24-item Academic Behavioural Confidence Scale and my Factor Analysis outputs

As reported elsewhere on the project webpages, Principal Component Analysis has been executed on the data collected in this project for both of the principle variables: Dyslexia Index and Academic Behavioural Confidence.

In using Sander & Sander’s (2003) original 24-item ABC Scale, a basic analysis was conducted which determined that there exists both a moderate effect size and a statistically significant difference between the mean ABC of the subgroup of students with identified dyslexia and the subgroup of students who dyslexia-like profiles had not be previously identified, according to their self-disclosure in the project questionnaire.

However, the most recent inspection of this data identified a fresh ‘angle’ that can be taken.

To date, all analysis has relied on the subgrouping of the datasets into two distinct subgroups: students who report their dyslexia (research subgroup DI), and students who do not report any dyslexia (research subgroup ND). As has been presented in many posts published throughout this StudyBlog, it is finding students with dyslexia-like profiles from amongst the datasets of students who do not report any dyslexia that is of particular interest because it is these students whose academic confidence is being explored in relation to the academic confidence of both the declared dyslexics and the most definitely ‘not dyslexic’ (as identified through this project’s Dyslexia Index Profiler).

It occurred to me that there may be value in looking at the complete research datapool in a different way. This has been to consider the complete datapool as a whole and differentiate those with dyslexia or dyslexia-like profiles purely on the basis of the score returned in the Dyslexia Index Profiler that was included as a major part of the research questionnaire. In other words, disregard whether any particular respondent had declared their dyslexia or declared no dyslexia but asign that dataset into a research subgroup entirely on the basis of the value returned by the Dyslexia Index Profiler. This meant that two alternative research subgroups have been established that nevertheless still apply the same discriminating principle established by the Dyslexia Index Profile value of Dx = 592.5. It is this value, Dx = 592.5, which has been used as the discriminator in research subgroup ND to identify students who have not declared any dyslexia but who nonetheless appear to be presenting a dyslexia-like profile. Recall that this distinct subgroup has been designated research subgroup DNI (dyslexia not identified).

From this, a fresh look at comparisons between the research subgroups’ academic confidence might declare alternative results that might be helpful in understanding differences observed.

The table below sets out these comparisons with results of particular interest highlighted in blue/green.

S&S ABC 24-item scale, Dykes 5-factors

S&S ABC 24-item scale overall results

study efficacy academic engagement academic output attendance debating
Complete dataset sorted according to Dyslexia Index (Dx) score and independent of dyslexia disclosure
sample means data subgroup Dx < 592.5
66.09 60.77 59.85 73.93 79.71 66.10
sample means data subgroup Dx > 592.5
59.84 57.33 48.97 62.18 82.14 67.35
sample standard deviations ~ Dx < 592.5
14.89 20.60 18.09 16.19 21.09 22.65
sample standard deviations ~ Dx > 592.5

14.77 20.13 19.62 20.42 16.87 21.75
sample size ~ Dx <592.5
n = 101
sample size ~Dx > 592.5
n = 65
TTest p value between means Dx < 592.5 and Dx > 592.5
p = 0.0030, sig at 1% 0.0899 not sig 0.0002 sig at 0.1% 0.000014 sig at 0.1% 0.3256 not sig 0.2394 not sig
Hedges ”g’ effect size between Dx < 592.5 and Dx > 592.5
g = 0.42 g = 0.17 g = 0.58 g = 0.65 g = -0.12 g = -0.06
Complete dataset sorted according dyslexia disclosure: ND=no dyslexia, DI=dyslexia disclosed
sample means data subgroup ND
67.21 62.52 61.05 73.87 80.85 68.09
sample means data subgroup DI
58.45 54.96 47.72 62.78 80.39 64.43
sample standard deviations ND
13.85 17.82 18.06 16.43 19.18 21.32
sample standard deviations DI

15.45 23.10 18.63 20.17 20.15 23.51
sample size ND
98
sample size DI
68
TTest p value between means ND <-> DI
0.00014, sig at 0.1% 0.0111 sig at 5% 0.000010 sig at 0.1% 0.000084 sig at 0.1% 0.3987 not sig 0.1714 not sig
Hedges ”g’ effect size between ND <-> DI
g = 0.60 g = 0.38 g = 0.73 g = 0.61 g = 0.02 g = 0.16

More to report here

Leave a Reply