their interactions with patients), higher scores on the Medical Practices Anxiety Questionnaire, higher scores on the Course Difficulties Inventory and lower scores on the Brief Measure of English Proficiency. Scores on the IDL scale are correlated significantly (and negatively) only with scores on the Course Difficulties Inventory.