Difficulty Ratings of Tooth Extraction Cases
The percentage of each group reporting the extraction in each case as easy or difficult, are shown in Table 2.
Table 2
Percentage of ratings of each case by Stage
| | BDS4 | BDS5 | Intern |
Case | Grade of Extraction | Difficult | Easy | Difficult | Easy | Difficult | Easy |
Case1 | I | 20.25 | 79.75 | 8.82 | 91.18 | 3.85 | 96.15 |
Case2 | II | 41.77 | 58.23 | 16.18 | 83.82 | 7.69 | 92.31 |
Case3 | I | 69.62 | 30.38 | 73.53 | 26.47 | 50.00 | 50.00 |
Case4 | II | 55.70 | 44.30 | 41.18 | 58.82 | 36.54 | 63.46 |
Case5 | IV | 94.94 | 5.06 | 97.06 | 2.94 | 94.23 | 5.77 |
Case6 | III | 54.43 | 45.57 | 60.29 | 39.71 | 50.00 | 50.00 |
Case7 | III | 49.37 | 50.63 | 57.35 | 42.65 | 48.08 | 51.92 |
Case8 | IV | 70.89 | 29.11 | 80.88 | 19.12 | 84.62 | 15.38 |
The effect of Grade of Extraction (the expert rating of cases), and Stage on difficulty ratings was assessed using a mixed 3 Stage (BDS4, BDS5, Intern) x 4 (Grades I-IV) ANOVA, with response (Difficult = 1, Easy = 0) as the dependent variable. The resultant mean for each cell of the design represents the percentage of respondents who perceived the cases as difficult in that category, and the ANOVA compares the differences in these means between groups.
The results showed that there was a main effect of Stage (F2,196=4.31, p = 0.015, ƞ2g=0.015), a main effect of Grade of extraction (F2.72, 533.84=115.57, p < 0.001, ƞ2g=0.279), and a significant interaction between the two (F5.45, 533.84=4.76, p < 0.001, ƞ2g=0.031); as depicted in Fig. 1.
Pairwise comparisons (Tukey HSD) showed that Grade 1 extractions, only the BDS4 and Intern years differ significantly (p = 0.030); for Grade II extractions, only the BDS4 and Intern years differ significantly (p = 0.004); and on Grade III and IV extractions, no differences were seen by stage.
BDS4 respondents’ judgements of the difficulty of cases suitable for Grade I, II and III extractions are comparable, but all of these judgements are significantly lower than their judgements of Grade IV extractions.
BDS5 respondents’ judgements of the difficulty of cases suitable for Grade I, II and III extractions are all significantly lower than their judgements of Grade IV extractions, and in addition, Grade III extractions were reported as more difficult than cases suitable for Grade II.
Intern respondents’ judgements show the same pattern as BDS5 respondents; judgements of the difficulty of Grade I, II, and III extractions are all significantly lower than their judgements of complicated cases, and in addition, Grade III extractions were seen as more difficult than Grade II extractions.
These patterns are found across both male and female respondents. Side-by-side plots are shown in Fig. 2. Statistically, the ANOVA described above, with gender factored in, showed a significant main effect of Gender (F1, 193=4.08, p = 0.045, ƞ2g=0.007), with females categorizing significantly more cases (54.60%) as difficult on average than males (48.00%), but no significant interaction effects with either Stage (p = .601), Grade of extraction (p = 0.065), or the two in combination (p = 0.186), i.e. the patterns shown in Fig. 1 and both of those shown in Fig. 2 are statistically comparable.
Self-reported Confidence Ratings
The percentage of each group reporting the perceived confidence for performing each extraction, by Stage, are depicted graphically in Fig. 3.
Correspondence analysis of a three-way contingency table (counts of each confidence-level response by Stage by Expert rating of cases) suggests a statistically significant association between the three factors (Χ2df=28 = 853.725, p < 0.001).
Chi-Squared tests of association between stage and response are significant for Grade I (Χ2df=4 = 17.497, p = 0.002), Grade II (Χ2df=4 = 16.517, p = 0.002), Grade III (Χ2df=4 = 14.916, p = 0.005), and Grade IV (Χ2df=4 = 22.703, p < 0.001) extractions.
Similarly, collapsing across stage and testing for associations between responses and expert rating of cases showed a significant relationship between the two (Χ2df=6 = 761.33, p < 0.001), and these are found within all permutations of the contingency tables, e.g., whether comparing responses between Grade I and II extractions, Grade II and III extractions, Grade I and IV extractions etc. All of these relationships hold across male and female respondents; a summary of responses for each are shown in Fig. 4.