Standardisation

The nine subtests in LASS 8–11 have been standardised so that teachers using the system can establish where any given student falls on any of the subtests, in relation to the population norms. This means that direct and meaningful comparisons can be made between the individual subtests that a single student takes. In addition, direct and meaningful comparisons can be made between students as well as between the student and national norms.

LASS 8–11 underwent a full national re-standardisation in May – July 2019. The standardisation was conducted in 44 schools (England n = 33; Northern Ireland n = 9; Scotland n = 1; Republic of Ireland n = 1). Of those schools where an Ofsted assessment has been published, 25% were rated as Outstanding, 63% were rated as Good and 12% were rated as Requiring Improvement (which compares reasonably well to national figures for the 2018/19 academic year: 20% Outstanding; 66% Good; 11% Requires improvement). The number of students on the roll for the sample schools ranged from 30 to 683, with an average of 273.

School characteristics (where these were available on Gov.uk or the equivalent websites for Scotland, Northern Ireland and the Republic of Ireland) for the sample schools were compared to the national average (for English state-funded Primary schools) – see Table 2. It can be seen that the schools overall included a slightly higher proportion of girls than the national average and a slightly lower proportion of pupils with an ECHP than the national average.

 

Table 2. Characteristics of schools within the standardisation sample

Within the selected schools, students were included in the standardisation on an entire class basis, to avoid any selection bias. The number of students completing each subtest, within each age group of the standardisation sample, are shown in Table 3.

 

Table 3. Students per age group for each subtest

Demographic information concerning the students within the standardisation sample are given in Table 4 (note that information was not provided for all students). Population parameters are also provided, but these are based only on English state-funded Primary schools, whereas the sample also includes students from Northern Ireland, Scotland and the Republic of Ireland, so the comparisons are limited. It can be seen that the sample included a slightly higher proportion of female students than the national average for English state-funded Primary schools. With regards to ethnicity, the sample has a higher proportion of Asian students than is found in the population and lower proportions of White, Black, Mixed and Other students, although ethnicity information was not provided for 10.1% of the sample. The number of students within the sample who are eligible for Free School Meals is slightly higher than in the population. However, it should be noted that the national average for Northern Ireland (where 9 of the schools were based) is 29.4%, which may account for the higher proportion of students eligible for FSM within the sample. With regard to language, the percentage of students within the sample speaking English as an Additional Language is very close to the population average. The proportion of students within the sample with a diagnosed SEN is slightly higher than within the population, whereas those with an Education, Health and Care plan reflects the national average. Again, it should be noted that the national average for SEN in Northern Ireland is 21.0%, which may account for the slightly higher proportion of students with SEN within the sample.

 

Table 4. Demographic details of sample

* Based on DfE school census data for English state-funded Primary schools, January 2019

Of the standardisation sample, 84% undertook the tests using desktop computers, whilst 16% used tablets. Analysis shows no evidence of a platform effect (with the exception of a negligible effect (Cohen’s d = 0.13) on Non-verbal reasoning, where desktops produced slightly higher scores than tablets).

The sample data has been weighted according to age, gender and SEND proportion against population parameters. Using a non-parametric age-standardisation model, the raw scores for each age group were transformed into Standardised Age Scores (SAS) with a mean of 100 and a standard deviation of 15. This builds on previous work conducted by Schagen (1990). SAS scores for the subtests range from 65 to 135, although on Single word reading, where there is a ceiling effect, the SAS is capped at 110 at the upper end (note that the cap does not distort the scoring at the lower end of abilities, which is what we are most commonly interested in).

Table 5 shows the correlations between all LASS 8–11 subtests. The correlations range from .310 to .732, with the majority being within the moderate range (.4 to .6). All correlations are significant at the p<.001 level.

 

Table 5. Intercorrelations between subtests

*all correlations are significant at p<.001; (N is shown in brackets)

In order to check for any gender bias, comparisons were made between males and females (where gender had been identified) on each subtest (see Table 6). Small effects were found on Verbal reasoning and Mobile phone (auditory sequential memory), with both subtests slightly favouring girls. There were no other gender effects.

 

Table 6. Gender differences

* Cohen’s d is a measure of effect size of the difference between two means

Checks were also made for ethnic group bias. Due to the small numbers in some ethnic minority groups, comparisons were made between White students and Other ethnic groups combined (where ethnicity had been identified) on each subtest (see Table 7). A small effect was found on Verbal reasoning, slightly favouring White students. There were no other ethnicity effects.

 

Table 7. Ethnic group differences

* Cohen’s d is a measure of effect size of the difference between two means