6 2 Experiment
Haskell 1 Extravert 7 out of 9 p<
Table 6.6 showed that in using Haskell 1, only 2 introverted participants out of 7 revisited the pages that they had already learned, whereas 7 extraverted learners out of 9 revisited the pages for more reading. We applied Fisher’s exact test (χ2 (1) = 3.87),
because of the small sample size, and found that significant difference. It can be seen that the introverted learned better with Haskell 1, which was thoroughly understood with one attempt.
In contrast, the extraverted learner needs to revisit pages more often. This could imply that when the extraverted spend less time on average in reading the materials, they may not fully understand the complexity of the learning material.
Table 6.7. Number of participants who revisited the previous pages In Haskell 2 use
System
Personality type
Number of participants
revisited pages Sig. Introvert 4 out of 7
Haskell 2
Extravert 3 out of 10
p<.01
In contrast, Table 6.7 showed that in using Haskell 2, only 3 extraverted participants out of 10 revisited the pages that they had already learned, whereas 4 introverted learners out of 7 revisited the pages for more reading. A Fisher’s exact test
(χ2 (1) = 1.25) showed the significant personality difference.It suggests that as the
extraverted learners tend to be global learners; they may benefit more from the structure of Haskell 2 materials.
We have identified that matching learner’s personality with the learning material designs might be important in terms of the task performance. One of the important aspects we should also consider is their learning experiences, i.e., how easily they remember what they have learnt. This can be examined by constructing a knowledge structure map (Smith & Riding, 1999), which can represent a deeper insight into participants’ comprehension of the learning materials. At the end of the experiment the participants were asked to draw the structure of what they had learnt from both Haskell 1 and Haskell 2. They made this as detailed as they could. The marking strategy was based on theGroup Embedded Figures Test (GEFT; Oh & Lim, 2005). The criterion used to measure the drawings was how many levels they used to draw the course structure. Simply, the more levels descriptions they drew, the more likely they have global understanding of the contents. Thus, if the participants only manage to describe one level of the structure, they are classified as weak performers. If they described two levels, they are thought of as on-average performers; otherwise they are good
Figure 6.5. Knowledge structure map: An example of weak performance (a) and good performance (b)
(a) a weak performer example (b) a good performer example
Table 6.8. Cognitive map for Haskell 1/Haskell 2
Number of participants System Personality
type
Weak performance Good performance
Sig. Introvert 2 5 Haskell 1 Extravert 7 2 p<.05 Introvert 5 2 Haskell 2 Extravert 2 8 p<.05
Table 6.8 showed clearly that the introverted outperformed the extraverted in Haskell 1. Two out of seven introverted participants were weak performers, whereas, seven out of nine extraverted were weak performers. In contrast, Haskell 2 is for the extraverted. Fisher’s exact tests supported these accounts.
These results also supported our assumption that the learners may perform better if they can employ the learning material matched to their own personality type in the learning process.
6.3. General Conclusions and Discussion
The assumption of this study was that the learner’s cognitive style may significantly influence their preferences for a particular learning material design. The findings from this study indicated that the task performances by the two different personality groups (introverted and extraverted) were significantly affected by the two different material designs. That is, the introverted with Haskell 1 outperformed the extraverted with Haskell 1. As opposed to this, the extraverted with Haskell 2 outperformed the introverted. These findings strongly indicated that the personality type could be an influential indicator of learning performance when learners were being taught by different learning strategies.
Both the Haskell 1 and Haskell 2 use cases revealed that introverted were interested in detailed understanding, concentrating on separate topics, which leads to taking a longer time to read materials. In contrast, the extraverted, according to their personality, tended to adopt a global approach to learning, concentrating on building a conceptual overview and fitting in the detail subsequently.
This understanding of the relationship between the personality type and the learning material structure is not new (e.g., Riding & Fanning, 1998; Riding & Rayner, 1999). However, the contribution of this chapter is to empirically identify this
relationship for the design of adaptive e-learning systems, which has not been shown before. The approach to encompassing personality in the design of structuring the contents is new, in these experiments, which clearly demonstrated that different learners may process the learning material using different strategies.
This study thus implies that the user model in adaptive e-learning system should accommodate learners’ different learning styles. For instance, for the introverted, it may be of great use to present more in-depth knowledge before global or associative
knowledge. This would ensure that any adaptive e-learning experience had a spread of activities that would appeal to a range of personalities.
Even though this empirical study showed that the personality type affected the learning process, there are some limitations to generalising these results to the design materials. Firstly, the number of participants was small, so they may not be
representative of a whole population. Secondly, the contents used in this experiment were personal and individual learning with computers rather than collaborative
understanding, which has been paid more attention in recent e-learning systems design. The next chapter addresses this collaborative learning experience in order to see the relationship between personality and collaborative work in designing adaptive e- learning systems. Also the sample size issue will be discussed in Chapter 9 which conservatively limits the interpretation of the thesis.