The Scottish Government

« Previous | Contents | Next »

Listen

SEED Sponsored Research: Children starting school in Scotland

5 Do children starting school in Scotland have the same age related development profile as children in England, Western Australia and New Zealand?

5.1 Comparison Of The Starting Point Of Children With Three Other Countries

In this section the starting points of Scottish children are compared with those from England, New Zealand and Western Australia. The study was restricted to children whose first language was English, which gave the sample sizes indicated in Table 6 (Samples of children from 2003 whose first language was English).

Table 6 Samples of children in 2003 whose first language was English

The data from England and Scotland have been shown to be representative of their countries but the Western Australia and New Zealand samples were not guaranteed to be representative of those countries as a whole. The schools in those two countries joined the project in much the same way as in England and Scotland and one might expect that they would be representative, but this has yet to be demonstrated.

Before starting the comparative analyses we used Rasch analyses to scale the items in the four sub-tests independently within each country. We then compared the relative difficulties of the items for children in Scotland with other countries. The purpose was to identify any items in the assessments that functioned differently in the four countries. The results showed very few differences. The strongest agreements were in mathematics, where the correlations of the difficulties of the items in the four countries were all 0.99. This is so high that no further preliminary action was needed before making comparisons.

The difficulties of the reading items were also strongly related but not quite so strongly. The correlations are shown in Tables 7, 8 and 9.

Table 7 Correlation between difficulties of 56 reading items⁴

** Correlation is significant at the 0.01 level (2-tailed).

The lowest correlation was for the items difficulties between Scotland and New Zealand but the figure (0.94) was felt to be sufficiently high to allow further analysis to proceed.

A similar analysis for the vocabulary items showed that two words were particularly difficult in the antipodes compared with the UK. They were pigeon and wasp and therefore these were excluded from further analysis. The correlations, excluding these items, are shown in Table 8.

Table 8 Correlation between difficulties of 17 vocabulary items⁵

** Correlation is significant at the 0.01 level (2-tailed).

As with the reading correlations it was felt that the lowest figure (0.95) was sufficiently high to proceed with further analysis.

The phonics section was less satisfactory as the correlations in Table 9 show.

Table 9 Correlation between difficulties of 17 phonics items⁶

** Correlation is significant at the 0.01 level (2-tailed).

The correlations between the item difficulties for Scotland and the other countries might appear to be high for social science work but figures of 0.91 leave something to be desired as a basis for comparing countries and the correlations with Scottish data were the lowest in the table. Further, when the difficulties of specific items were compared it was not possible to pick out just one or two items as being problematic with a view to dropping them from the analysis. The point of this analysis was to ascertain if any items seemed to be culturally biased and relatively more easy or difficult for children in a particular country. The items in the phonics section gave different outcomes for pupils in Scotland compared with pupils in England and New Zealand and as a result it was thought to be inappropriate to proceed with comparison of phonological awareness.

For each of the three other sub-tests (mathematics, reading and vocabulary) the datasets for Western Australia, New Zealand, England and Scotland were combined and the Rasch scales were created. The children were then put into eleven age categories corresponding to increments of 3 months each. The lowest had a mean age of 4.2 years and the highest of 6.6 years. The average scaled scores and 95% Confidence Intervals were then plotted against age to produce Figures 7, 8 and 9 below. The first thing to notice from the figures is the difference in the age of children starting school in different countries. The pupils in England tend to be the youngest although there is some overlap with Scotland and New Zealand. The New Zealand sample is quite interesting in that all children start school immediately after their fifth birthday. The three charts show generally very similar patterns indicating that the cognitive growth patterns of children whose first language is English is similar in Scotland when compared with England, Western Australia and New Zealand. But, despite the general pattern, there were differences and these are examined in more detail below.

The vocabulary scores rise steadily with age. When the error bars (95% Confidence Intervals) are taken into account the Scottish data in blue are seen to be very much in line with the data from the other three countries, which are also in line with one another. There is suggestion that the vocabulary scores of the younger children are higher than similar children from England but this is a small effect. It amounts to an Effect Size of 0.2 or about three and a half months of development.

Figure 7

Figure 8

The reading chart shows some clear differences from the vocabulary chart. The Scottish data now fall increasingly behind the data from the three other countries as the age increases. The English and Western Australian results form an unbroken continuum and the New Zealand scores are higher. The younger children are in line with the scores of children of the same age in England but for the oldest children there is a discrepancy of 0.38 standard deviation units (Effect Size) or about just over five months of development compared with similarly aged children in Western Australia.

Figure 9

The pattern for mathematics is very similar to the pattern for reading. The English and Western Australian data appear to form a continuum and the Scottish data show increasing discrepancies from Western Australia for older children. For the oldest group starting school in Scotland there is a discrepancy of about 0.31 standard deviation units (Effect size). This is equivalent to nearly four months of development. Unlike the reading data, the New Zealand results are in line with the results for the other countries.

5.2 Further investigation

Factor analyses of the PIPS BLA maths and reading scores suggested that those things which would be expected to be taught at school, such as word recognition, reading simple sentences and more formally presented maths problems, and which are usually not taught to children before the start of school form a different factor from those that are acquired more naturally (Tymms 1999). By developing 'more naturally' we mean that for areas such as vocabulary, a child is interacting with their environment and engaging in conversations with other children and adults. An understanding of the concepts of print is once again developed from interaction with adults and exposure to a range of literature. Other sections of the PIPS BLA (Repeating Words, Rhyming Words, Ideas about Maths, Counting and Numerosity, Shape Identification, Addition and Subtraction problems presented without formal notation, and Single Digit Identification) reflect areas that will develop as a consequence of a child's interactions with their environment, adults and other children.

It was therefore hypothesised that the higher scores of older children outwith Scotland were the result of more formal teaching in the pre-school setting. This would fit with the findings reported in the PIRLS encyclopaedia written in 2001/2 which compares pre-school curricular among the PIRLS countries. In an effort to explore this hypothesis, the scores on sub-tasks that contributed to the reading and mathematics scores were further investigated by plotting the same kinds of diagrams shown above but for actual (raw) scores rather than Rasch scaled scores. Figures 10-16 below show the outcomes of this analysis. It was expected that the Scottish children would show the greatest discrepancies for formal sums and the reading of words, sentences and passages.

Figure 10	Figure 11

Figure 12	Figure 13

Figure 14	Figure 15

Figure 16

It had been expected that the Scottish children would show the greatest discrepancies for formal sums and the reading of words, sentences and passages since these might have been specifically taught. However, the charts do not support this hypothesis. The most formal parts of the assessment (formal sums, the identification of numbers greater than nine and the reading of words, sentences and paragraphs) showed little difference from Western Australia, although the scores from England and New Zealand were a little higher for older children. Further, the identification of single digit numbers, counting and simple informal sums were also more or less in line with the results for England and Western Australia, although once again the scores for Scotland were a little lower for older children.

But the greatest discrepancies were for Concepts about Print and letter identification. Interestingly, Concepts about Print were particularly high for New Zealand, the country in which the work of Marie Clay, who pioneered Reading Recovery and invented the Concepts about Print scale, has been so important.

5.3 Summary: Comparison of Scottish children starting school with three other countries

Scottish children start school at the age of 5 years on average. This is six months later than in England but the same as in New Zealand. In that country however, children start on or after their fifth birthday whereas in Scotland they generally start between the ages of 4.5 and 5.5. In Western Australia they start about six months later.

After taking age into account the starting points of children in the four areas is very similar. However, older Scottish children appear to be behind in mathematics and early reading but not in vocabulary.

5.4 Interpretation

It is not easy to come to clear conclusions as to why the data for Scotland should differ in some cases quite markedly from other countries. However, earlier in this report it was noted that the relationship between the outcome measures on the baseline assessment and pre-school experience was weak, and it could be that there is a connection here. It may be that the kind of pre-school provision being provided outside Scotland is different from the kind of provision provided within Scotland, but without further information about the pre-school curricula of the different countries it is difficult to draw any conclusions. It is also difficult to know exactly what the long-term consequences of the differences seen above are likely to be. However, the data do raise questions about possible differences between pre-school in Scotland and the other countries.

« Previous | Contents | Next »