GROWING UP IN SCOTLAND: THE CIRCUMSTANCES AND EXPERIENCES OF 3 YEAR OLD CHILDREN LIVING IN SCOTLAND IN 2007/08 AND 2013

Chapter 4 Child Development

This section focuses on two areas of child development: physical development milestones and cognitive ability.^[13] These measures are important indicators of developmental progress. Previous research on GUS has demonstrated stark differences in these measures according to a child's socio-economic background from an early age. Gender is also known to be strongly associated with differences in early child development; as such this has been included as an additional comparison for this section.

4.1 Physical development milestones

Parents were asked whether or not the cohort child was able to achieve the following developmental tasks by the time of the interview (shortly before the child's third birthday):

• Walk up steps like an adult (alone or with help)
• Balance on one foot
• Hop on one foot
• Undo big buttons
• Draw a circle
• Copy a square
• Drink from a cup
• Brush teeth without help
• Put on a t-shirt without help
• Get dressed without any help.

In both cohorts, almost all children were able to drink from a cup (98%) and walk up steps like an adult either on their own (89%) or with help (9%). A large proportion were also able to balance on one foot (86%). Most children were also able to undertake the remaining tasks. The exceptions were copying a square, drawing a circle, putting on a t-shirt without any help and getting dressed without any help (38%), where fewer than half of children were reported as being able to complete these tasks (Table 4.1).

In five out of the ten milestones included, there were no statistically significant differences in the proportion of children in each cohort who were able to complete the task. The remaining five (indicated in Table 4.1) did show some differences between cohorts which were statistically significant. On one of these - brushing teeth without help - the size of the difference is very small. However, the remaining measures show a greater - though still small - change. The pattern of change is not consistent in direction nor in the developmental area which the task measures. On two milestones - drawing a circle and getting dressed without help - the proportion of children able to complete the task is lower in BC2 than in BC1 by around 4-5 percentage points. For the other two milestones - copying a square and putting on a t-shirt - the opposite is true with the proportion in BC2 being higher than BC1 by around 5-6 percentage points.

Table 4.1 Percentage of children able to complete selected developmental tasks

	BC1	BC2
	%	%
Walk up steps like an adult (alone or with help)	98	98
Balance on one foot	87	86
Hop on one foot	75	75
Undo big buttons	78	77
Draw a circle*	24	19
Copy a square*	39	44
Drink from a cup	99	98
Brush teeth without help*	96	95
Put on a t-shirt without help*	27	33
Get dressed without any help*	42	38
Unweighted bases	4094	4931

Differences by cohort on items marked * are statistically significant at p < .001. All other differences are not statistically significant.

For the purposes of comparison, a summary variable was created which counted the number of tasks the child was unable to accomplish. 14% of children could complete all tasks, 20% were unable to complete one, 22% were unable to complete two and 44% were unable to complete three or more (Table 4.2). There was no statistically significant difference between cohorts on this measure.

Table 4.2 Number of developmental tasks not achieved, by cohort

	BC1	BC2
	%	%
None	14	14
One	22	20
Two	22	22
Three or more	42	44
Unweighted bases	4032	4887

Differences by cohort are not statistically significant.

Table 4.3 shows that girls were more likely than boys to be able to achieve all tasks - 19% could do so compared with 9% of boys.

In contrast, boys were more likely than girls to be unable to complete three or more tasks - 53% compared with 31% (Table 4.3). Differences by gender are statistically significant - in both cohorts, boys were more likely than girls to be unable to complete three or more tasks.

Table 4.3 Number of developmental tasks not achieved, by gender and cohort

	Boys	Girls
	%	%
BC1
None	9	19
One	17	27
Two	21	23
Three or more	53	31
BC2
None	9	19
One	16	23
Two	21	24
Three or more	54	33
Unweighted bases - BC1	2064	1968
Unweighted bases - BC2	2471	2416

Tested on category one: differences by gender - p < .001; differences by cohort p < .05; cohort*gender p = NS.

The proportion of children who could complete all tasks did not vary much across the sub-groups considered. Most variation occurred in the proportion of children who were unable to complete three or more of the tasks. In terms of household income, in both cohorts the proportion of children unable to complete three or more tasks increased as household income increased. Differences by income were statistically significant. The proportion of children in each income group unable to complete three or more tasks was very similar in both cohorts. Any differences shown between cohorts were not statistically significant.

Table 4.4 Number of developmental tasks not achieved, by equivalised household income (quintiles) and cohort

	Lowest quintile	2nd quintile	3rd quintile	4th quintile	Highest quintile
	%	%	%	%	%
BC1
None	15	14	13	13	12
One	25	21	24	19	19
Two	23	23	21	22	22
Three or more	36	42	42	46	46
BC2
None	14	17	12	13	12
One	20	20	20	19	20
Two	23	22	24	23	22
Three or more	43	40	44	45	47
Unweighted bases - BC1	754	772	731	831	694
Unweighted bases - BC2	954	763	796	762	1000

Tested on category 'three or more': differences by income - p < .001; differences by cohort p = NS; cohort*income p = NS.

The relationship between parental level of education and development was different in each cohort (Table 4.5). In BC1, children whose parents were degree educated were more likely than those with any other qualifications to be reported as unable to complete three or more tasks. These differences are statistically significant. In BC2, differences by parental level of education were not significant. There were some small differences between comparable sub-groups in each cohort, which were statistically significant. However, there is no consistent pattern of change across all sub-groups nor an obvious trend by level of education.

Table 4.5 Number of developmental tasks not achieved, by parental level of education and cohort

	No qualifications	Lower Standard Grades or VQs or Other	Upper level SGs or Intermediate VQs	Higher grades and upper level VQs	Degree level academic and vocational qualifications
	%	%	%	%	%
BC1
None	14	12	15	16	11
One	31	20	24	22	20
Two	17	26	21	24	21
Three or more	39	41	41	39	47
BC2
None	18	15	16	15	13
One	12	20	19	21	19
Two	26	23	25	21	22
Three or more	44	43	40	43	47
Unweighted bases - BC1	189	197	770	1342	1527
Unweighted bases - BC2	152	203	701	1403	2262

Tested on category 'three or more': differences by education - BC1 p < .001, BC2 p = NS; differences by cohort p < .001; cohort*income p < .01.

Children with older mothers were more likely to be unable to complete three or more tasks than those with younger mothers. 52% of children whose mother was aged 40 or older were unable to complete three or more tasks compared with 41% of those whose mother was under 20 at the birth and 42% of those whose mother was in her twenties (Table 4.6). Differences by maternal age were statistically significant. There were no statistically significant differences between cohorts.

Table 4.6 Number of developmental tasks not achieved by maternal age at child's birth and cohort

	Under 20 years old	20 to 29 years old	30 to 39 years old	40 or older
	%	%	%	%
BC1
None	14	14	13	13
One	27	23	20	22
Two	21	23	22	16
Three or more	38	40	44	49
BC2
None	17	15	13	14
One	20	20	19	16
Two	22	23	22	19
Three or more	41	42	46	52
Unweighted bases - BC1	220	1516	2133	143
Unweighted bases - BC2	213	1927	2506	219

Tested on category 'three or more': differences by maternal age - p < .001; differences by cohort p = NS; cohort*income p = NS.

The proportion of children unable to complete three or more tasks increased as area deprivation decreased (Table 4.7). 41% of children living in areas in the most deprived quintile could not complete three or more tasks compared with 49% of those living in areas in the least deprived quintile. Differences by area deprivation were statistically significant. This pattern is evident in both BC1 and BC2. However, in all deprivation groups other than the fourth quintile, there was an increase of around three percentage points between BC1 and BC2 in the proportion of children unable to complete three or more tasks. Nevertheless, as noted earlier, there was no statistically significant overall rise in the proportion of children who were not able to complete three or more tasks.

Table 4.7 Number of developmental tasks not achieved, by area deprivation (quintiles) and cohort

	1 Most deprived	2	3	4	5 Least deprived
	%	%	%	%	%
BC1
None	14	15	16	11	12
One	25	21	22	20	20
Two	23	22	20	23	23
Three or more	37	42	41	46	46
BC2
None	15	16	14	11	14
One	21	19	20	21	16
Two	23	21	22	24	21
Three or more	41	44	44	44	49
Unweighted bases - BC1	813	672	843	834	870
Unweighted bases - BC2	920	905	1001	1040	989

Tested on category 'three or more': differences by area deprivation - p < .05; differences by cohort p < .05; cohort*area deprivation p = NS.

Analysis of data on other health and development indicators tends to show poorer health and development amongst children in more disadvantaged circumstances. In this instance, and consistently for both cohorts, the relationship is the opposite; children in more disadvantaged circumstances are reported to have more advanced development than those in more advantaged circumstances. This may be a result of the nature of the questions themselves and how parents choose to answer them or the particular tasks which they cover. To explore this further, additional analysis was undertaken to examine the extent of any differences by household income on each of the individual tasks.

The results^[14] indicated that two tasks were the main drivers of the differences in the summary variable: whether the child could put on a t-shirt alone and whether the child could get dressed alone. On each of these items, and in each cohort, as income increased the likelihood of a child being able to complete the task decreased (Table 4.8). Given the nature of these items - each involving the child getting dressed - differences here may be more a reflection of parenting practices and/or the wider context of the household rather than down to differences in the child's development. For example, parents in higher income households are more likely than those in lower income households to be employed and to work longer hours. In these circumstances, parents have less time to allow a child to dress themselves and may help them more often.

Table 4.8 % of children who could achieve selected tasks, by equivalised household income (quintiles) and cohort

	Lowest quintile	2nd quintile	3rd quintile	4th quintile	Highest quintile
	%	%	%	%	%
Put on a t-shirt without help*
BC1	78	74	72	69	67
BC2	69	70	63	64	64
Get dressed without any help*
BC1	51	43	40	37	35
BC2	42	43	33	32	34
Unweighted bases - BC1	773	798	755	857	718
Unweighted bases - BC2	968	774	809	773	1019

Put on t-shirt - tested on category 1 'yes': differences by income - p < .001; differences by cohort p < .001; cohort*income p = NS.
Get dressed - tested on category 1 'yes': differences by income - p < .001; differences by cohort p = NS; cohort*income p = NS.

4.2 Cognitive ability

Cognitive ability was measured in both cohorts via two assessments: the naming vocabulary and picture similarities subtests of the British Ability Scales (BAS). These two assessments measure, respectively, language development and problem solving skills. The assessments are individually administered. Numerous tests of ability and intelligence exist but the BAS is particularly suitable for administration in a social survey like GUS. Children in BC1 were administered tests from the BAS second edition (BAS-II). This version was updated between cohorts. As such, children in BC2 were administered tests from the third edition (BAS-III). Both editions are virtually identical in terms of administration protocols. However, the particular items within each test vary slightly. This has implications for producing comparative scores, which are discussed below.

The naming vocabulary assessment measures a child's language development. The test requires the child to name a series of pictures of everyday items and assesses the expressive language ability of children. The picture similarities assessment measures a child's problem solving ability (or non-verbal reasoning). In the assessment children are shown a row of four pictures on a page. They are asked to place a free-standing card with a fifth picture underneath the picture with which the card shares a similar element or concept. There are a little over 30 items in total in both assessments. 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. Children were not provided with any feedback on their individual performance.

For analysis purposes, the child's raw score (that is, the count of the number of items they answered correctly) is converted to an ability score. The ability score reflects the range and, more importantly, the difficulty of the specific items a child is asked. Look-up tables for the transformation from raw scores to ability scores are provided in the BAS testing materials. The ability scores are then adjusted for the child's age at the time of assessment using scores from the 'norming' sample which are also supplied with the assessment materials. Standardising the scores in this way avoids older children obtaining higher scores due to their more advanced stage of cognitive development and greater educational experience, rather than their ability. In practice, the vast majority of children in GUS were assessed within a month or two in age.^[15] Nevertheless, the standardised test score takes account of how a child is performing on the sub-tests in relation to other children of the same age.

The standardised scores for each of the sub-tests have a mean of 50 and standard deviation of 10, and the scores are bounded between 20 and 80. A child whose standardised ability score is equal to the norming sample will have a score of 50, a child with a score of 40 has an ability score one standard deviation below the mean score of the norming sample, and a child with a score of 60 has an ability score that is one standard deviation above the norming sample.

Because of differences in the content of the BAS-II and BAS-III assessments, the BAS-II scores for BC1 need to be adjusted before they can be compared with the BAS-III scores for BC2. This was done using information supplied by the assessment authors. Note that, because of this adjustment, it is not possible to convert differences in average cognitive ability scores to developmental age in months, as has been done in a previous GUS report (Bradshaw, 2011).

The mean standardised scores on each assessment for each cohort are shown in Table 5.9. As the data show, children in BC2 had a slightly higher vocabulary score than children in BC1. This difference is statistically significant. There was no difference in problem solving ability. As noted above, whilst children in each cohort undertook different editions of the BAS assessments, a comparative score was derived through consultation with the assessment developers. As such, this increase is not considered to be a function of different editions of the assessments being used.

The increase in average language ability between the cohorts warrants some further investigation. As noted in chapter three, there are some key differences in the socio-economic characteristics of parents of 3-year-olds in Scotland - most notably, an increase in the proportion of parents qualified to degree-level - which may have impacted on average language levels. We know from previous GUS analysis that parental education is highly correlated with their child's language development. It is possible then that the improvements we see at age three have been driven by differences in the characteristics of parents.

The differences in early language ability between the cohorts are being considered in more detail in a separate report (Bradshaw et al, forthcoming). That report considers a range of other differences in circumstances and experiences between the cohorts which may have influenced language ability, such as the frequency of parent-child activities and the introduction of national interventions aimed at improving early development such as Play, Talk, Read. Analysis undertaken for that report showed that after differences in level of parental education between cohorts are controlled for, children in BC2 are still more likely than those in BC1 to have a higher vocabulary ability.

The analysis below, which explores the extent of change between cohorts amongst children in all socio-economic sub-groups, provides some further insight into the differences between cohorts. In the main, these show that differences in average ability between children of different demographic and socio-economic characteristics are broadly similar in both cohorts. Girls and those in more advantaged circumstances show higher average ability than boys and those in less advantaged circumstances, with only small changes in the difference between children in the most and least advantaged groups.

Table 4.9 Mean standardised cognitive ability scores, by cohort

	BC1	BC2
Naming vocabulary	47.6	50.0
Unweighted bases	3930	4625
Problem solving	46.2	46.0
Unweighted bases	3946	4678

Differences by cohort: naming vocabulary p < .001, problem solving p = NS.

In both cohorts, and for both assessments, girls scored higher on average than boys (see Table 4.10). Differences by gender are statistically significant on both assessments. Between cohorts, there was a statistically significant increase in the mean vocabulary score for both boys and girls. There were no statistically significant differences in problem solving ability for either sex.

Table 4.10 Mean standardised cognitive ability scores, by gender and cohort

	Boys	Girls
Naming vocabulary
BC1	45.8	49.4
BC2	48.5	51.4
Unweighted bases - BC1	1980	1950
Unweighted bases - BC2	2310	2315
Problem solving
BC1	45.2	47.3
BC2	44.8	47.2
Unweighted bases - BC1	1993	1953
Unweighted bases - BC2	2345	2333

Naming vocabulary: differences by gender - p < .001; differences by cohort - p < .001; cohort*gender p = NS.
Problem solving: differences by gender - p < .001; differences by cohort - p = NS; cohort*gender p = NS.

As has been found in previous analysis of GUS cognitive ability data (Bradshaw, 2011; Bromley, 2009) there was a relationship between cognitive ability and household income. As income increased ability scores also generally increased for each cohort and each assessment (Table 4.9). Differences by household income are statistically significant.

The increase in vocabulary score between BC1 and BC2 at age 3, occurred across all income sub-groups. As Table 4.11 and Figure 4-A show, in BC2, children in each income sub-group returned a higher average score than their peers in BC1. Furthermore, the level of change was very similar in each group. In Figure 4-A, the level of change between cohorts is indicated by the distance between the two lines. As can be seen, the lines are almost parallel, reflecting that scores in each sub-group increased by a similar margin. Whilst change is slightly greater in the lowest income group than the highest - meaning that the difference in mean scores between the highest groups is a little smaller in BC2 than it was in BC1 (6.7 points compared with 7.8 points) - there was no particular pattern in the extent of the improvement by income level.

Table 4.11 Mean standardised cognitive ability scores, by equivalised household income (quintiles) and cohort

	Lowest quintile	2nd quintile	3rd quintile	4th quintile	Highest quintile
	%	%	%	%	%
Naming vocabulary
BC1	43.2	46.6	48.4	49.9	51.0
BC2	46.7	49.0	50.8	53.2	53.3
Unweighted bases - BC1	711	757	722	815	690
Unweighted bases - BC2	896	700	744	745	988
Problem solving
BC1	42.4	46.0	46.7	48.2	49.1
BC2	44.4	45.9	47.0	46.8	47.8
Unweighted bases - BC1	717	761	722	819	694
Unweighted bases - BC2	908	711	754	748	992

Naming vocabulary: differences by income - p < .001; differences by cohort - p < .001; cohort*income p = NS.
Problem solving: differences by income - p < .001; differences by cohort - p = NS; cohort*income p < .001

Figure 4 A Mean standardised vocabulary scores, by equivalised household income (quintiles) and cohort

Figure 4 B Mean standardised problem solving scores, by equivalised household income (quintiles) and cohort

The pattern of change in problem solving scores amongst income sub-groups between cohorts is different to that seen with vocabulary scores. As noted earlier, at an overall level, there was no statistically significant difference in average problem solving scores between BC1 and BC2. Nevertheless, data in Table 4.11 and Figure 4-B indicate that income is less strongly related to problem solving ability in BC2 compared with BC1. Looking at Figure 4-B, the slope of the BC2 line is flatter than it is for BC1 reflecting the smaller difference between average scores in the lowest and highest income groups. Whilst this change has partly occurred because children in the lowest income group in BC2 showed an increase in their average problem solving scores, it also results from a decrease in average scores amongst children in the two highest income groups.

Children whose parents had higher educational qualifications tended to have higher average ability scores than those whose parents had lower qualifications (Table 4.12).This pattern is evident in both cohorts and for both assessments. Differences by parental level of education are statistically significant.

There was an increased vocabulary score between BC1 and BC2 among children from all parental educational backgrounds except those in the lower Standard Grade group. The level of increase is broadly similar in most sub-groups so that the difference between children with degree educated parents and those whose parents have no qualifications is similar for BC1 and BC2 with only a slight reduction (8.3 for BC1 compared with 8.1 for BC2).

Table 4.12 Mean standardised cognitive ability scores, by parental level of education and cohort

	No qualifications	Lower Standard Grades or VQs or Other	Upper level SGs or Intermediate VQs	Higher grades and upper level VQs	Degree level academic and vocational qualifications
	%	%	%	%	%
Naming vocabulary
BC1	42.1	44.8	45.0	47.7	50.4
BC2	44.6	43.3	46.6	50.1	52.7
Unweighted bases - BC1	174	179	745	1309	1516
Unweighted bases - BC2	124	168	652	1322	2204
Problem solving
BC1	40.7	44.2	43.8	47.0	48.3
BC2	42.3	43.4	44.0	46.4	47.3
Unweighted bases - BC1	177	182	749	1308	1523
Unweighted bases - BC2	131	175	663	1333	2217

Naming vocabulary: differences by education - p < .001; differences by cohort - p < .001; cohort*education p < .05
Problem solving: differences by education - p < .001; differences by cohort - p = NS; cohort*education p = NS.

On both assessments and for both cohorts, children with mothers aged 30 or older had higher average ability scores than children with younger mothers (Table 4.13). Differences by maternal age are statistically significant. Mean vocabulary scores increased for children in all maternal age groups between BC1 and BC2. This change was smallest for children whose mothers were aged 40 or older. As a result, the difference in ability score between children whose mothers were in the youngest and oldest age groups decreased a little between cohorts from 5.7 in BC1 to 4.9 in BC2. However, comparing the differences between children whose mothers are in the youngest age group and those whose mothers are aged between 30 and 39 shows almost no change between cohorts (5.1 in BC1 compared with 4.9 in BC2) and as a result maternal age remains similarly related to vocabulary in BC2 as it did in BC1.

Table 4.13 Mean standardised cognitive ability scores, by maternal age at child's birth and cohort

	Under 20 years old	20 to 29 years old	30 to 39 years old	40 or older
	%	%	%	%
Naming vocabulary
BC1	44.1	46.1	49.2	49.8
BC2	46.4	49.1	51.3	50.4
Unweighted bases - BC1	201	1456	2106	146
Unweighted bases - BC2	195	1777	2424	209
Problem solving
BC1	42.8	45.4	47.5	46.4
BC2	44.5	45.7	46.5	46.7
Unweighted bases - BC1	202	1461	2115	147
Unweighted bases - BC2	197	1810	2440	211

Naming vocabulary: differences by age - p < .001; differences by cohort - p < .001; cohort*age p = NS.
Problem solving: differences by age - p < .001; differences by cohort - p = NS; cohort*age p < .001

Table 4.14 shows that children living in less deprived areas had higher than average ability scores than those living in more deprived areas. This trend applies on both assessments and in both cohorts. Differences by area deprivation are statistically significant. In relation to naming vocabulary, mean scores for children living in all deprivation quintiles were higher in BC2 than in BC1. This change was statistically significant. However, the increase was slightly lower for children living in areas in the most deprived quintile compared with those living in all other areas. As a result, the difference in average scores for children living in the most and least deprived areas was slightly higher for BC2 than it was for BC1 (5.1 for BC1 compared with 6.2 for BC2). Nevertheless, area deprivation remains similarly related to vocabulary in BC2 as it did in BC1.

Table 4.14 Mean standardised cognitive ability scores, by area deprivation (quintiles) and cohort

	1 Most deprived	2	3	4	5 Least deprived
	%	%	%	%	%
Naming vocabulary
BC1	45.1	45.3	48.5	49.0	50.2
BC2	46.4	48.3	51.1	52.2	52.6
Unweighted bases - BC1	749	643	831	845	862
Unweighted bases - BC2	829	852	950	997	964
Problem solving
BC1	43.1	45.1	47.0	47.6	48.9
BC2	43.8	45.1	46.8	47.4	47.2
Unweighted bases - BC1	751	648	830	850	867
Unweighted bases - BC2	841	861	967	1007	969

Naming vocabulary: differences by area deprivation - p < .001; differences by cohort - p < .001; cohort*area deprivation p = NS.
Problem solving: differences by area deprivation - p < .001; differences by cohort - p = NS; cohort*are deprivation p = NS.