Potential discriminative factors for dyslexia: A predictive statistical model based on the Pavlidis questionnaire distinguishing 8-9 year-old dyslexic and non-dyslexic-control Greek children. Validity's and potentialpredictive efficiency's considerations potential predictive efficiency's considerations

Abstract

The aim of the thesis was to identify the most significant early signs of dyslexia, independent of reading and to use them to differentiate between diagnosed dyslexic and non-dyslexic-control Greek children, 8-9 years old. Many authors (e.g., Nicolson & Fawcett, 1996; Muter, 1996; Singleton, et al. 1995; Grogan, 1995; Wenner, 1995; Hurford et al. 1994; Coleman & Dover, 1993) highlighted the importance and advantage of early identification, in terms of human and financial resources. Educationally, most teachers acknowledge the greater ease of working with younger children who have not yet experienced excessive frustration and feelings of failure. The findings of this thesis noted that could screen the dyslexics from non-dyslexics on the basis of a non-reading procedure, based on the parent-reported information for their children. The information was originated from a quick, economical, easy to administer, sensitive checklist related to the developmental history; laterality; sequential problems; behaviour and personal traits; ADD characteristics and family's history (heredity). This checklist, named Pavlidis Checklist could be used as screening tool for dyslexia. It could correctly discriminate the dyslexics with 95.2% accuracy, the non-dyslexics-controls with 97% and the overall accuracy was 96.3%. The validity of this predictive tool was tested using the comparison of the discrimination rate between this predictive tool and the spelling errors of dyslexics and non-dyslexics. Spelling errors could discriminate these two groups, because it is widely accepted (Hornsby, 1995; Megalokonomos 1983) that they are (spelling errors) a part of the diagnostic procedure for dyslexia. The results showed that this tool had the similar, very high, overall discriminative accuracy with the spelling errors, so, it was accepted as valid. Moreover, the within-test consistency of this checklist was very high too and also it seemed to be potential predictive efficient.