Growing Up in Scotland: changes in language ability over the primary school years

2 Methods

2.1. Sample and data overview

The analysis presented in this report uses data from the first GUS birth cohort (Birth Cohort 1 or 'BC1'). BC1 is comprised of a nationally representative sample of 5217 children living in Scotland when they were 10 months old who were born between June 2004 and May 2005.

Starting in 2005/06, data were collected annually from when the children were aged 10 months until they were just under 6 years old, and then biennially at age 7-8 and when the children were in Primary 6 (age 10-11). At the time of writing (2018), the ninth sweep of face-to-face data collection with this cohort has finished. At this ninth sweep the cohort children were in their first year of secondary school (age 12-13). This report draws primarily on data collected, firstly, at the time the children were aged just under 5 when most children were in Primary 1 or about to enter primary school (in 2009/10), and secondly, at the time they were in Primary 6 and aged 10-11 (in 2014/15), although for a number of the factors examined in chapter 4 it also draws on data collected out with these two time points (see Table 2-1 for details). Because the cohort is comprised of a nationally representative sample of children the results should be understood to represent all children of the respective age living in Scotland at the time point in question who were also living in Scotland when they were 10 months old. For example, the results presented for the GUS children at the time they were in Primary 6 are roughly representative of all children in Scotland who attended Primary 6 in 2014/15^[5].

The main data collection on GUS takes place through annual or biennial 'sweeps' of face-to-face interviews with children and parents in their homes. This report draws on data collected from several sources. First, it draws on data collected from the cohort child's main carer at various age points. Second, it draws on objective measures of the child's vocabulary at the time most children were in or about to enter Primary 1 and when they were in Primary 6 (see further details in section 2.2). Third, it draws on data collected from the children themselves when they were aged 8, around the time most children were in Primary 4. Finally, it draws on administrative data concerning the child's Primary 1 school (further details are provided in Appendix A).

Table 2-1 provides an overview of the sources of data used in the report. Note that the analysis draws on data for children who took part in language assessments at both time points only (n = 2944) - that is, children who undertook language assessments both around their 5th birthday and when they were in Primary 6. Furthermore, data were weighted using the GUS longitudinal survey weight, meaning that only cases which have taken part in every face-to-face sweep of GUS up to and including sweep 8 were included. In total, 2726 children were included in the analysis.

Table 2‑1 Data overview

2.2. Expressive language ability

Put simply, language development refers to children's use of words, sentences, gestures and vocalisations to convey meaning, communicate with others and gain knowledge (Law et al., 2017). The ability to use language underpins many aspects of children's activities, including their social interactions and intellectual pursuits, and thereby impacts on various elements of their non-physical development. For example, it contributes to their ability to manage emotions, communicate feelings, form and maintain relationships and read and write. Consequently, as highlighted by Save the Children's (2014) 'Read On. Get On' campaign, solid foundations in early language are the foundation on which children's future education and learning are based. Indeed, a Save the Children study using data from the Millennium Cohort Study (MCS) demonstrated that while socioeconomic disadvantage predicted children's academic performance, 'the most important factor in reaching the expected levels in English and Maths at seven was children's language skills at age five', which was greater than the link to poverty or poor parental education (Finnegan et al., 2015).

The importance of early language development for children's educational attainment has meant that there is also a substantial body of evidence linking low levels of early literacy to poor outcomes in adulthood. For example, research has documented correlations between poor early language development and poor labour market outcomes such as low pay and unemployment (Howieson and Iannelli, 2008; McIntosh and Vignoles, 2001). In addition, a study using data from a UK birth cohort of over 17,000 children born in 1970 found that those with poor vocabulary skills at age 5 were four times more likely to have reading difficulties, three times more likely to have mental health problems and twice as likely to be unemployed by the time they were 34, when controlling for other factors (Law et al., 2009). There is also evidence to suggest that poor early literacy can also be a risk factor associated with criminal behaviour in adulthood (Devitt, 2011; Institute of Education, 2002).

In this report, the focus is on expressive language ability, or vocabulary. As outlined above, children's early vocabulary ability has been found to be associated with later outcomes across several parameters (e.g. Law et al., 2009).

In GUS, expressive language ability has been measured three times for children in BC1: when they were just under 3 years old, when they were just under 5 years old and again when they were in Primary 6 (aged 10-11). In this report we focus on the measures obtained at the latter two age points. Across these two age points, the children's vocabulary was measured using two different assessments. These are described below.

As part of the fifth sweep of interviews undertaken with families in BC1 (when the children were aged just under 5 and most were in or about to enter Primary 1), the child's language ability was measured using the naming vocabulary subtest of the British Ability Scales Second Edition (BAS-II). This is a cognitive assessment which forms part of the Early Years battery designed for children aged between 2 years and 6 months and 7 years and 11 months. Though numerous tests of language ability exist, the BAS is particularly suitable for administration in a social survey like GUS. The naming vocabulary test requires the child to name a series of pictures of everyday items to assess their expressive language ability. There are 36 items in total in the assessment. However, to reduce burden and to avoid children being upset by the experience of repeatedly failing items within the scale, the number of items administered to each child is dependent on their performance. For example, one of the criteria for terminating the naming vocabulary assessment is if five successive items are answered incorrectly. As already noted, children in BC1 were asked to complete this assessment when they were just under five years old. As such, the BAS assessment scores offer a snapshot of children's expressive vocabulary ability around the time they started primary school.

When the GUS children were in Primary 6, as part of their eighth GUS interview, their language ability was measured using the listening comprehension subtest of the Wechsler Individual Achievement Test - Second UK Edition (WIAT-II^UK). More specifically, the expressive vocabulary measure of the subtest was used, which assesses speaking vocabulary and word retrieval ability. This subtest is part of a comprehensive individually administered test for assessing the achievement of children and adolescents aged between 4 years and 16 years and 11 months. As with the BAS, WIAT is also suitable for administration in a study like GUS, with the version used especially adapted for social surveys. During the expressive vocabulary element of the listening comprehension subtest children are shown a single picture and given an oral description. They then have to provide one word that matches the picture and the description. The assessment has a total of 15 items. As with the BAS, to reduce the burden on the child the number of items administered is dependent on their performance, with the assessment discontinued if the child gives a wrong answer on 6 consecutive occasions. Because children in BC1 were asked to complete this assessment when they were in Primary 6, the WIAT assessment score provides a picture of children's expressive language ability at the time they were nearing the end of primary school.

To make the scores from the two measures of expressive language ability comparable, the vocabulary score from each age point was standardised into a z-score. Z-scores are derived from the survey data. They count the number of standard deviations from the score mean and have a mean of 0. Therefore, a child with a z-score of 0 at either Primary 1 (age 5) or Primary 6 (age 10-11) has an average ability across all children in that age group. Those with a z-score greater than 0 scored above average and those with a score of less than 0 scored below average. The size of the z-score indicates how far above or below average the child's score was. By using the standardised scores, it is possible to compare ability at the two age points and to consider whether children who scored above, below or about average around the time they entered primary school (aged just under 5) continued to do so when they were in Primary 6.

Throughout the remainder of this report, the terms 'expressive language ability', 'language ability', 'expressive vocabulary ability' and 'vocabulary ability' will be used interchangeably.

2.3. Analytical approach and interpreting the results

Much of this report is concerned with exploring expressive language ability for different groups of children according to a number of socio-economic characteristics (annual equivalised household income; highest level of parental education in the household; and area deprivation (SIMD)). Definitions of these measures are provided in Appendix A.

Not all families who initially took part in GUS did so for all subsequent sweeps. There are a number of reasons why respondents drop out from longitudinal surveys and such attrition is not random. Therefore, the data were weighted using specifically designed weights which adjust for non-response and sample selection. All results have been calculated using weighted data and all comparisons take into account the complex clustered and stratified sample structures. Note that because results were calculated using weighted data, the results and bases presented cannot be used to calculate how many respondents gave a certain answer.

Unless otherwise indicated, only differences which were statistically significant at the 95% level or above are commented on in the text.

Notes on how to interpret tables and charts are provided in the text. A brief description of the analysis undertaken is also provided in the text. However, readers interested in the analytical approach should refer to Appendix C.