## The Bangor Dyslexia Test II

Combining a clinical approach with a statistical one was not always easy. At an early stage in the research I presented a paper to my colleagues with the title 'How Do I Score the "crikey?" ' The situation which I envisaged was one in which a subject, presented, for instance, with a request to say the months of the year, responded, 'Oh, crikey!' - and then proceeded to say the months of the year correctly. Was I to ignore the 'crikey' on the grounds that it was a one-off remark which could not be quantified? On the other hand, if I simply ignored the 'crikey', I would clearly be throwing away useful information. I did not at the time produce any answer to this question. However, I now think that a suitable answer would be to specify that, for instance, 'any exclamation of dismay, puzzlement etc.' should be recorded and counted.

I suggested in the last chapter that clinical judgements were based on the use of a combination of cues, sometimes small ones, which required to be pieced together. This, of course, is what happens in any medical diagnosis. If I could operationalise these cues, I would be in a position to make a judgement as to whether a particular indicator of dyslexia was present or absent and quantify the numbers of each. In the event I found that I would sometimes be presented with a response which I regarded as marginal - not clearly indicative of the presence of dyslexia but also not clearly indicative of its absence. I therefore devised the following notation: a clearly dyslexia-positive response would be scored as 'plus', a marginal response as 'zero', and a dyslexia-negative response as 'minus'.

If the subject answered the question with no difficulty, hesitation or special strategy, this qualified as 'minus'. If there were hesitations and other signs of uncertainty or if the subject needed a special strategy in order to work out the answer, the response was scored as 'zero', while, if I judged a response to be typically characteristic of dyslexics, it was scored as 'plus'. When totting up the number of dyslexic indicators, I decided to score a 'zero' response as half a 'plus', which meant that two 'zeros' were equivalent to one 'plus'. If there was a single hesitation, this was regarded as insignificant, but if there were two or three hesitations this was scored as a 'zero'. One of the decisions which I had to make in scoring the test was whether there were, in conjunction, enough positive signs to be significant.

A complete list of what responses should be scored as 'plus', 'zero' and 'minus' will be found in the test manual (Miles, 1997). Examples of 'plus' responses included turning in one's seat in order to work out the tester's left and right sides, asking if the months of the year needed to be said in order, losing the place during the recitation of tables ('Was it six sevens I was up to?') and making a mistake in trying to repeat, say, five digits but repeating six correctly.

'Zero' responses included more than a single hesitation or request for the question to be repeated and 'slips' in the reciting of tables, for example by saying 'eight eighties' instead of 'eight eights'. The result was also scored as zero if the subject responded correctly but only as the result of a special compensatory strategy, for instance being able to show his or her right hand correctly because it is 'the hand I write with'. It seemed to me that hesitations over the Left-Right item might indicate a general uncertainty, not over direction, as I first thought, but over which label, 'left' or 'right', was the correct one to use. I found, incidentally, that among those who scored 'plus' on the 'Left'-'Right' item there was no excess of left-handers or of those with unusual handedness or eyedness (Miles, 1993a, Chapter 21).

Some of the responses which were scored as 'zero' were of the sort that might be thought insignificant in ordinary conversation, for example asking for the question to be repeated. It is possible that in the early stages of the research I missed their significance. Later I appreciated that if you have a memory problem you may well lose track of what has been said and ask for it to be repeated, or possibly you may try to keep track of what has been said by repeating it to yourself subvocally. A memory weakness of this kind is clearly an important diagnostic indicator.

Sometimes it seemed that hesitations indicated the need to take time to work the answer out. Similarly requests for the instruction to be repeated, subvocal rehearsal, echoing the question ('Let me see - my left hand with your right, was it?') might all indicate a failure to grasp instantly the import of the question, and this might be indicative of the dyslexic's language-processing difficulties.

If there were a sufficient number of wrong answers, the result had, of course, to be scored as 'plus'. However, what particularly interested me was not wrong answers as such but the way in which the subject arrived at the answer. This was why I had to take note of small signs such as hesitations and requests for the question to be repeated. A tick was, of course, always acceptable if the subject produced the correct answer with no hesitation, but I always discouraged test users from simply putting a cross if the answer was wrong; one needed more information about the circumstances in which the error occurred.

At an early stage of the research it was essential to try the items out on children in an ordinary classroom. There would be no point in scoring a response as 'dyslexia-positive' if I found that a large number of non-dyslexic children also made the same response. I am grateful to a former honours student of mine, Ian Pollard, for collecting data from schools in the Manchester area on most of the items which I was planning to include in the test. These were adequate spellers who on the basis of two items (Similarities and Picture Completion) from the Wechsler (1949) test were within the average range for intelligence. I was thus in a position to have some idea of what non-dyslexic children of different ages were able to achieve. Inevitably, however, there was the problem of drawing boundaries: how was I to determine what were normal limits for a non-dyslexic child? At what point should I decide that this or that response merited a 'zero' or a 'plus' as opposed to a 'minus' ?

The basic requirement was to achieve the fewest possible number of false positives (non-dyslexics who came out as dyslexic) and false negatives (those who were dyslexic but were not picked up by the test). The data collected by Pollard showed me that the drawing of boundaries, though not wholly arbitrary, could not, from the nature of the case, rest on totally firm foundations.

Here are some examples. Quite a number of the children tested by Pollard made one error in the Subtraction items. This meant that if 'one error' was to be scored as a 'plus' or even as a 'zero' I would be opening the way for an excess of false positives - since you could make an error over subtraction without being dyslexic. On the other hand, if I required, say, four errors before scoring the result even as a 'zero', this would open the way to a large number of false negatives - those who were dyslexic but were not picked up. In the last resort I had simply to plump for what I judged to be the optimum place in which to draw the boundaries and trust to there being enough redundancy in the test for users to be able to avoid serious misclas-sification. In this particular case I decided that two errors in the Subtraction items should count as a 'zero' and three errors as a 'plus'.

I found that the ability to repeat polysyllables correctly increased with age. The scoring therefore had to take account of this, with the result that three words not repeated accurately counted as 'plus' at age 10 and over, whereas at age 15 and over only two failures were needed.

It was at this point that I was able to bring in Popper's (1963) principle of falsifiability. The children whom I had been seeing were, in my view, showing the manifestations of dyslexia, but what would be the predicted results if I was wrong? I would find that among normal spellers of average intelligence there would be as many dyslexia-positive responses as there were among those whom I judged to be dyslexic. This could be put to the test with minimal trouble and expense by giving the Bangor Dyslexia Test to these other children as a control group.

As a result of a shortage of resources it was possible to make comparisons only on seven out of the 10 items in the Bangor Dyslexia Test - checking school books for 'b'-'d' confusion and arranging to meet families to discover about familial incidence would have been very time-consuming, and, as a time saver, the Digits Forwards subtest was omitted. Full results have been published in the test manual (Miles, 1997) and in Miles (1993a, Chapter 7).

It was possible to collect data for 132 controls. In what follows I have included a table which originally appeared in Miles (1993a, p. 56). It relates to 80 dyslexics and 80 controls between the ages of nine and 12.

The dyslexics in this age range were found to have a mean number of 'pluses' (out of a possible seven) of 5.14, standard deviation 1.20, while the controls had a mean of 2.24, standard deviation 1.37. Statistically this difference was highly significant (note 8.1). With regard to the individual items, the percentage of dyslexics and controls showing 'pluses' is given in Table 8.1.

This meant that my attempt to falsify my hypothesis had been unsuccessful. The items in the Bangor Dyslexia Test were differentiating those whom I had judged to be dyslexic from normal achievers. Had this result not been obtained, my whole enterprise would have been broken-backed.

What in fact I found was not that non-dyslexics never produced the specified responses or that dyslexics always did so. It was rather that, overall, dyslexics were more vulnerable. It also appeared to be the case that items which were difficult for

 Item Dyslexics Controls Left-Right 78 42 Polysyllables 56 24 Subtraction 58 19 Tables 96 51 Months Forwards 60 13 Months Reversed 86 28 Digits Reversed 80 48

Adapted from Miles (1993a)

Adapted from Miles (1993a)

non-dyslexics would be extra difficult for dyslexics and items which were easy for non-dyslexics would be not easy but at any rate less difficult for dyslexics.

By taking the relative percentages of dyslexic and control children who scored 'plus' on the various items it was possible to make judgements about the diagnostic value of these items. Thus 96% of the dyslexics came out as positive on the tables item compared with 51% of the controls. In contrast, only 13% of the controls failed to recite the months of the year correctly, compared with 60% for the dyslexics. Thus a 'minus' on tables would be a strong counter-indicator of dyslexia, while a 'plus' on months of the year would be a strong positive indicator.

Two final points require discussion. The first is that on the Bangor Dyslexia Test, as in most other tests, correct responses can be learned. The second is to point out how the Bangor Dyslexia Test gives subjects the opportunity to go wrong.

With regard to the first point, what I have come to realise is that, for the most part, the correct answers to items in the Bangor Dyslexia Test are not taught in school. In particular it is no part of the syllabus to train children to distinguish 'left' and 'right' - it is something which the majority just pick up. Nor is practice normally given in repeating polysyllables, saying the months of the year or repeating strings of digits. It was a revealing experience when I assessed a boy who had nine 'pluses' out of 10 on the Bangor Dyslexia Test and was successful only in repeating the months of the year correctly. When I mentioned to his mother that I was surprised, in view of his other difficulties, at his success with this item, she gave a wry smile and said: 'We have spent ages and ages learning them.'

I see no reason why in principle it should not be possible for a child to learn correct responses to any item in the Bangor Dyslexia Test, and if this happened on any large scale some different items would be needed or greater weight would have to be attached to reports of earlier difficulties and the use of compensatory strategies. It is no accident that the Bangor Dyslexia Test does not include tests of reading and spelling, since these are taught in schools and the variety which I encountered, particularly in the case of scores on the Schonell Word Recognition test (Schonell and Schonell, 1952), could be attributed to environmental factors - the quality of the teaching received and the opportunities for learning available - more than on the constitutional factors in which I was interested. The children whom I assessed were mostly poor readers and spellers in spite of the opportunities which they had received.

In the case of the Left-Right items and the Subtraction items 'special strategies' were allowed for in the scoring, while the reciting of the harder tables (6x, 7x, and 8x) and the confusing of 'b' and 'd' show up dyslexic limitations in spite of there having been opportunities for learning. Familial Incidence is the one item in the 10 where opportunity for learning can play no part at all.

With regard to the second point, I have come to realise that dyslexia is a disjunctive concept - it can show itself by either this manifestation or that, or rather by specified manifestations in a given case (as it might be, manifestations a, b, d, f and h, and in another case by manifestations b, c, d, f and k). This is why I have never been happy to say that any one item in the Bangor Dyslexia Test is a decisive indicator of dyslexia; a diagnosis is reached by considering the pattern as a whole. A 'plus' or 'zero' response here or there may be of no special significance, but if there are several such responses one starts to wonder if they are merely coincidental, and with increasing numbers of positive indicators the hypothesis that the person is not dyslexic becomes increasingly difficult to maintain.

In the 10 items between them there is arguably enough redundancy to compensate for the fact that in a given case some of them may not be usable for diagnostic purposes, for example the Tables item in the case of children who had never been taught their tables or the 'months of the year' item in the case of a child who had been systematically taught them. In general the Bangor Dyslexia Test took the form it did because of the culture in which it originated, and it determines the form which dyslexia takes in that particular culture. Dyslexia needs a particular culture in which to manifest itself, but as the culture varies so will the manifestations.

I have never been willing to place emphasis on the precise number of positive indicators found on the Bangor Dyslexia Test by a given individual. I am not prepared to say that someone with six 'pluses' is 'more dyslexic' than someone else with five. What I do say is that in our particular culture to give someone the Bangor Dyslexia Test provides them with the opportunity to display dyslexic manifestations. Not all the manifestations will be present in any one individual, and there exist many opportunities other than those supplied by the Bangor Dyslexia Test for individuals to manifest their dyslexia (note 8.2).