CHAPTER 4: RESULTS
1.1 Introduction
4.1.3 ADOS sub-category analyses
This analysis employed the same values as the previous equation except for the value
𝛇
1= -.10, which was employed in the numerator as we were evaluating the left boundary of equivalence.Z =
√Z =
√
=
= - 1.30065
Since the obtained value Z = 1.3 was less than the tabled critical one-tailed Z.05 = 1.65, the null hypothesis (Ho: θ ≤ - 0.10) designating non-equivalence, could not be rejected. Thus they were equivalent within the designated margin of error. The slight variations between groups could be due to chance variations also. If the correction for continuity was made once again, the results would not have affected the outcome.
So, the second of the null hypotheses designating non-equivalence cannot be rejected. Thus, since both null hypotheses cannot be rejected, we conclude that the ADOS diagnoses are equivalent to the clinician diagnoses. The lower range of the 90% confidence interval falls below the minimum tolerable difference between the diagnoses of the two groups. Both null hypotheses would have had to have been statistically significant at the 0.05 level, and if the latter had occurred, the lower limit of the confidence interval would have been some value greater than 0.10 and -0.10 for the tests to be non-equivalent. The results therefore support the main null hypothesis of this study.
Thus, each ethnic group was compared on each algorithm to note whether diagnosis on the ADOS compares across ethnic lines.
An analysis of variance (ANOVA) was run using SPSS (version 19) to consider whether significant effects existed between ethnic groups regarding their responses on the ADOS algorithm scores. This statistical procedure was used to answer the following hypotheses:
Ho1: There is no significant difference between children of White, Black or Indian cultures with ASD and those with other special needs based on the sub-group coding for communicative behaviours on the ADOS.
Ha1: There is a significant difference between children of White, Black or Indian
cultures with ASD and those with other special needs based on the sub-group coding for communicative behaviours on the ADOS.
Ho2: There is no significant difference between children of White, Black or Indian cultures with ASD and those with other special needs based on the sub-group coding for Reciprocal Social Interaction behaviours on the ADOS.
Ha2: There is a significant difference between children of Black, White or Indian
cultures with ASD and those with other special needs based on the sub-group coding for Reciprocal Social Interaction behaviours on the ADOS.
Ho3: There is no significant difference between children of Black, White or Indian cultures with ASD and those with other special needs based on the sub-group coding for Imagination/Creativity behaviours on the ADOS.
Ha3: There is a significant difference between children of Black, White or Indian
cultures with ASD and those with other special needs based on the sub-group coding for Imagination/Creativity behaviours on the ADOS.
Ho4: There is no significant difference between the children of Black, White or Indian
cultures with ASD and those with other special needs based on the sub-group coding for Stereotyped Behaviours and Restricted Interests on the ADOS.
Ha4: There is a significant difference between the children of Black, White or Indian cultures with ASD and those with other special needs based on the sub-group coding for Stereotyped Behaviours and Restricted Interests on the ADOS.
Table 4.
Homogeneity of variance
Test of Homogeneity of Variances
Levene Statistic df1 df2 Sig.
Communication .302 2 23 .743
Reciprocal Social Interaction
.699 2 23 .507
Imagination/Creativity 1.281 2 23 .297
Stereotypical
Behaviours/Restricted Interests
.961 2 23 .397
In the Levene’s Test above, the F-value is not significant since all values are greater than p = 0.05. Thus, suggesting that we cannot reject the null hypothesis of equality of variance. Thus, the assumption of homogeneity has not been violated.
Table 5.
ANOVA summary table for analysis of data
ANOVA
Sum of Squares df Mean Square F Sig.
Communication Between Groups 31.405 2 15.703 3.607 .043
Within Groups 100.133 23 4.354
Total 131.538 25
Reciprocal Social Interaction
Between Groups 106.128 2 53.064 6.621 .005
Within Groups 184.333 23 8.014
Total 290.462 25
Imagination/Cre ativity
Between Groups 5.785 2 2.892 2.247 .128
Within Groups 29.600 23 1.287
Total 35.385 25
Stereotypical Behaviours/Res tricted Interests
Between Groups .328 2 .164 .134 .875
Within Groups 28.133 23 1.223
Total 28.462 25
In studying the summary table, significant effects were found. It was noted that the source of variance in the data was between ethnic groups and lay in the sub-categories of communication and reciprocal social interaction. Since their significance values were p = .043 and p= .005 respectively. These values were less than the alpha level of p= 0.05 indicating statistical effects in the data.
Further multiple comparisons in the form of post hoc tests needed to be run so as to identify where the precise difference between means lay (Durrheim, 2002). This study used Tukey’s Honestly Significantly Difference test (HSD), since it was the test used most widely in the Social Sciences (Durrheim, 2002).
Tukeys HSD tests (Table 6.) of multiple comparisons confirmed that the significant differences were found between groups in the communication and reciprocal social interaction sub-categories. More specifically, differences were noted between the Black and White/Indian ethnic groups. For the communication sub-category the significance value was 0.034 at the significance level of p = 0.05. For Reciprocal Social Interaction the significance value was 0.004 at the significance level of p = 0.05. The other sub-categories of imagination/creativity and stereotyped behaviours/restricted interests had significance scores greater than p = 0.05, thus no significant differences in behavioural responses on the ADOS were found between Black, White and Indian participants for the sub-categories of imagination/creativity and stereotypical behaviours/restricted interests.
Table 6.
Multiple comparisons using Tukey’s (HSD) test results
Multiple Comparisons Tukey HSD
Dependent Variable (I) Ethnicity (J) Ethnicit y
Mean Differen
ce (I-J)
Std.
Error Sig.
95% Confidence Interval Lower Bound
Upper Bound
Communication Black White 2.500* .933 .034 .16 4.84
Indian 1.067 1.077 .590 -1.63 3.77
White Black -2.500* .933 .034 -4.84 -.16
Indian -1.433 1.077 .393 -4.13 1.27
Indian Black -1.067 1.077 .590 -3.77 1.63
White 1.433 1.077 .393 -1.27 4.13
Reciprocal Social Interaction Black White 4.600* 1.266 .004 1.43 7.77 Indian 2.033 1.462 .362 -1.63 5.69 White Black -4.600* 1.266 .004 -7.77 -1.43 Indian -2.567 1.462 .207 -6.23 1.09
Indian Black -2.033 1.462 .362 -5.69 1.63
White 2.567 1.462 .207 -1.09 6.23
Imagination/Creativity Black White .700 .507 .368 -.57 1.97
Indian -.500 .586 .674 -1.97 .97
White Black -.700 .507 .368 -1.97 .57
Indian -1.200 .586 .123 -2.67 .27
Indian Black .500 .586 .674 -.97 1.97
White 1.200 .586 .123 -.27 2.67
Stereotypical
Behaviours/Restricted Interests
Black White -.200 .495 .914 -1.44 1.04
Indian -.267 .571 .887 -1.70 1.16
White Black .200 .495 .914 -1.04 1.44
Indian -.067 .571 .993 -1.50 1.36
Indian Black .267 .571 .887 -1.16 1.70
White .067 .571 .993 -1.36 1.50
*. The mean difference is significant at the 0.05 level.