However, it is possible to get an indirect indication of the depth of the critical thinking going on in each stage, by relating each indicator to the stage in which it is most expected (see Appendix A). For example, we would expect new problem-related information (NP+) to be introduced in Garrison's stage 2. By mapping our indicators to Garrison's 5 stages, we get an estimate of how deep the learning style is at each stage of critical thinking or problem solving. The depth of critical thinking has been plotted against Garrison's stages, for each seminar group.
Content analysis of Group B's discussions showed an overall increase in the depth of critical thinking across Garrison's stages when using computer conferencing. A similar pattern was found for the other two seminar groups.
But for them, the advantage of computer conferencing was least during the problem exploration phase, stage 3. During problem exploration, the participants should be creatively exploring new ideas. This is a somewhat different task to the structured problem-solving found in the problem identification, problem definition and problem integration stages. Once again, we find that computer conferencing helps the more structured, less creative, parts of critical thinking process. It is in stages 1 and 5 where computer conferencing shows a significant consistent gain over face-to-face seminars, as shown by the difference plot in Figure 12 and confirmed by matched sample t-tests (Table 5).
| G1 | G2 | G3 | G4 | G5 | |
|---|---|---|---|---|---|
| A | 0.18 | 0.19 | -0.36 | 0.15 | 0.43 |
| B | 0.09 | 0.44 | 0.28 | 0.13 | 0.31 |
| C | 0.19 | 0.67 | 0.14 | 0.35 | 0.22 |
| t | 4.6 | 3.1 | 0.1 | 3 | 5.5 |
| t-crit | 4.3 | 4.3 | 4.3 | 4.3 | 4.3 |
| p | 4.4% | 9.0% | 93.0% | 9.8% | 3.2% |
If we now examine the privacy and office automation seminars in these figures, we once again find a common pattern. There is a difference in the depth of learning style adopted in stage 5, the integration of the problem back into the world. There is good evidence for this integration in the office automation discussions, but the privacy discussions seem to have degenerated at this stage. So this looks like an effect of the discussion subject on critical thinking.
To test this, we compared the CT ratios for the seminars in which the same subjects (office automation and privacy) had been discussed. An analysis of variance of these seminars on the same subject shows that all the relationships were significant (see Table 6). There was an overall difference between the depth of learning over the stages, and a significant interaction between the subject and the stage, confirming the impression given in Figures 7-9. Looking in more detail at each stage, none of the differences with subject at different stages are significant at 5% in this small sample, but stages 5, 4 and 1 come closest.
| Source of Variation | SS | df | MS | F | P-value | F crit |
|---|---|---|---|---|---|---|
| Subjects | 0.17 | 1 | 0.17 | 8.69 | 1.50% | 4.96 |
| G1-5 stages | 0.43 | 4 | 0.11 | 5.42 | 1.40% | 3.48 |
| Interaction | 0.37 | 4 | 0.09 | 4.69 | 2.20% | 3.48 |
| Within | 0.2 | 10 | 0.02 | |||
| Total | 1.18 | 19 |
| Variable | Hypoth. MS | Error MS | F | Sig. of F | ETA Square | Power |
|---|---|---|---|---|---|---|
| G1 | 0.014 | 0.002 | 7.1 | 0.12 | 0.88 | 0.34 |
| G2 | 0.030 | 0.065 | 0.5 | 0.69 | 0.31 | 0.07 |
| G3 | 0.007 | 0.009 | 0.7 | 0.58 | 0.42 | 0.08 |
| G4 | 0.021 | 0.003 | 8.4 | 0.11 | 0.89 | 0.38 |
| G5 | 0.210 | 0.021 | 10.1 | 0.09 | 0.91 | 0.43 |
