Academic Attainments of Children with Down's Syndrome: A Longitudinal Study


Stephen Turner¹ and Alison Alborz²
¹Centre for Social Research on Dementia, University of Stirling, UK
²National Primary Care Research and Development Centre, University of Manchester, UK
British Journal of Educational Psychology (2003), 73, 563-583
© 2003 The British Psychological Society
  Reprinted with permission from the British Journal of Educational Psychology © The British Psychological Society granted by Claire Shinfield, Journals Department

Background. Studies of the academic progress of schoolchildren with Down's syndrome have given only limited indication of attainments at different ages. Such normative data, it is argued, could guide professionals and parents in respect to reasonable expectations and typical achievement.
Aims. The aim of the paper is to provide a descriptive account of levels and range of academic attainments reached by a representative sample of children with Down's syndrome over time.
Sample. Data relate to 106 children from three studies with the same cohort of families. Mean age was 9.1 years at the first study (1986), 13.7 at the second (1991), and 21.1 at the third (2000). Mean IQ was 40.29.
Methods. The Academic Attainments Index (AAI) covers reading, writing and numeracy, and was designed for teacher completion. In order to compare attainments to wider norms, scores were compiled by age group, and related to age-specific National Curriculum attainment levels, which are similar to US school grades.
Results. AAI scores were obtained for 102 individuals in 1986, 101 in 1991 and 79 in 2000. Comparisons across age groups indicate that higher scores were achieved by each successive age group up to age 20, and that this pattern held for both more and less able children. Once adjusted for differences in IQ scores between age groups, however, there was some evidence of a levelling off towards the end of the school career. By school-leaving age, the more able children reached at least some National Curriculum Key Stage 2 targets in reading and writing, and some Key Stage 3 targets in number work.
Conclusion. Results provide a detailed picture of attainments across ability and age groups, and broadly agree with other studies. Such data may be useful for new UK Government initiatives regarding standard setting for children with learning disabilities.

Down's syndrome is the most common chromosomal cause of intellectual disabilities. It occurs in approximately 1 in 600 live births, incidence increasing with maternal age. Memory, language and communication impairment are core features of Down's syndrome (Laws, Byrne, & Buckley, 2000), and neuromotor and sensory disabilities are more common than in the non-disabled population. Cicchetti and Ganiban (1990), in a review on the organisation and coherence of developmental processes in children with Down's syndrome, comment that the higher level information-processing skills that motivate and direct attention in non-handicapped infants are impaired in these children. Although they have the ability to perceive information accurately they are delayed in their ability to use and interpret this information. They are also inclined to become fixated on a single dimension or event. Consequently their experience of, and interaction with the world is limited. Cicchetti and Ganiban (1990) also report significant delays in semantic development. Vocabulary and other communication skills tend to increase with the child's increasing cognitive complexity but mean length of utterance, a measure of language fluency, remains poor. The reasons for this are unclear but overly directive parental interaction in response to the child's relative 'placidity' and problems with motor control or language structures were suggested as possible influences.

However, there is a great deal of variation in these deficits. The severity of intellectual disability varies by the order of 50-60 IQ points among both children and adults (Carr, 1988), and Pueschel and Hopmann (1993) report wide variation in communication skills across a wide age-spread of children with Down's syndrome. Such variation, caused by both biological and environmental factors, impacts on the level of academic attainments of these children (Laws et al., 2000).

In the UK, the number of children of school age (5-19) with Down's syndrome has stabilised at around 8-9,000 (Steele & Stratford, 1995). The abilities of children with Down's syndrome in reading, writing and number work has received considerable attention in the learning disability and special education research communities. However, UK studies tend to focus on primary school age children (Casey, Jones, Kugler, & Watkins, 1988; Nye, Clibbens, & Bird, 1995; Philps, 1993), to be limited to one area of attainment, such as literacy or numeracy (Buckley, Bird, & Byrne, 1996; Nye et al., 1995), and/or to be based on cohorts of children born in the 1960s, whose education was largely prior to the 1981 Education Act (Buckley & Sacks, 1987; Carr, 1988; Casey et al., 1988; Shepperdson, 1988). In addition, writing ability has been relatively under-researched, and few studies have used measures of academic abilities appropriate to the wide range of abilities found among children with Down's syndrome, or reported attainment levels for children in different age groups.

As a result, there remains much imprecision regarding what level and range of competence can be expected at different ages in children with Down's syndrome. This uncertainty may be part of the cause for the persistence of misconceptions regarding the educability of such children. One example of an extremely pessimistic view comes from a widely available medical reference book:

It has been estimated that 6% of Down's children are probably capable of profiting appreciably from attendance at schools for the educationally handicapped. Practically all who survive to school age gain some benefit. Most eventually acquire some degree of speech, and about 5% learn to read. They practically never learn to write.
(Macpherson, 1999)

Such a pessimistic view is challenged by Rynders et al. (1997) who argue that misinformation and too little information about the educational potential of schoolchildren with Down's syndrome have led school teachers, psychologists and other school personnel to have low educational expectations of these children. In general, the findings of studies of the academic progress of children with Down's syndrome through secondary school and beyond support this viewpoint. Three non-UK studies report particularly high levels of literacy. Bochner and Pietrese (1996), in a study of 66 13-20 year-old students with Down's syndrome living in New South Wales, report that 86% of parents said their adolescent child 'could read'. However, it is difficult to determine the level of competency this statement describes: 22% reported that their child could 'only read a few words', while 26% said their child 'often' read books written for teenagers. In the US, Peuschel and Hopmann (1993) used a parentcompleted checklist on communication and language skills with 154 parents of children with Down's syndrome aged 1-21 years. They report that two-thirds of the 17- 21 age group in their study (N = 31) could read sentences, and half could read books. However, no information on IQ or severity of developmental disability was reported.

Rynders et al.'s (1997) report results from a combined three data set study of 171 individuals which generated both longitudinal and cross-sectional data covering the age range 5-18+. Mean IQ ranged from 45 to 55 depending on age group - comparable to that of the young people in the present study. Attainments were reported as school grade equivalents to be more pertinent to teachers and parents. The authors report little change across age groups in number grade equivalency scores, but that reading recognition and comprehension scores increased steadily as age increased. By the age of 18+ (n = 15) mean scores represented ability to read newspaper sections, menus, TV guides, etc.. They conclude that reading comprehension scores, which they regard as representing the 'acid test' of literacy, reflect relatively good academic achievements in school years, and that progress continued into young adulthood.

In a UK study involving of children with Down's syndrome aged 11 to 17 years, Buckley and Sacks (1987) report data on reading, writing and number skills, in terms of percentages of children attaining different levels of skills, e.g., can read sentences, can write own name, can do simple addition. They report that 16% (15) were 'quite good' readers, capable of reading simple books; half could add numbers up to 10, and about one-third could subtract. Six percent could multiply, and 3% divide. Two UK studies extend into adult years. Carr (1988) reports results of reading and number tests for 41 21-year-olds with Down's syndrome who took part in a longitudinal study. Only 39% (16) of these were able to gain reading ages on the Neale Analysis of Reading Test (mean age 7 years 8 months for accuracy; 6 years 9 months for comprehension). Reading levels were computed for another 15 young people unable to achieve a reading score on the test but able to name at least some letters. No examples of different levels of attainment are given, although Carr reports that five young people were said by their mothers to read books, newspapers or comics for pleasure. Numeracy was tested using the Vernon's Arithmetic Test: 83% (34) were able to score. Of these, 63% could only recognise numbers and count; 23% could add two figures 0-9, and 15% could subtract figures 0-9.

Shepperdson's (1994) study involved 49 young people who were assessed by their teachers as teenagers and again when they were young adults. At teenage, 37% (18) had 565 Academic attainments and Down's syndrome no reading ability; 18% (9) could read their own name; and 45% (22) could read some words. Ten years later 45% (22) were assessed as having no reading ability; 23% could read their own name; and 33% (16) could read some words. Numerical ability was assessed through a teacher-completed checklist of 26 items. At teenage, 18% (9) had no numerical abilities: by early adulthood, this had increased to 31% (15). However, the same study reports higher reading and numerical scores for a smaller cohort (n = 26) of children born in the 1970's (i.e., around 10 years later than the larger cohort), who were assessed as teenagers (Shepperdson, 1994).

It is reasonable to suspect that much of the research evidence based on studies with children who were at school during the 1970s and 1980s may still underestimate the potential for academic attainment. Shepperdson (1994) suggests that some children in her study could only read a little because they had not been taught. Buckley et al. (1996), in a review of studies of reading attainment among children with Down's syndrome, caution that data from such less recent studies are difficult to interpret as many of the children will not have been taught to read, and that it would be dangerous to assume that reported attainment levels represented the upper limits of abilities. This argument is supported by recent research by Laws et al. (2000) who report that all but two of 22 children with Down's syndrome aged 7 to 14 who were receiving reading tuition in mainstream schools were classified as readers on a word identification reading test.

Pueschel and Hopmann (1993, p. 350) argue that there remains a need for normative data about the communication and language development of people with Down's syndrome to guide professionals and parents in respect to reasonable expectations and typical achievement. The last 20 years in the UK has seen a wide-ranging imposition of standards and assessment in public services. The 1988 Educational Reform Act introduced a National Curriculum for England which all pupils in state-maintained schools are expected to follow, including those with special educational needs. The curriculum established levels of attainment by subject and age by which progress may be judged through Standard Assessment Tests. For example, Key Stage 2 assessment tests are conducted at age 11, in the final year of primary school, and cover English (including reading and writing skills), mathematics and science. Nitko (1980) characterised such targets as criterion-referenced, that is, they are based on evidence of skill acquisition on subject-matter increasing incrementally in difficulty or complexity. There is some controversy over the reliability and validity of this type of testing, and the assumptions which underpin assessment in general have been challenged (see, for example, Delandshere, 2001). Despite these criticisms, it has recently been recognised that an assessment regime is also appropriate for pupils with special educational needs, albeit based on more limited or specific targets below and within Levels 1 and 2 of the National Curriculum (DfEE, 2001).

The aim of the present paper is to use a specially developed measure of academic attainment to identify levels reached by a representative sample of children with Down's syndrome at different stages of their school career. Such data may provide a basis for monitoring the educational progress and attainment of children with learning disabilities, as proposed recently by the Government (Dept. of Health, 2001). Agerelated attainments in the cohort are then related to specific skill levels laid down in the National Curriculum. This approach reflects that of the Department for Education and Employment (now the Dept. for Education and Skills, DfES) which has developed differentiated assessment criteria for pupils with special educational needs within the overall structure of the National Curriculum (DfEE, 2001).

While this current paper is descriptive in content, a further paper will examine the predictors of academic attainments at adulthood, following the transactional family model adopted with these families by Sloper, Cunningham, Turner, and Knussen (1990).

The Manchester Down's Syndrome Cohort

The current study is the latest in a series of studies with the Manchester Down's Syndrome Cohort (Byrne, Cunningham, & Sloper, 1988; Cunningham, 1986; Sloper et al., 1988; Sloper & Turner 1994; Turner & Alborz, 2000). Initially, 181 families of children with Down's syndrome born between 1973 and 1980 were recruited, representing approximately 90% of all such births in the Greater Manchester area over this period. For the first few years of the children's lives, research was conducted within the context of a preschool early intervention programme. The families represent a broad and representative range of demographic and socioeconomic characteristics (Sloper et al., 1990).

Sample size and attrition

The nature of longitudinal research is such that the number of people for whom data are available fluctuates. Within each study, the use of multiple sources also means that there are slight variations in the numbers involved in analyses. We have striven to make the reason explicit in each case. By age 5, the numbers in the cohort had declined from the original 181 to 134. Attrition was entirely due to deaths (20 - mostly in the first year) and family moves from the area (27). Since then attrition due to deaths and moves has continued. In addition a small number of families have declined to take part in specific studies. Thus 120 families took part in the 1981 Mother Interview study, 123 in the Family Adaptation study of 1986-87, and 106 in the Parent Child and Sibling study of 1991. A total of 107 families were approached for the 2000 Coming of Age study, and data were obtained from 90. This represents 85% of the families approached, and 67% of all the families in the cohort some 15 years before. Of the 17 families not included in the study, five were not traceable or had moved from the area. Seven declined to take part, and four were not contacted because of known family difficulties, including severe illness. In one family the young person had recently died.

For the purposes of this paper, the longitudinal data set is limited to 106 individuals with at least one measure of academic attainment from one of the three studies from the period 1986 to 2000. For brevity, these three studies are referred to as the 1986 study, the 1991 study, and the 2000 study. A total of 271 academic attainment scores are available for analysis from these 106 individuals, i.e., 85% of the possible total.

Of these 106 young people, 62% (66) were male and 37% (40) female. Table 1 gives the chronological age of the individuals at the time of the three studies. Age groups shown cover up to 24 months (age shown in years plus or minus 12 months).


Table 1. Age distribution; 1986, 1991 and 2000 studies
Number in each age group (± 12 months): mean age, SD in months; N
Study        Age 8     Age 10     Age 12     Age 14     Age 16     Age 18     Age 20     Age 22     Age 24  
1986 57 36 8 1  
  Mean: 9.1yrs, SD: 19.9mths, N:102  
1991   36 43 21 1  
  Mean: 13.7yrs, SD: 19.1mths, N:101  
2000   48 21 9
  Mean: 21.1yrs, SD: 16.1mths, N:78
Totals 57 36 44 44 21 1 48 21 9

IQ scores

The 1986 study collected data from 104 of the 106 in the longitudinal dataset; of these, 102 had both IQ and AAI (see Measures) scores. Of the other two, one was not testable on the mental age tests used, and one had missing AAI scores (non-response from teacher). These 102 children had a mean IQ score of 40.29 (SD = 10.95). This compares with a mean of 37.2 (SD = 11.9) among 44 11-years-olds with Down's syndrome in Carr's study (Carr, 1995). Table 2 shows 1986 IQ scores placed into groups corresponding to profound, severe, moderate and mild developmental disability.


Table 2. IQ scores in 1986
20 or less       over 20-50       over 50-70       over 70       Total
4% 80% 17% 100%
4 81 17 102

Educational history

In 1986, 63% of the children attended a school for pupils with Severe Learning Disabilities (SLD)¹, 16% attended an school for pupils with Moderate Learning Disabilities (MLD), 18% were in a mainstream class or unit in a mainstream, and 3% had some other educational placement (N = 100; not known:4). By 2000 these figures (for last school attended) were: 63% in SLD schools; 17% in MLD schools; 15% in mainstream classes or units, and 5% (4) in other placements (N = 79: not known: 1; number of schools involved: 36). Fourteen percent (11) had attended mainstream schools throughout the period covered by the three studies. This is similar to the 15% of 14-year-olds in mainstream schools reported in a recent large scale UK study (Cuckle, 1997). Seventy-one percent (56) had left school at age 18 or 19, 25% (20) had left before 18, and 4% (3) had left at age 20.

Measures

The Academic Attainments Index (AAI) is intended specifically for use with children with Down's syndrome, and was developed in earlier work with this sample (Sloper et a1., 1990). It is based on the measure developed by Lorenz (Lorenz, Sloper, & Cunningham, 1985) in a study of reading among children with Down's syndrome. In order to reflect the range of abilities that could be expected in primary school age children with Down's syndrome, the Index was designed to cover early stages of prereading, number and writing skills as well as more advanced skills. No standardised measures which covered such a range were identified at that time. For example, only 17% of the sample in Sloper et al.'s study of these families achieved a reading score on the Spar Reading Test (Sloper et al., 1990). The Index is a criterion-referenced assessment comprising three separate subscales (checklists) covering reading (17 ordered items), writing (19 items) and numeracy (22 items), designed for completion by the teacher who knew the individual child best (see Appendix A).

In the study by Lorenz et al. (1985), items were selected to reflect a graded sequence of skills, and the authors report that none of the 115 children with Down's syndrome in the study passed a higher item while failing a lower, indicating that a cumulative score could be used to reflect skill level achieved. However, it cannot be assumed that the AAI measure is truly interval in nature, in that a 1 point change in the lower end of the scale is necessarily equivalent to a 1 point change at the upper. If, for example, the steps between items become progressively greater, this may lead to an apparent plateau effect towards the end of the school career. While mean scores at different ages are compared in this paper (see Table 5 and Figure 2), these comparisons should be evaluated with caution. However, the AAI distribution in the 2000 study contained 43 different values and approximated to the normal distribution (Figure 1).

Distribution of scores on Academic Attainment Index
Figure 1. Distribution of scores on Academic Attainment Index, 2000 study

Some changes were made in the 2000 study to reflect the fact that one third of the young people were not in tertiary education, and others were not following courses that involved direct work relating to number, reading or writing. If parents were not able to identify a tutor with relevant current knowledge of the young person's academic abilities, parents themselves were asked to complete the checklist. The need to involve parents required the rewording of nine items to clarify the language used (e.g., phonics; sight words). A change was also made to the way the scoring system was computed. In all three studies, respondents were asked to select one of three responses to each item: 'can do', 'can do with help'; 'can't do', and were provided with instructions to guide their choice. It was decided that at this stage of the young people's lives it would be more meaningful to base Index scores on 'can do' responses only (i.e., discounting 'can do with help' responses). Thus the measure would more closely represent independent skills achieved by the start of adult life, when support in such skills may be less available. 1986 and 1991 scores were recomputed to ensure comparability. The two forms of AAI scoring correlated at 0.99 (1986 scores), 0.98 (1991 scores) and 0.96 (2000 scores). The resulting measures have possible ranges of 0-58 (AAI), 0-17 (reading), 0-19 (writing), and 0-22 (number).

Comparability with the 1986 and 1991 studies was maximised by giving precedence to scores obtained from educational professionals in the 2000 study. After eliminating returns with more than 10% of items missing, the result was: 79 Academic Attainments Index scores (36 from tutors only; 38 from parents only; 5 from a combination of tutors and parents); 80 reading scores (38 from tutors, 42 from parents); 80 writing scores (40 from teachers: 40 from parents); 79 numeracy scores (36 from tutors: 43 from parents).

Key attainments

While the overall rate of progress as indicated by total and subscale mean scores should be treated cautiously, it may be appropriate to use AAI scores to indicate the individual skills typically attained by children in different age groups. The scale was designed by its authors to be cumulative, with items ordered in difficulty, which allowed quartile scores to be calculated for each age group and related to individual items. Thus the median score of 13 in number work at age 16 was taken to mean that 50% of the cohort children could complete the 13th item 'subtracts from written numbers up to nine without materials' by that age. The validity of this assumption was checked by comparing quartile score levels and attainment of the corresponding item. In a few cases, children were judged to need help with the relevant item, but in the majority attainment of the item conformed with the quartile score. It was concluded that the items shown in Table 5 give a reasonable reflection of particular attainments within each age group. Comparison with National Curriculum target skills was made by reference to Curriculum Guidance for the Foundation Level (http://www.qca.org.uk) and guidance on Key Stage 1, 2 and 3 attainment levels found at (www.nc.uk.net). An analysis of the reliability and validity of the measures is presented in Appendix B.

Results

Table 3 shows the distributions of the AAI and its three components at the three points in time. The range of AAI scores was 0-50 (of a maximum possible 58) in 1986, 0-54 in 1991, and 0-58 in 2000. 1986 scores were skewed towards the lower frequencies (skewness statistic = 1.2); for 1991 scores this statistic was 0.2, and for 2000 scores it was -0.0. 1986 scores correlated with 1991 scores at r = 0.82, N = 99, and 1991 scores correlated with 2000 scores at 0.78, N = 76 (both significant at .001, Spearman Rank Coefficients). Changes in Index means over time were also significant (1986/1991: t(98) = 12.78, p = .001; 1991/2000: t(75) = 5.06, p = .001).

While Table 3 shows increases in all mean scores over time, not all children showed improvement. None of the four children with IQs below 20 (Table 2) showed any change from the extremely low ( < 4) 1986 AAI scores in 1991 or 2000. However, the five children with the next lowest 1986 IQ scores (over 20 but less than 30) showed considerable variation in 2000 AAI scores (ranging from 3 to 24). It should be remembered that the data include the small number of children for whom no meaningful level of academic skills was achievable.


Table 3. Academic Attainments Index scores: 1986; 1991; 2000
Measure      Year of study
1986
Mean (SD)
(N)
1991
Mean (SD)
(N)
2000
Mean (SD)
(N)
AAI    12.86 (10.13)   
(102)
   23.18 (13.00)   
(101)
   29.56 (14.10)   
(79)
Reading 4.83 (3.67)
(103)
7.38(4.53)
(101)
9.14 (4.82)
(80)
Writing 4.62 (2.92)
(103)
7.76 (4.06)
(101)
10.08 (4.70)
(80)
Number 3.38 (4.13)
(102)
8.04 (5.32)
(101)
10.32 (5.67)
(79)

Given that there was a mean gap of 6½ years between the first and second studies, and a mean gap of 7½ years between the second and third, there is some indication that improvement in scores was at a lower rate during the second period. For example, mean AAI score added over 11 points between the first two studies, but only 6 between the second and third. This is despite any inflation of Time 3 scores due to a higher drop out rate of low 1986 scorers, and the possible overestimate of abilities on the part of parents in 2000 (see Appendix B for a discussion of these issues).

In order to clarify the relationship between chronological age and AAI score, the data were reorganised into age groups as described above. Table 4 shows these data, omitting the 18 and 24 year-old age groups due to their small numbers.


Table 4. Academic Attainments Index mean scores by age group
     Age 8   
n = 57
   Age 10   
n = 36
   Age 12   
n = 44
   Age 14   
n = 44
   Age 16   
n = 21
   Age 20   
n = 48
   Age 22   
n = 21
AAI 10.75
(8.83)º
15.22**
(11.53)º
20.73
(11.23)
23.27**
(12.51)
25.81
(16.33)
30.83
(14.19)
25.81
(15.35)
Reading      4.26
(3.47)
5.47*
(3.98)
6.97
(4.18)
7.11
(4.42)
8.05
(5.23)
9.73
(4.83)
7.57
(5.09)
Writing 4.02
(2.47)
5.08**
(3.25)º
7.09*
(3.21)
8.15**
(4.17)
8.00
(5.08)
10.41
(4.83)
8.81
(4.77)
Number 2.43
(3.48)º
4.67**
(4.76)º
6.52
(4.73)
8.00**
(4.94)
9.76
(6.78)
10.63
(5.84)
9.43
(6.17)
Notes: Age groups = No. years ±12 months;
*mean score sig. higher than predecessor at p < 0.05 (one sample t test)
**mean score sig. higher than predecessor at p < 0.01 (one sample t test)
ºskewness statistic > 1.0 (distribution skewed towards 0)

Figure 2 shows the same information in line graph form. Given variations in mean IQ scores between age groups (see Appendix B), Figure 2 also shows mean AAI scores adjusted for inter-group differences in IQ.


Distribution of scores on Academic Attainment Index
Figure 2. Academic attainment scores by age group

The IQ adjusted data shown in Figure 2 appear to indicate a post-16 'plateau effect'. However, this could merely reflect that greater cognitive leaps may be required between items at the higher end of the checklist. In order to investigate this, the 1991 and 2000 scores of the children who were in the 12 year-old age group in 1991 were compared. The same comparison was then made for those who were in the 14 and 16 year age groups in 1991. If an age-related plateau effect existed, these older children would be less likely to show increases in scores between the two studies than the younger children. This proved to be the case. There was no significant increase in AAI scores from 1991 to 2000 for the 14 and 16 year age groups, t(29) = 1.84, p = .08; t(17) = 1.08, p < .29 respectively, but the increase in 2000 scores over 1991 scores for the children in the age 12 in 1991 group was significant at t(52) = 5.62, p < .001.

It may be argued that this result is still consistent with the possibility that there is greater distance between items at the top end of the scale and it is also relevant that the time gap between the two assessments was greater for younger than older children (97 months for the 12 year-old age group, 84 months for the 14 year-old age group, and 82 months for the 16 year-old age group) i.e., the younger children had more time to progress between the two studies. However, the distribution of AAI scores over the maximum range (0-58) follows a broadly normal distribution (see Figure 1) with no skew (skewness statistic -0.0) which is contrary to what one would expect if it required greater cognitive leaps to increase scores at the higher end of the scales. The implication is that the plateau represents diminishing skill acquisition. This may be due to some individuals beginning to reach the limits of their 'academic potential', or a school curriculum changing in emphasis to life rather than academic skills towards the end of their school career.

Key attainments

Table 5 shows individual checklist items corresponding to the 1st, 2nd and 3rd quartiles of subscale scores (rounded to the nearest whole number for each age group). The 1st quartile skill level represents the level achieved by 75% of the group, the 2nd quartile level is that achieved by 50%, and the 3rd quartile level is that achieved by 25%. The age 22 group has been omitted as little change was found for any of the three quartile scores for the subscales vis-à-vis the age 20 group. In order to show the actual attainments of the children, scores have not been adjusted for differences in IQ between age groups. The effect of adjusting for IQ would be to inflate the level of items attained in 21 of the 54 cells in the table (by one item in all but 6 cases), and to reduce it by one item in another 6 cells. Higher adjusted scores are concentrated in the 16 year old age group, due to the low mean IQ score (36) for that age group.

Where possible, these items have been matched to National Curriculum attainment targets, indicated in Table 5 by F (Foundation level), KS1 (Key Stage1), KS2 (Key Stage1), and KS3 (Key Stage 3). Where there is a close match, Table 5 shows the item in bold; where there is no close match, the item has been assigned to the level indicated by preceding matched items.


Table 5. Reading, writing and number attainments compared to National Curriculum targets: By age group and quartile score
  Age 8
n = 54
Age 10
n = 36
Age 12
n = 43
Age 14
n = 44
Age 16
n = 21
Age 20
n = 48
Reading
1st Quartile
Item 2:
Matches words of up to 5 letters
Item 3:
Recognizes his/her own name written down
Item 4:
Recognizes 5 to 10 sight words
Item 3:
Recognizes his/her own name written down
Item 2:
Matches words of up to 5 letters
Item 6: F
Recognizes up to 25 sight words
2nd Quartile Item 3:
Recognizes his/her own name written down
Item 4:
Recognizes 5 to 10 sight words
Item 7: F
Reads simple sentences
Item 6: F
Recognizes up to 25 sight words
Item 9: KS1
Reads simple picture books
Item 11: KS1
Builds simple words through knowledge of letter sounds
3rd Quartile Item 7: F
Reads simple sentences
Item 8: F
Knows most letter sounds of alphabet
Item 10: KS1
Reads & acts on written signs e.g. 'danger'
Item 11: KS1
Builds simple words through knowledge of letter sounds
Item 13: KS2
Reads and follows a line of instructions e.g. 'cut along line'
Item 13: KS2
Reads and follows a line of instructions e.g. 'cut along line'
Writing
1st Quartile
Item 2: F
Scribbles purposely with pencil or crayon
Item 3: F
Copies a circle
Item 5: F
Copies first name
Item 5: F
Copies first name
Item 4: F
Copies letters
Item 7: F
Writes first name and surname independently
2nd Quartile Item 3: F
Copies a circle
Item 4: F
Copies letters
Item 8: F
Copies a simple sentence of 4 or more words
Item 7: F
Writes first name and surname independently
Item 7: F
Writes first name and surname independently
Item 9: F
Writes a simple sentence of 4 or more words
3rd Quartile Item 6: F
Writes first name independently
Item 7: F
Writes first name and surname independently
Item 9: F
Writes a simple sentence of 4 or more words
Item 10: KSI
Writes own name, address, and telephone number
Item 11: KS1
Writes simple sentences when dictated
Item 14: KS2
Writes a short personal letter
Number
1st Quartile
No items Item 1: F
Discriminates between largest & smallest group of objects
Item 1: F
Discriminates between largest & smallest group of objects
Item 4: F
Makes a group of 9 objects
Item 3: F
Makes a group of 5 objects
Item 7: F
Names and matches symbols, 0 to 9
2nd Quartile Item 1: F
Discriminates between largest & smallest group of objects
Item 4: F
Makes a group of 9 objects
Item 7: F
Names and matches symbols, 0 to 9
Item 7: F
Names and matches symbols, 0 to 9
Item 11: KS2
Add written numbers up to 10, without materials
Item 10: KS1
Add written numbers up to 10 with materials
3rd Quartile Item 4: F
Makes a group of 9 objects
Item 6: F
When given a group of objects, takes one away
Item 10: KS1
Add written numbers up to 10 with materials
Item 13: KS2
Subtracts from written numbers up to 9 without materials
Item 14: KS2
Adds two numbers up to 20 with materials
Item 15: KS3
Adds two numbers up to 20, without materials
Note. Items in bold correspond to National Curriculum attainment targets
F - item corresponds to National Curriculum Foundation level abilities
KS1 - item corresponds to National Curriculum Key Stage 1 level abilities
KS2 - item corresponds to National Curriculum Key Stage 2 level abilities
KS3 - item corresponds to National Curriculum Key Stage 3 level abilities

Table 5 indicates that a 'plateau effect' is most marked in writing, where median scores remain level across four age groups. Median number attainments level off around age 16. However this is not the case for the lower quartile scores, where significant post-16 progress is indicated in all three areas.

Some comparative data on reading attainments are available from Pueschel and Hopmann's 1993 study. The parent-completed checklist used in that study contained several items which correspond to reading items from the AAI. As Table 6 shows, the two studies obtained broadly similar results.


Table 6. Reading attainments: Pueschel and Hopmann (1993) and present study
     Pueschel and Hopmann (1993)       Manchester Down's Syndrome Cohort   
  1991 Data 2000 Data
Reading attainment items 11-16 years
n = 34
17-21 years
n = 31
11-16 years
n = 96
19-21 years
n = 60
Can use phonics to build new words (AAI item 11) 44% 53% 27% 47%
Reads sentences (AAI item 7) 61% 67% 52% 57%
Reads books (AAI item 9) 56% 50% 53% 58%

Discussion

The exploitation of the age range within each of the three studies in a longitudinal series has provided a more detailed picture of the nature and pace of academic attainment within a relatively large and representative group of young people with Down's syndrome. The availability of data on developmental level also permits intergroup control of this factor. However, the smaller numbers in the age 16 group (21, compared to 36-53 in other age groups) suggests that caution should be used in interpreting these data. Although the study would be strengthened by inclusion of an 18 year age group, it should be noted that the age 20 group (n = 38) includes 20 individuals aged 19 who had only recently left school.

Cunningham, Glenn, Lorenz, Cuckle, and Shepperdson (1998), in a review of the extent and impact of mainstream placements, suggest that overall the evidence is for greater educational progress among children with Down's syndrome in mainstream schools. A recent UK study by Laws et al. (2000) reports higher achievements in grammar and numeracy among children in mainstream schools compared to children in special schools matched for age and receptive vocabulary. The present study is predominantly of pupils in special schools, with only 11% in mainstream schools. In this study, 11 of the children with IQ scores of 50 or more attended mainstream schools, and seven attended special schools. The relatively small number of the cohort in mainstream schools suggests that the overall attainment levels reported are lower than what might have been achieved if the proportion attending mainstream schools was closer to that reported in some later studies (Shepperdson, 1997). In this respect the present study mirrors that of Rynders et al. (1997) in the US, who comment that, as most of the individuals in their studies were in segregated placements, their academic attainments must be seen in the context of receiving an education programme orientated toward self-care rather than literacy or numeracy skills. The implication for the present study is that is that some of the children could have achieved higher attainments in mainstream schools.

Nevertheless, the levels of attainment found in the present study compare favourably with those reported by Carr (1988) and Shepperdson (1994) in similar age ranges, and are consistent with those reported by Pueschel and Hopmann (1993) (Table 6). The comparison with National Curriculum targets (Table 5) suggests that by the time they left school: (1) 75% of the young people in the study had achieved at least some Foundation level skills in all three areas of academic attainment; (2) 50% had attained some Key Stage 1 skills in reading and number work; and (3) 25% demonstrated Key Stage 2 level skills in all three areas, and Key Stage 3 skills in some aspects of number work. These levels may represent attainment levels one might expect for children with Down's syndrome of above average ability. National Curriculum Foundation and Key Stage 1 and 2 targets relate to attainments expected of (but not necessarily achieved by) 5, 7 and 11 year-olds respectively in the general school population.

Interpreting these results requires caution. The low mean IQ score in the 16 year-old age group suggests that attainment levels shown in Table 5 for this group may underestimate levels achievable in the wider population of children with Down's syndrome of this age. There is some evidence of variation in achievement by subject area, with writing skills reaching only Foundation level for 75% of the group. Writing is known to lag behind reading in terms of skill acquisition, and this may be exacerbated in this group by associated poor physical development. Morris, Vaughan, and Vaccaro (1982) found that children with Down's syndrome aged 4 to 17 years had lower muscle tone and inferior grip strength compared to children without disabilities. Arguably this would affect writing due to the fine motor control movements required in writing. Predictors of attainment in these measures in this group of young people are examined in a further paper in preparation.

Rynders et al. (1997) argue that there was little evidence of a plateau in attainments in their samples, and that teachers should be reassured that their efforts will be justified by the long-term progress shown and literacy levels achieved by many of their pupils. The evidence from the present study is less clear-cut. Improvement tailed off post-16, although lack of information on the 18 year age group means this observation should be treated with caution. Bias towards higher performers due to sample attrition does not necessarily imply, however, that a plateau effect would be reinforced if lower performing individuals had remained in the study, as lower quartile scores continued to increase post-16. Cognitive leaps between items for those attaining skills at the higher end of the checklist are an issue. Increasing item complexity and/or changing educational priorities in the later years for pupils at special schools could provide alternative explanations for an apparent 'plateau effect'. More generally, it is relevant to note that many typically developing children do not reach Key Stage Assessment targets.

The present study, in common with many of the studies referenced in this paper, include individuals who were unable to achieve a significant level of literacy or numeracy. The number that failed to make any meaningful progress over the time span of the three studies was small. There were three young people whose 2000 AAI score was at or more than one standard deviation below the 1986 mean for the sample as a whole, and another nine whose 2000 scores were below the 1986 mean - approximately 12% of the 1986 sample. (These children had a mean age of 10 years in 1986, compared with just over 9 for the sample as a whole, so their performance is not explained by age differences.) The needs of these children should not be overlooked in any attempt to set standards or targets in the education of children with learning disabilities. The academic attainments measured in this study are not the only relevant measure of an individual's educational progress, nor indeed necessarily reflect their development as individuals or as social actors. While Bird and Buckley (1994) rightly argue that preconceived ideas about limits on performance may be restrictive, and lead to self-fulfilling prophecies, there is also the danger that a new emphasis on achievement will undervalue the education of those unable to make use of new opportunities. More broadly, Farrell (1996) warns against equating literacy acquisition with the mastery of discrete skills, and ignoring 'questions about what part literacy might play in the lives of students, especially as they move into adolescence and adulthood' (p.296).

In England, the Government has from 2002 established a revised statistical system to monitor the progress and attainment of children, including those with learning disabilities. All schools are expected to set targets for the achievement of children working below age-related expectations on the National Curriculum, following guidance from the Qualifications & Curriculum Authority on target setting and assessment for children working below these age-related norms. The aim is to encourage higher expectations of and higher attainment by all children with special needs (DfEE, 2001).

This paper takes forward previous studies involving schoolchildren with Down's syndrome with the aim of providing an evidence base for such initiatives. Further work may be beneficial at this stage to evaluate the appropriateness of the age-related attainment levels reported in this paper in the light of recent changes in educational provision. No attempt has been made in this paper to identify positive or negative predictors of progress relating to the individual or their family. However the data contain much information relevant to this issue, and is the subject of a further paper currently in preparation which may enable a clearer picture to emerge of the factors, including mainstreaming, that may influence continued progress in the later teen - early adult years.

Acknowledgements

The authors would like to thank the young people, parents and teachers who have given their time and interest to this work over many years. We would also like to thank ex-colleagues at the Hester Adrian Research Centre for advice and support, and to Professors Cliff Cunningham and Sheila Glenn at the Department of Health Studies, John Moores University, Liverpool, for invaluable assistance and cooperation. This study was supported by the Department of Health and the ESRC (Award nos. R000231975 and R000237012).

References

Bayley, N. (1969). The Bayley Scales of Infant Development. New York: The Psychological Corporation.

Bird, G., & Buckley, S. (1994). Meeting the educational needs of children with Down's syndrome. Portsmouth: University of Portsmouth.

Bochner, S., & Pietrese, M. (1996). Teenagers with Down syndrome in a time of changing policies and practices: Progress of students who were born between 1971 and 1978. International Journal of Disability, Development and Education, 43, 75-95.

Buckley, S., & Sacks, B. (1987). The adolescent with Down's syndrome: Life for the teenager and for the family. Portsmouth: Sarah Duffen Centre, Portsmouth Polytechnic.

Buckley, S., Bird, G., & Byrne, A. (1996). Reading acquisition by young children. In B. Stratford & P. Gunn (Eds.), New approaches to Down's Syndrome (pp. 268-279). London: Cassell.

Byrne, E. A., Cunningham, C. C., & Sloper, P. (1988). Families and their children with Down's syndrome: One feature in common. London: Routledge.

Carr, J. (1988). Six weeks to 21 Years: A longitudinal study of children with Down's syndrome and their families. Journal of Child Psychology and Psychiatry, 29, 407-431.

Carr, J. (2000). Intellectual and daily living skills of 30-year-olds with Down's syndrome: Continuation of a longitudinal study. Journal of Applied Research in Intellectual Disabilities, 13, 1-16.

Casey, W., Jones, D., Kugler, B., & Watkins, B. (1988). Integration of Down's syndrome children in the primary school: A longitudinal study of cognitive development and academic attainments. British Journal of Educational Psychology, 58, 279-286.

Cicchetti, D., & Ganiban, J., (1990) The organization and coherence of developmental processes in infants and children with Down syndrome. In R.M. Hodapp, J. A. Burack & E. Zigler (Eds.), Issues in the developmental approach to mental retardation (pp. 169-225). Cambridge: Cambridge University Press.

Cuckle, P. (1997). The school placement of pupils with Down's syndrome in England and Wales. British Journal of Special Education, 24, 175-197.

Cunningham, C. C. (1986). Early intervention: Some findings from the Manchester Cohort of 577 Academic attainments and Down's syndrome Children with Down's syndrome. In M. Bishop, M. Copley & J. Porter (Eds.), Portage: More than a teaching programme (pp. 89-106). London: NFER-Nelson.

Cunningham, C. C., Glenn, S., Lorenz, S., Cuckle, P., & Shepperdson B. (1998). Trends and outcomes in educational placements for children with Down's syndrome. European Journal of Special Needs Education, 13, 225-237.

Delandshere, G. (2001). Implicit theories, unexamined assumptions and the status quo of educational assessment. Assessment in Education: Principle, Policies and Practice, 8, 113-133.

Department of Health (2001). Valuing people. A new strategy for learning disability for the 21st century. A White Paper (Cm 5086). London: HMSO. Department for Education and Employment (2001). Supporting the target setting process (DfEE ref. 0065:2001). London: DfEE.

Farrell, C. (1996). Continuing literacy development. In B. Stratford & P. Gunn (Eds.), New approaches to Down's syndrome (pp. 280-299). London: Cassell.

Laws, G., Byrne, A., & Buckley, S. (2000). Language and memory development in children with Down syndrome at mainstream and special schools: A comparison. Educational Psychology, 20 (4), 447-457.

Lorenz, S, Sloper, T., & Cunningham, C. C. (1985). Reading and Down's syndrome. British Journal of Special Education, 12, Research Supplement, 65-67.

Macpherson, G. (1999). Black's Medical Dictionary (39th ed.). London: A. & C. Black.

McCarthy, D. (1972). Manual for the McCarthy Scales of Children's Abilities. New York: The Psychological Corporation.

Morris, A. F., Vaughan, S. E., & Vaccaro, P. (1982). Measurements of neuromuscular tone and strength in Down syndrome children. Journal of Mental Deficiency Research. 26, 41-46.

National Curriculum Online (n.d.). National Curriculum Documents. Retrieved from http://www.nc.uk.net

Nitko, A. J. (1980). Distinguishing among the many varieties of criterion-referenced tests. Review of Educational Research, 50, 36-43.

Nye, J., Clibbens, J., & Bird, G. (1995). Numerical ability, general ability and language in children with Down's syndrome. Down's Syndrome: Research and Practice, 3, 92-102.

Philps, C. (1993). A comparative study of the academic attainments and language development of children with Down's syndrome placed in mainstream and special schools. Unpublished M. Phil. thesis, University of Wolverhampton.

Pueschel, S. M., & Hopmann, M. R. (1993). Speech and language abilities of children with Down syndrome. In A. P. Kaisen & D. B. Gray (Eds.), Enhancing children's communication: Research foundations for intervention (pp. 335-362). London: Brookes.

Qualifications & Curriculum Authority (2000). Curriculum guidance for the Foundation stage. Retrieved from: http://www.qca.org.uk/ca/foundation/guidance.

Rynders, J., Abery, B. H., Spiker, D., Olive, M. L., Sheran, C. P., & Zajac, R. J. (1997). Improving educational programming for individuals with Down syndrome: Engaging the fuller competence. Down Syndrome Quarterly, 2, 1-11.

Shepperdson, B. (1988). Growing up with Down's syndrome. London: Cassell.

Shepperdson, B. (1994). Attainments in reading and number of teenagers and young adults with Down's syndrome. Down's Syndrome: Research and Practice, 2, 97-101.

Shepperdson, B. (1997). Changes in the school placements of pupils with Down's syndrome. Research in Education, 53, 1-9.

Sloper, P., & Turner, S. (1994). Families of teenagers with Down's syndrome. Parent, child and sibling adaptation. Final Report to ESRC, award No. R000231975. Hester Adrian Research Centre, University of Manchester.

Sloper, P., Cunningham, C. C., Knussen, C., & Turner, S. (1988). A study of the process of adaptation in a cohort of children with Down's syndrome and their families. Final Report to DHSS. Hester Adrian Research Centre, University of Manchester.

Sloper, P., Cunningham, C. C., Turner, S., & Knussen, C. (1990) Factors related to the academic attainments of children with Down's syndrome. British Journal of Educational Psychology, 60, 284-298.

Steele, J., & Stratford, B. (1995) The United Kingdom population with Down syndrome: Present and future predictions. American Journal of Mental Retardation, 99, 664-682.

Turner, S., & Alborz, A. (2000). Young adults with Down's syndrome and their families: Predictors of well-being and adaptation. Final Report to ESRC, award No. R000237012. Hester Adrian Research Centre, University of Manchester.


Received 22 May, 2001; revised version received 20 July, 2002

Appendix A

Academic Index Checklist

This checklist aims to produce a picture of this child/young person's academic skills. Please tick the appropriate column for each question based on your observations of the child/young person. If s/he has grown out of, or gone past a certain behaviour please tick the CAN DO column.

The response columns are:-

CAN'T DO Even with help the child/young person is unable to do this.
CAN DO This means you have seen the behaviour and are sure that the child/young person can do it on his/her own 9 times out of 10.
CAN DO WITH HELP    This means s/he is still learning the skill and needs physical or verbal prompts, but not reminders to do it.

USE OF COMMENT COLUMN: You do not need to write a comment for every item. Sometimes you will not know if the child/young person can do something. Please put DON'T KNOW in this column. Sometimes you know the child/young person can do it but s/he refuses to do it most of the time. Please write in WON'T.

Sometimes there is no opportunity for the child/young person to show s/he has the skill. If you are sure the child can do this if given the opportunity, write in NO OPP.

Reading / Use of Written Information  CAN'T DO   CAN DO  CAN DO
 WITH HELP 
 COMMENT 
Matches pictures of most common objects        
Matches words of up to five letters        
Recognises his/her own name written down        
Recognises five to ten sight words        
Recognises and picks out labels, trade names etc (e.g., brand names for soft drinks, chocolate bars, breakfast cereals)        
Recognises up to 25 sight words        
Reads simple sentences        
Knows most letter sounds of the alphabet        
Reads simple picture books        
Recognises and acts appropriately to written signs, e.g., 'Danger' 'Bus stop' 'Exit'        
Builds simple words through knowledge of letter sounds        
Reads and acts appropriately to signs giving directions, e.g., street names, notices in shops        
Reads and follows a line of instructions, e.g., 'cut along dotted line'        
Uses complex phonics, e.g., digraphs, sound blends etc.        
Reads and follows a sequence of instructions, e.g., recipes, rules of a game        
Reads books, magazines for pleasure        
Reads with understanding to get information, e.g., newspapers, brochures        
 
Writing  CAN'T DO   CAN DO  CAN DO
 WITH HELP 
 COMMENT 
Holds pencil or crayon and attempts to scribble        
Scribbles purposefully with pencil or crayon        
Copies a circle        
Copies letters        
Copies first name        
Writes first name independently        
Writes first name and surname independently        
Copies a simple sentence of four or more words        
Writes a simple sentence of four or more words        
Writes own name, address and telephone number        
Writes simple sentences, when dictated        
Writes more than one sentence on own        
Writes short notes, e.g., messages, shopping list        
Writes short personal letter        
Fills in printed forms, coupons        
Writes and addresses personal letters        
Writes short factual statements, e.g., accounts of events        
Writes short, descriptive passages        
Writes imaginative pieces, creative writing        
 
Number Skills  CAN'T DO   CAN DO  CAN DO
 WITH HELP 
 COMMENT 
Discriminates between largest and smallest groups of objects        
Identifies groups with the same number of objects        
Makes a group of five objects        
Makes a group of nine objects        
When given a group of objects, adds one        
When given a group of objects, takes one away        
Names and matches symbols, 0 to 9        
Writes symbols, 0 to 9        
Writes symbols, 10 to 20        
Adds written numbers up to 10, with materials (e.g., fingers, counting apparatus)        
Adds written numbers up to 10, without materials        
Subtracts from written numbers up to 9, with materials        
Subtracts from written numbers up to 9, without materials        
Adds two numbers up to 20, with materials        
Adds two numbers up to 20, without materials        
Subtracts from numbers up to 20, with materials        
Subtracts from numbers up to 20, without materials        
Adds numbers, 20+, without materials        
Subtracts from numbers up to 50        
Does simple multiplication sums e.g. 3 × 12        
Does simple division sums, e.g., 8 ÷ 4        
Can do more advanced number work. Please specify        

Appendix B

Reliability and validity of the Academic Ability Index measure

Recourse to parents and tutors rather than teachers as informants in the 2000 study raised the possibility of reducing the reliability of the measure. Internal consistency reliability for the three subscales across the three studies is high, varying between. 91 and. 98 (Cronbach alpha coefficients), and there is no evidence that scores based on data from any particular source were less reliable in these terms.

Tutor and parent scores for those young people with 2000 scores from both showed the following correlations: AAI scores: r = .81, p = .01, n = 28: reading: r = .76, p = .01, n = 34; writing: r = .82, p = .01, n = 36; number work: r = .56, p = .01; n = 36 (Spearman rank coefficients). However, t tests of differences in means indicated some differences in levels of scoring, shown in Table 7.


Table 7. Paired samples t test: Tutors' and parents' scores, 2000 data
Measure      Tutors'
  score Mean  
Parents'
  score Mean  
Std.
  Deviation  
        t              df      Sig.
     (2-tailed)     
AAI 24.79 28.14 8.40 -2.12 27 .044
Reading 7.68 9.65 3.45 -3.33 33 .002
Writing 8.89 10.06 2.72 -2.57 35 .014
Number 8.76 9.59 4.68 .95 28 .349

The widest discrepancy in means was in reading skills. Lower scores from tutors may reflect lack of opportunity to assess these skills (missing values were more frequent among tutors' responses). As Pueschel and Hopmann (1993) point out, it is entirely possible that people with Down's syndrome may exhibit setting-related variability in their communication skills. It is also possible that parents' scores may reflect a positive bias in assessing their sons' and daughters' ability.

2000 AAI tutor scores correlated with 1991 AAI teacher scores at r = .84, p = .01, n = 36; 2000 AAI parent scores correlated with 1991 AAI teacher scores at r = .75, p = .001, n = 73 (no parent scores were obtained in earlier studies) (Spearman rank coefficients). Equivalent subscale scores correlated at r = .84 (reading), r = .84 (writing), r = .67(number) for tutor-teacher scores; and r = .75, .76 and .52 respectively for parent-teacher scores (all significant at p = .01 or less, Spearman rank coefficients).

A further check of the measure's reliability was made early in the study by obtaining parallel checklists from parents and teachers of eight young people still at school. Results of this check were satisfactory, with overall scores from the two sources correlating at r = .92 p = .001 (Spearman rank coefficient), and no significant differences between means, t(7) = 1.19, ns. Similar results were obtained for two of the three subscores: reading: r = . 89, p = .01; t(7) = 1.17, ns; writing: r = .97, p = .001; t(7) = 1.21, ns). Number work ratings failed to correlate significantly (r = . 56, ns), but differences between ratings were not significant, t(7) = 0.38, ns (Spearman rank coefficients). The nine items with changed wording for parents had satisfactory inter-rater reliability scores (using the revised scoring). Correlations between parent and teacher ratings on these items ranged from r = .36, p = .05 to r = .85, p = .001 (Spearman rank coefficients). The only non-significant correlation was for the item assessing the young person's ability to recognise brand names. There were no significant parent-teacher differences in means for any of the reading or writing items, which ranged from t(33) = 0.70, p = .49 to t(33) = 1.79, p = 0.08. Parent and teacher ratings on the two reworded items on the number checklist were significantly different, t(34) = 2.10, p = .05 - item 1, and t(33) = 3.19 p = .01 - item 2). However, closer examination of the data revealed that in cases where there were differences of opinion these extended throughout the checklist and were not confined to the reworded items.

There are a number of other biases which could affect the data. Firstly, where a specific age group is composed of children tested either in 1986 or in 1991 (predominantly the 12 year-old age group), it is possible that factors other than chronological age (CA) could have affected scores - for example, changes in educational provision. There was no indication that time of testing (i.e., 1986 or 1991) was related to scores for children in the 12 year-old age group (mean 1986 AAI score 19.00, n = 7, mean 1991 score 21.42, n = 36: t(41) = -0.52, ns).

A second possible source of bias is differences in IQ levels between age groups, which would have implications for comparing AAI scores. To test this, a series of independent sample t tests were performed. IQ scores were calculated at the time of the 1986 study, using McCarthy Scales of Children's Abilities (McCarthy, 1972) or Bayley Scales of Infant Development (Bayley, 1969). The 8-year-old age group had a significantly higher mean IQ than the 10-year-old age group, t(88) = -2.00, p = .05. This difference is repeated for 1991 scores, when the same children largely fell into the 12 and 14 year-old age groups respectively, and again in 2000, in relation to the 20 and 22 year-old groups. The differences in IQ scores between these groups were in the range 5.53-9.26 points. Given the accepted imprecision of IQ scoring, it was felt that differences of less than 10 points in magnitude were unlikely to significantly affect the findings. Figure 2 in the main text compares AAI raw scores with those adjusted for differences in IQ between age groups. Adjusted scores were calculated by dividing individual scores by the mean IQ score for that age group, and then multiplying by 41.51, the mean IQ score for the young people with a 2000 AAI score. Elsewhere, raw scores only are reported.

Finally, children assessed in 1986 but missing in 1991 or 2000 for whatever reason may have scored differently on the checklist or have other important differences (IQ, CA). For example, Carr (2000) reports a 'survivor effect' in her longitudinal study due to deaths in the cohort, which biased the group towards higher ability over time. Differences in mean 1986 AAI scores between those with valid 2000 AAI scores and those with missing 2000 scores were tested using independent sample t tests. Those with missing 2000 AAI scores had a mean 1986 AAI score of 9.36 (n = 25) compared to 14.00 (n = 77) for those with valid 2000 scores. This difference just reached statistical significance, t(100) = -2.02, p = .046. There were only three missing 1991 scores for children with valid 1986 AAI scores, and so these tests were not repeated for these cases. The results of these tests imply that a bias towards higher abilities over time reported by Carr (2000) is also likely in this study.

Two responses to this situation were considered. The first was to include cases with valid early scores but missing later scores, which would result in a steeper gradient in scores over time, as lower scorers drop out. The second was to exclude these cases from all analyses, which would inflate early AAI scores, without removing the bias towards higher scorers in older age groups. It was decided that for this reason the first course was preferable.

Other comparisons were made to establish whether the sample had become biased by attrition in any significant way. The missing in 2000 group had lower 1986 IQ scores (36.7 cf. 41.5), but the difference did not quite reach significance, t(101) = -1.98, p = .051). No significant differences were found between 2000 respondents and non-respondents in terms of earlier data on gender, age, personal and community self-sufficiency, behaviour problems or health of the young person. There were no differences in mother's age, educational background, stress level, father's occupational classification, educational background or stress level, or family housing or car ownership.

Support for the reliability and validity of the measure also comes from the results of two smallscale studies which used the AAI with schoolchildren with Down's syndrome. Philps (1993) obtained teacher reports for 30 children aged 5-11 split evenly between mainstream and special schools. She reports acceptable levels of internal consistency (Cronbach alphas of between 0.89 and 0.93 for the three subscales). Nye et al. (1995) report a significant correlation between AAI number subscale scores and scores on two standardised measures of numerical ability (The British Ability numerical scale and the Kaufman Assessment Battery), in a sample of 16 children with Down's syndrome aged between 7 years and 12 years 6 months. They conclude that this supports the construct validity of the AAI, and report that it may be particularly suitable in assessing children who do not perform well in conventional test situations.


¹Special education for children in the UK has typically been provided at SLD schools catering for children with IQs below 50 or at MLD schools for children whose IQ is in the range 50 to 70 or who are unable to cope in a mainstream classroom.