01 NRDC Dyslexia 1-88 update - Texthelp
01 NRDC Dyslexia 1-88 update - Texthelp
01 NRDC Dyslexia 1-88 update - Texthelp
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Developmental dyslexia in adults: a research review 165<br />
Screening tests, whether behavioural checklists, cognitive test batteries or hybrid methods,<br />
have one purpose: to reduce the time and expense of diagnosis. In theory, the outcome of a<br />
screening test entails referral of true and false positives and non-referral of true and false<br />
negatives. Tests are therefore required to be sensitive to the condition under investigation by<br />
including as high a proportion of true positives as possible and to be specific to that condition<br />
by excluding as high a proportion of true negatives as possible. In a rule-of-thumb fashion,<br />
the tests serve a purpose, but it is improbable that they do so with equal efficiency. That is to<br />
say, their predictive values are likely to differ, in that one test may yield a higher proportion<br />
than another test of ‘test positives’ who are found to be ‘true positives’.<br />
One complication is that the predictive value of any screening test is unlikely to be a constant.<br />
It may vary strikingly according to the context in which it is used, so that an acceptably high<br />
predictive value for a clinic population (where the condition screened for is relatively common<br />
and where the test might have demonstrated its usefulness) may become an unacceptably<br />
low predictive value when the same test is used in a general population survey (where the<br />
condition screened for is relatively uncommon).<br />
A further complication is that the test might have been normed in a way that makes it<br />
unsuitable for use in an epidemiological study. For example, it might omit to take age or<br />
maturational effects into account where there should be norms for different age-groups. Or it<br />
might omit to take sex differences into account where there should be norms for each sex. Or<br />
it might need variant forms to take cultural or linguistic differences into account. Where timed<br />
tests are used, it might be that norms derived from a high-achieving sample of university<br />
students are unsuitable for use with a sample of low-achieving young adults. This will be the<br />
case where reaction times are assessed, where failure to take the difference into account<br />
would refer an excessive proportion of people whose attainments, however low, could be<br />
expected on the basis of their general ability.<br />
What criteria would a screening instrument need to satisfy in order to offer an adequate way<br />
of assessing the prevalence of dyslexia in an epidemiological study? A counsel of perfection<br />
would require a predictive value so high that the instrument would be, in effect, a diagnostic<br />
test. Unless and until such a test becomes available, a realistic set of criteria must prioritise<br />
optimal predictive value and availability of demographically-adjusted norms. There might be a<br />
trade-off between these two criteria.<br />
If or when a screening instrument has been identified on realistic criteria, a decision can be<br />
made as to whether the data obtained through its use could be entered into any analyses. It is<br />
unlikely that any screened individuals could be identified as dyslexics for the purpose of data<br />
analysis, for the reasons already stated. However, it is possible that self-reported behaviours<br />
or scores on test items could be included in data analyses with no less confidence than<br />
applies to self-report in general.<br />
If the desired unit of analysis is the dyslexic individual, then it is essential to supplement any<br />
screening test data. This might be done in two ways.<br />
If the screening instrument is a behavioural checklist, the conventional wisdom would follow<br />
it with a full-scale diagnostic assessment by an educational psychologist. The feasibility of<br />
this procedure in a large-scale epidemiological survey would first need to be established. It is<br />
a moot point, in any case, whether full-scale diagnostic assessment could succeed in<br />
identifying dyslexic individuals to an acceptable level of accuracy, given the inutility of