Two reviews have investigated learning styles with special focus on health science education. Romanelli stresses the lack of a conceptual framework for both learning style theory and measurement and conclude that faculty members should make concentrated efforts to teach in a multi style fashion [14
]. Cook concludes that further research in web-based learning could clarify the feasibility and effectiveness of assessing and adapting to learning [15
]. Others have criticized the use of learning styles as predictors of learning preferences with the argument that there are more important factors involved in the learning process. Some work has been done since Cooks review was published, but there are only a few papers relevant for medical web-based education [16
]. Further, the findings of these studies were not consistent and it is thus still not known why students prefer to use online learning materials or not.
In this study we found no evidence supporting that students' learning styles, according to ILS, influence the choice to use the web-based ECG-interpretation programme or not in a blended learning setting. This result is in accordance with those of Cook et al who found no association between ILS scores and different web-based format preferences in medical residents [11
]. However, other studies indicate other possibilities. For example, McNulty, studying medical students, collected entry logs for two different web-based applications (a discussion forum and a tutorial). It was found that students with "Sensing" preference tended to use the web-based applications to a larger extent than the ones with an "Intuitive" preference. Further, differences in the usage of web-based applications for the "Perceiving/Judging" dimension in the instrument were also found. Another learning style instrument, the Mayers-Briggs type indicator (MBTI) was used in that study, but an association between the sensing dimension in MBTI and ILS has been described suggesting that these two learning styles indices bear a close resemblance to each other for this dimension.
Even when using one specific instrument, different types of format in the evaluating questionnaires (i e forced-choice preference or Likert scale) may influence the ability to compare the results between studies. Using a self report survey Brown et al came to the conclusion that the learning styles of health science students, as measured by the ILS, can be used, although to a limited extent, as a predictor of students' attitudes towards E-learning. Both Brown and Cook primarily used indirect measures assessed on an Online Learning Environment Survey (OLES) and an end of course questionnaire using a scale ranging from 1-6 as preference. Thus, they did not use recorded activity in the programme as a denominator, as was done in the present study.
An indirect measure, asking students forced-choice preference items, was used by Johnson & Johnson. They found a statistical difference for college students in the active -reflective dimension. Active students preferred face-to-face study groups rather than online study groups, but online quizzes rather than pencil and paper quizzes. However, Johnson & Johnson used the individual result in the four ILS dimensions as a continuous variable and compared the group average instead of the more commonly used scored categorisation [24
]. Their approach lend support by the finding of Cook, that up to one third of learners change style classification although the mean score change is not large [19
Blended learning can be defined in a broader way as a combination of face-to face instruction and computer-mediated instruction [25
However the blend can be of different categories and using a blended learning approach could mean that the web-based component of the course needs to have a specific instructional design to match the personal learning style of the individual student [25
In this study we have defined the blended learning setting as the voluntary opportunity for a student to use web-based learning as a complement to the traditional teaching.
However, in the setting of a self-directed stand-alone web-based course learning styles might have affected the usage differently. A study using ILS found a difference between a self-directed version and a collaborative version of an online course [26
]. In that study significant associations between students' learning styles and success in distance education were found, suggesting a relationship between learning style and ability to respond to web-based learning.
The inconsistency in research results regarding learning styles and their relation to different studying preferences could also relate to differences between studied groups of students (e g different academic level). The Swedish admission system for medical studies is mainly based on high grades and on expected high theoretical academic performance. It might be assumed that these students have a high capacity to adapt their studies to different learning situations. This may affect the ability to generalize our results to other groups of students.
The concept of learning styles in pedagogical research has been criticised from different perspectives. Massa and Mayer state that there are instruments that correctly distinguish between verbalizers and visualizers. However, data do not provide support for the idea that different instructional methods should be used for these groups [27
]. Others have reviewed the field and conclude that an impact of learning style theory on teaching and learning efficiency is unproven by current empirical work [11
]. However the same authors also recognise that the learning styles theories may still be of importance to pedagogy; personalized education and students' self-awareness (learning to learn).
The results of Cronbach's alpha analysis with a coefficients above 0.7 in all of four dimensions is comparable to others, and indicate that the Swedish translation of the ILS is usable in this study.
One potential bias of our study could be that the definition of a user could be incorrect. To test for this we performed a sensitivity analysis using three different user time cut-offs with the same result. Thus it is not likely that the definition of users have influenced our results.
Another potential limitation is the sample size. The pre-study power analysis used data from a pilot study measuring mean usage time of the web-base course according to learning styles which indicated a need for 8 persons in each group. Based upon this we calculated that a sample size of 60-100 persons was needed to achieve a power of 80% to detect a difference in usage time according to learning style at the p < 0.05 level. A post-hoc analysis with the achieved dichotomous data indicated that 12 persons were needed in each group to achieve 80% power to detect a 50% relative difference in learning style between users and non-users.
In order to preserve statistical power the four dimensions of learning styles were each divided into three categories of learning style (according to Cooke) instead of five (as described by Felder). This modification increased the number of students in each group with a preference for a certain learning style and should thus if anything enhance the possibility to find differences with the actual sample size.
Out of the 123 students, 76% answered the ILS instrument and 73% the general questionnaire. A fairly good proportion of the students thus participated in the study. In the studied group 59% were defined as users. Using data from Ping Pong the number of users of the web-based ECG learning programme among the group not answering the ILS-instrument was only 10. Thus, we believe that possible differences between participants and non-participants in the survey did not affect the results of our study to a substantial degree.
Other factors than learning style may influence the choice to use the web-based ECG learning programme or not. Our experience from discussions with students is that time seems to be an important factor. To spend time to learn to use a fairly new medium, not knowing the effectiveness, may be an obstacle in applying web-based learning. Time for studies without computers exist in different environments, for example on the bus, in bed or in non-computerised areas within the university. The social situation of individual students could affect the time to spend on the Internet as a medium.