Of 255 students, 209 returned the questionnaire (response rate 82.0%). Of these, 95 students were in their fifth year, and 114 were in their sixth year. Almost twice as many were female (134) as male (75). Most were not married (183). One-half (107) were housed in dormitories or home stays, about one-third (76) lived with parents and a small proportion lived with other relatives (17) or alone (9). It is typical for out-of-town students to live with relatives, in dormitories or in home stays.
The means and standard deviations for the subscales of students’ approaches to learning and estimates of internal consistency are presented in Table
. The scores for the deep learning approach were higher than those for the surface or achieving approach. Preclinical students in medical school are considered to have to be very competitive, high achievers to obtain high marks in examinations that test a wide range of factual knowledge. However, once a student has passed the early part of the course, success in the clinical part is assumed. Student comments during focus group discussions suggested that they are less competitive in their clinical years than during the earlier, more theoretical part of the course. Hilliard found that senior students were more interested in learning for the sake of learning and in acquiring the skills of a competent physician [12
]. Furthermore, in the departments involved in the present study, assessments tend to require explanation and comprehension (oral examinations, case presentations and written assignments) and not merely memory recall. This practice may require students to use more deep learning.
Mean scores on the subscales of approaches to learning
Several studies of medical students in their non-clinical years have used either the Biggs SPQ instrument to measure surface, deep and achieving learning approaches, or the Entwistle–Ramsden Lancaster Approaches to Studying Inventory to measure reproducing, meaning and achieving orientations to learning [4
]. The scoring methods used, numbers of items and medical school classes differed between these studies. All studies except Martenson’s found deep or meaning scores to be the highest [23
]. The Martenson study, conducted in a traditional medical school, reported surface scores as the highest.
The overall alpha value and the alpha value for each scale indicated a high level of internal consistency (0.91) comparable with that in other studies [7
]. In the present study, there were only four items per subscale. This could have been the reason for the higher alpha reliability (range 0.65–0.84) compared with the range (0.60–0.75) reported in studies with seven items per subscale [9
]. Compared with a similar questionnaire applied in medicine, the alpha reliability was higher (0.57–0.67 in Hilliard [12
]). Factors that may have influenced the higher reliability in this study were number of items (this study used the shorter version) and homogeneity of the students (age, marital status, and living arrangement) who completed the questionnaire. These results validate use of the modified questionnaire for assessing approaches to learning in a clinical setting. Further studies with larger samples are recommended to investigate the applicability of the questionnaire in other clinical settings.
To evaluate the construct validity of the questionnaire, the six subscale scores were subjected to principal component factor analysis followed by an oblimin rotation to a simple structure. Eigenvalues greater than 1.0 were adopted as a criterion for determining the number of factors to be extracted. Two factors with eigenvalues above 1.0, which accounted for 67.7% of the variance, were obtained and are shown in Table
. Examination of the screen test supported the two-factor solution.
Factors resulting from the factor analysis of approaches to learning subscales
On the basis of Biggs’s original theory, there should be a three-factor solution identifying three learning approaches: deep, surface and achieving. However, subsequent research [9
] suggested that there are only two solutions, deep and surface. The achieving subscales were usually found to either load on one of the other two factors or be divided between them. In 2001, Biggs and colleagues verified the revision of the SPQ with a two-factor structure [10
]. The revised instrument assesses deep and surface approaches only and uses fewer (20) items. The present study also supports a two-factor solution, with the “achieving strategy” loaded on factor 1 (deep approach) and the “achieving motive” loaded on factor 2 (surface approach). These two factors explain 45.1% and 22.5%, respectively, of the variance. Further discussion therefore considers only the deep and surface approaches to learning.
The relationships between the two scales and their subscales are shown in Table
. There was a positive correlation (.845; .887; and .872) between the deep approach and its three subscales and also between the surface approach and its subscales (.765; .719; and .756). The correlations were significant at the 0.01 or 0.05 level. Weak correlations between approaches and subscales show that the two approaches were dissimilar.
Correlation between approaches to learning scales (coefficient correlations)
A surface approach will work if a student is trying to memorize information for reproduction [1
]. However, excessive use of the surface approach can lead to missing interconnections between elements, or the meanings and implications of what is learned. Students who use a surface approach tend to have an external locus of control (e.g. are highly oriented toward grades), study less and target their study toward material they believe will be included in the examination. Students who are less oriented toward grades have a deep approach to learning and study material in a broader sense.
The students’ approaches to learning did not change significantly as they rotated through the three departments involved in this study. Table
presents the repeated measures analysis of variance results.
Comparison between approaches to learning scores in three medical departments
The small differences in students’ approaches in this study suggest that contextual or environmental variables as well as individual perceptions of these variables influence students’ use of deep or surface approaches to learning tasks. The lack of significant results in the longitudinal study is possibly attributable to the small sample used (n=39) or the short period of time between measurements (six weeks for neurology and 12 weeks for internal medicine and surgery).
Despite there being no significant differences, the pattern of fluctuating surface and deep approach scores across different departments supports the idea that students have individual combinations of approaches when facing different tasks in the departments. These have been called study “orchestrations” by Meyer [26
] and imply that students use strategies flexibly, according to which is most appropriate to the learning environment. In the present study, students had higher deep approach scores in all three departments, which may be explained by their being senior, clinical students. Lindblom-Ylanne and Lonka [27
] suggested that an increasingly high degree of deep-level learning is needed toward the end of medical studies, and indeed found that this occurred in their study.