Analysis of second data set.
The results of the analysis of the first set of data clearly showed that learning styles did not differentiate between the students perceptions of lectures and learning in the experimental lectures. This cast doubt on the construct validity of Kolb’s learning style measure. ‘Does it really measure the difference in students approach to the learning environment’? It was felt that further study was
necessary to gain insight in this centrally important question concerning the validity of Kolb’s learning styles.
A series of measures were carried out during an additional phase of data
collection, each of which addressed a question generated by the first data set. Due to the fact that the additional phase of the study was not originally designed into the project it was not possible to gain access to all of the original degree course groups of students used during the first stage. The statistics and psychology students for example, had broken into options groups and did not meet as a whole. The nursing students still met as a whole group and were used for this additional data gathering. As the first phase of the analysis showed that there was no significant differences in the learning styles or ratings of the different degree groups it was felt that the nurses could be used as a representative sample of the original population.
The additional phase of the study presented the students with four measures: • Rotter’s measure of locus of control
• A rank order questionnaire
• Second administration of Kolb’s learning style inventory
Rotter’s locus of control was used to discover if it was possible that some other form of classifying the students would allow distinctions in perceptions and ratings of the lectures to develop, for example it could be that internal and external control groups from Rotter would allow distinctions in approach to
comparative choice between the lectures. As the original design of the study was, it was possible that the students were rating the lectures on an ‘of its types basis’. That is, they may not personally have found the lecture useful and perceiving they did not learn particularly well, but of its generic type the presented lecture was a good example and they learnt well considering the limitations. The forced choice rank ordering would over come this difficulty and any concealed differences in attitudes towards the lectures would be brought out. Finally, Kolb’s learning style inventory was re-administered, this was done to discover how stable the measure was across time.
Learning Styles Across Time.
Summary Statistics for first and second learning style assignment
L.S..1 Count Percent L.S..2 Count Percent
Diverger - 1 1 6 .25 1 1 6.25 Assimilator - 2 3 18.75 2 1 6.25 Converger - 3 5 31.25 3 1 6.25 Accomodator - 4 4 25 .00 4 1 6.25 Fifth style - 5 3 18 .75 5 12 75.00 N= 16 N= 16
L.S.l - first measured learning style L.S.2 - second measured learning style
From the above summary tables it can be seen that between the first and second assessment of Kolb’s learning styles there has been a shift in the distribution of the learning styles, particularly towards learning style five. So dramatic is the shift to learning style that only four of the sixteen subjects are not learning style five in the second assessment.
Sign test for first learning style rating with second learning style rating.
Cases
0 - Diffs (LS2 LT LSI)
11 + Diffs (LS2 GT LSI) (Binomial)
5 Ties 2-tailed p = 0.0010
Critical value of s = 1 (N=ll)
16 P<=0.05 (two tailed)
Key - LS2 - Second learning style rating LSI - First learning style rating LT - Less than
GT - Greater than
As the observed value of ‘s’ is less than the critical value it can be concluded that there is a significant difference in the first and second learning styles. In summary this means there is no association between the first and second measure learning style. If the learning style inventory is reliably testing different approaches to learning, then the differences must be state rather than trait dependant, as the differences have fluctuated over a short period of time. Indicating that what is being measure is an approach at any one moment in time, rather than underlying features which effect students approach to learning in all conditions and across time. Hence, if Kolb’s LSI is to be used effectively it can no longer be used to note association between degree course and learning style as learning style have been shown to be readily flexible.
N.B. The Binomial Sign test was used because the learning style ratings represent catergorial or nominal data. As the first and second learning style rating are for a single individual the design is repeated measures. Hence the most appropraite test of whether there is a significant difference between the first and second learning style rating is a binomial sign test.
Comparison of the first TLSIS~ and second TLSIS
Hj - There will be a significant difference in the TLSIS between first and second assessment.
t-tests for paired samples comparing mean of first TLSIS with second TLSIS
Number of 2-tail
Variable pairs Corr Sig Mean SD SE of Mean
TOTAL1 63.1250 2.419 .605
16 .370 .158
TOTAL2 62.5625 3.864 .966
19 Total learning style inventory score (TLSIS) is established by adding the scores from each of the four learning mode scores together. The learning mode scores are gained from the LSI, which has nine sets of four words, each set has a word which reflects each learning mode. However, Kolb only uses the six most associated words (gained via factor analysis) to establish the learning mode scores. The six words used vary for each learning mode - reflective observation uses words from sets; 2, 3, 4, 5, 7 and 8, whereas concrete experience uses words 1, 3, 6, 7, 8, 9. Hence for
Paired Differences
Mean SD SE of Mean t-value df 2-tail Sig
.5625 3.723 .931 .60 15 .555
95% Cl (-1.422, 2.547)
critical value of t = 2.131 P<= 0.05(two-tailed)
No significant difference was found between the first and second TLSIS mean hence the analysis failed to be reject the null hypothesis.
In conjunction with the analysis which shows that the measured learning style has changed between the two assessments, it is interesting to discover that the TLSIS has not significantly changed. That is, the sum of the learning modes, from which the learning styles are derived, have not fluctuated between the measures. The implications of this are that the distribution of the TLSIS shifts, hence causing different learning style but does not significant differ in total. It would appear therefore that students will be more ‘flexible’ in their approach than any one learning style will give them credit for. Yet students will be limited by an overall TLSIS which must be distributed between the learning modes. This TLSIS is relatively more stable than Kolb’s learning styles, but it is not argued that it is fixed, as this would undermine any point in education if overall ability really were fixed.
Learning styles and forced choice rank ordering of lectures.
If the original notion that attuned lectures would be rated more positively than none attuned lectures is kept, then learning style groups should rank their attuned lectures more highly (1 being highest, 4 being lowest) than the none attuned lectures.