200 likes | 966 Views
A Validation Of The Dyslexia Adult Screening Test (DAST) in A Post Secondary Population. Allyson G. Harrison, Ph.D, C.Psych. Dyslexia Adult Screening Test DAST. Developed by Fawcett & Nicholson in UK Published by Psych corp 1998
E N D
A Validation Of The Dyslexia Adult Screening Test (DAST) in A Post Secondary Population Allyson G. Harrison, Ph.D, C.Psych
Dyslexia Adult Screening Test DAST • Developed by Fawcett & Nicholson in UK • Published by Psych corp 1998 • Normative data collected on 550 “normal” students & 618 adults (age 17-65). • ? # Dyslexic subjects-No reference at all. • At Risk Quotient (ARQ) calculated based on performance on 11 subtests
DAST cut-offs • ARQ of .7 or more = slightly at risk • ARQ of 1 or more = highly at risk Fawcett & Nicholson used the cut off of 1.0 in their normative study!
Problems with initial norms • “Dyslexic Student” data consists of only 15 people • Dyslexia validated by ADI: ACID pattern on WAIS; spelling, nonsense passage reading & previous hx dyslexia. • ARQ calculated dividing by 9 instead of 11 • How well can DAST accurately identify Dyslexic students based on this limited sample?
Current study • LOTF project in Ontario, Canada • Improve services & supports for “Dyslexic” students in post-secondary. • Pilot students had to meet rigorous diagnostic criteria for inclusion: • 2 std dev difference between measure of intellectual potential & achievement &/or specific information processing skill, + consistent history
Subjects • 117 “well validated” Pilot students • 122 volunteer controls • Sex ratio equal in Pilot students, but 75% of controls were female. • Mean age of two groups equal
Results Using .7 as cut off (mild risk): • 85% of Dyslexics correctly identified • 15% missed. 3 subjects had ARQ < 0.01 • 25% “controls” identified as mild risk Using 1.0 (high risk) as cut off: • 74% Dyslexics correctly identified • 15% controls identified as high risk
Information about control subjects • Recruited from first-year courses, posters, and work-study student population • Completed self-rating scales & DAST • Correlation between self-rated reading pleasure and ARQ=.40 • Correlation between self-rated reading skills and ARQ=.52
Relationship Between Pleasure from Reading and ARQ (Control gp only) 3.00 2.50 2.00 2 R = 0.1585 1.50 At Risk Quotient (ARQ) Highly at risk cut off 1.00 0.50 0.00 1 2 3 4 5 Self-rated reading enjoyment (1=very pleasurable; 5=no pleasure)
Relationship Between Self-rated Reading Skills and ARQ (control gp only) 3.00 2.50 2.00 2 R = 0.2722 1.50 At Risk Quotient (ARQ) Highly at risk cut off 1.00 0.50 0.00 1 2 3 4 5 Self-rated Reading skill (1=strongest; 5=poorest)
ARQ scores for control subjects who reported reading problems ARQ 0.20 0.36 0.63 0.27 0.36 0.18 0.82 0.82 0.90 1.45 1.09
Recalculation of DAST • Remove postural stability (least consistent & lowest inter-rater agreement) • Remove subtests with largest group overlap • Remove subjects with NVLD • Recalculate ARQ based on 7 subtests
Recalculated DAST (Excluding subtests 3, 8 & 11 + NVLD) Using .7 as cut off (mild risk): • 88 % of Dyslexics correctly identified • 12 % missed. • 27 % “controls” identified as mild risk Using 1.0 (high risk) as cut off: • 77 % Dyslexics correctly identified • 17 % controls identified as high risk
Conclusions • DAST in present form is not acceptable as screening for LD • Good screening test should identify almost ALL of true Dyslexic subjects. This does not. • Removal of subtests with questionable discriminate validity improves hit rate slightly, but still misses 12% of Dyslexic students • Relationship between ARQ & criterion variables (such as self-rated reading skill) an issue
Suggestions • Investigation of IQ-ARQ correlation • Establish criterion validity of ARQ & subtests in non-disabled control group • Recalculation of normative scores and cut offs using larger Dyslexic sample. • Don’t throw the baby out with the….