Report Rate and Length
Responses of children with PDD-NOS and HFA were not significantly different, and therefore were combined in the analyses. Three comparison children and seven children with HFASD reported not to know at all how they would describe themselves. The frequency of these non-responses did not differ between HFASD children and comparisons (χ
2 = 1.81, p > 0.05).
Equal numbers of words were used by children with HFASD (M = 59.7, SD = 39.7) and comparisons (M = 56.0, SD = 34.3), F(1, 84) = 0.43, ns, and there were no significant differences between the baseline and self-promotion conditions within the autism group (M = 32.1, SD = 26.3 and M = 27.6, SD = 20.4, respectively; F < 1) or the typically developing group (M = 29.6, SD = 19.0 and M = 24.8, SD = 22.0, respectively; F < 1).
Valence of Self-Statements
Table
2 shows the valence of the self-statements for the baseline and self-promotion condition. A 2 (Group: HFASD and comparison) × 2 (Condition: baseline and self-promotion) × 3 (Valence: positive, neutral and negative) analysis of variance indicated no main effect for Group,
F(1, 84) = 0.00,
p > 0.05. A main effect was found for Condition,
F(1, 84) = 23.18,
p < 0.01, indicating that the overall mean number of self-statements was lower in the self-promotion condition than in the baseline condition. Furthermore, effects were found for Valence,
F(2, 168) = 62.23,
p < 0.001, Group × Valence,
F(2, 168) = 7.25,
p < 0.001, Condition × Valence,
F(1, 84) = 15.39,
p < 0.001 and Group × Valence × Condition,
F(2, 168) = 3.52,
p < 0.05.
Table 2
Mean values (SD) of positive, neutral, negative and other self-descriptions in both conditions (range: 0–∞)
Base-line condition | Comparisona
| 4.18 (2.22) | 1.84 (1.68) | 1.58 (1.37) | 0.19 (0.45) |
HFASDa
| 3.30 (2.24) | 0.88 (1.26) | 2.26 (1.71) | 0.16 (0.43) |
Self-promotion condition | Comparisona
| 2.61 (1.38) | 1.49 (1.33) | 0.61 (0.82) | 0.02 (0.15) |
HFASDa
| 2.56 (1.76) | 1.44 (1.52) | 0.93 (1.08) | 0.07 (0.26) |
To elucidate the nature of the critical three-way interaction, we tested the simple effect of Group × Valence within each Condition. The simple effect of Group × Valence was significant for the baseline condition, F(2, 168) = 8.40, p < 0.001, but not for the self-promotion condition, F < 1. The contrast between results for the different conditions is due to the fact that the HFASD children provided significantly fewer positive self-statements [t(84) = 2.99, p < 0.01] and significantly more neutral self-statements [t(84) = 2.17, p < 0.05] than the comparison group in the baseline condition, representing medium sized effects (r = 0.31 and 0.21, respectively). No significant group differences were observed in the self-promotion condition (all ps > 0.10).
Furthermore, paired analyses within both groups showed that the number of positive self-statements increased in the HFASD group, t(42) = 1.95, p < 0.06, r = 0.29, but not in the typically developing group t(42) = 1.21, ns, r = 0.18. Given the absence of group differences in the self-promotion condition, as noted above, these results indicate that the experimental manipulation brought the number of positive self-statements of the HFASD group in line with that of the comparison group.
Goal-Directedness of Positive Self-Statements During Self-Promotion
As noted earlier, positive self-statements in the self-promotion condition were further categorized according to their goal-directedness as game related or not-game-related (Table
3). A 2 (Group: comparison and HFASD) × 2 (Goal-directedness: game-related and not-game-related) MANOVA showed no main effects for Group or Goal-directedness but an interaction effect was found for Group × Goal-directedness,
F(1, 84) = 6.72,
p < 0.05. Although children with HFASD did sometimes report game-related features, they did so less often than typically developing children
t(84) = 2.17,
p < 0.05.
r = 0.23. Furthermore, it was of particular interest to see that children with HFASD included very similar numbers of game-related and not-game-related self-statements in the self-promotion condition,
t(42) = 0.72,
ns, whereas comparison children seemed to focus specifically on game-related features
t(42) = 3.36,
p < 0.005,
r = 0.46.
Table 3
Goal directedness of positive self-statements in the self-promotion condition (range: 0–∞)
Comparisona
| 1.07 (1.08) | 0.41 (0.73) |
HFASDa
| 0.63 (0.78) | 0.81 (1.40) |
In addition to being matched on age and IQ, children with HFASD and comparisons performed similarly on the second order false belief task (percentage passing, 0.58 vs. 0.72, respectively), χ
2
(1) = 1.84, p > 0.10. Correspondingly, when age, IQ and false belief performance were included as covariates, the key findings regarding the valence and the goal-directedness of children’s self-statements remained the same (three-way interaction of Condition × Group × Valence, F(2, 162) = 4.93, p < 0.01; two-way interaction of Group × Goal-directedness, F(1, 81) = 9.45, p < 0.005). Moreover, in both the HFASD and the comparison group correlations between IQ scores and any of the dependent variables were absent.