Early learning and childcare at age five: comparing two cohorts

5 ELC Use and Outcomes at Age 5

5.1. Introduction

As already noted, improving children's outcomes and closing the gap between advantaged and disadvantaged children are important objectives of the Scottish Government's commitment to increasing the entitlement to ELC (e.g. Scottish Government, 2016a). To date, the evidence on the effects of increasing the number of funded hours of ELC on children's outcomes has been mixed. For example, previous GUS analysis using BC1 data found no association between weekly ELC duration and children's cognitive or social development at age 5 (Bradshaw et al., 2014). Additionally, the EPPE study found no evidence that children who attended full-time ELC provision had better outcomes at age 5 than children attending part-time (Sylva et al., 2004). That said, existing research has suggested that attending ELC (as opposed to not attending ELC) can have beneficial effects on children's outcomes (Melhuish et al., 2015).

It is widely acknowledged that if part of the aim of providing funded ELC is to improve children's outcomes, this provision must be of high quality (e.g. Scottish Government, 2016a). As noted above, existing research has suggested that attending a high quality ELC provider can have a positive effect on children's outcomes, even after other factors are taken into account (Melhuish et al., 2015). However, what constitutes 'high quality' ELC provision is itself a subject of research, as illustrated by the EPPE project which focuses specifically on identifying the key aspects of high quality pre-school provision (Sylva et al., 2012). Previous GUS analysis used data from Education Scotland and the Care Inspectorate alongside BC1 data to examine whether any quality measures appeared to be particularly important for child outcomes (Bradshaw et al., 2014). This analysis found a positive association between the Care Inspectorate care and support grading of the ELC setting and children's vocabulary at age 5.

Drawing on data collected from families in BC2, this chapter examines whether there was any relationship between the average number of hours a child attended ELC during the pre-school period ^[52] and their cognitive and social development upon entry to primary school, as well as their adjustment to primary school. Further to this, it examines whether there is any evidence that the quality of the ELC setting - as assessed by the Care Inspectorate - is associated with children's outcomes.

5.2. Key findings

• The analysis found no statistically significant associations between either the average number of hours per week a child spent in ELC or the quality of the ELC provider and children's adjustment to primary school, or their cognitive development at age 5 once other factors were taken into account. However, it did show some associations between ELC use and certain aspects of children's social and behavioural development.
• Children who spent 30 hours or more in ELC on a weekly basis were more likely to display above average levels of behaviours associated with hyperactivity at age 5 than children who spent 12.5 hours or less in ELC. This association was evident after controlling for differences in social background and the level of hyperactivity recorded when children were aged 3. Notably, though, the association was only apparent for children in lower and middle income groups - it was not statistically significant among children in the wealthiest 40% of households.
• The analysis found no other statistically significant associations between weekly ELC duration and child outcomes at age 5 after controlling for differences in social background.
• Among children who used ELC, attending a provider with higher staffing grades appeared to be associated with a very small decrease in the likelihood of exhibiting above average levels of peer problems at age 5.
• Children who attended an ELC setting that achieved at least 'very good' grades across all four of the Care Inspectorate's quality themes were less likely to have raised levels of peer problems at age 5, and were less likely to display below average levels of pro-social behaviour, than children who attended a setting that did not achieve these grades.

5.3. Analysing associations between ELC use and child outcomes

The main aim of this chapter is to examine whether the average weekly number of hours children attended ELC and the quality of the ELC setting were associated with outcomes at age 5. For each ELC measure, initial analysis was undertaken to look at associations between selected ELC measures and each of the following child outcomes at age 5: adjustment to primary school, vocabulary and problem solving ability, level of social, emotional and behavioural difficulties (measured through the total difficulties scale) and level of pro-social behaviour. Where a statistically significant association was found in relation to the total difficulties measure, further analysis was undertaken for each of the individual difficulties subscales - conduct problems, emotional difficulties, hyperactivity/inattention, and peer problems.

The main ELC measures of interest for this analysis are related to ELC provided by the child's main ELC provider (their pre-school provider). This includes both funded and unfunded hours. However, it does not include any childcare received from other providers (e.g. childminders used for wrap-around childcare, see also sections 3.3 and 3.4). The measure of weekly duration of ELC attendance used five categories similar to those used in a previous GUS report (Bradshaw et al., 2014) but adjusted to reflect the change in ELC attendance pattern between the two cohorts.

Where the bivariate analysis showed a statistically significant ^[53] association between the ELC measure and the outcome in question, multivariable regression models were fitted. ^[54]

In the first step, the equivalent outcome measured at age 3 was added to the model. Where the association between the ELC measure and the outcome was subsequently not statistically significant but the outcome at age 3 was, this indicated that the differences in outcomes observed at age 5 were by and large explained by earlier differences.

ELC use - including the type of ELC provider children attend - is associated with a number of factors which are also associated with children's outcomes. For example, analysis in previous sections showed that children in higher income households are much more likely to attend private or voluntary ELC providers and also tend to have different patterns of cognitive and behavioural outcomes than those in lower income households. Thus, we should account for these differences to properly explore the associations between ELC measures and outcomes at age 5. Where an association between the ELC measure and child outcome at age 5 still held once age 3 outcome was taken into account, a number of social background variables were therefore added to the model. This was done in order to control for any relationships between these and the ELC measure which might be explaining the association found in the bivariate analysis - for example, higher vocabulary ability at age 5 being explained by other factors not already captured in the age 3 outcome, rather than by characteristics of ELC.

The multivariable analysis controlled for the following social background characteristics: household income (equivalised), highest parental level of education (household level), socio-economic classification (household level), level of area deprivation, urban/rural location and the child's gender. Details about these variables are provided in Appendix A.

In cases where an association between the ELC measure and the outcome in question was still statistically significant even after controlling for differences in social background, further tests were carried out. First, where an association with average weekly ELC duration was found, a measure of ELC quality was added to the model to test whether the association still held once the quality of the ELC setting was taken into account. Conversely, where an association with ELC quality was still statistically significant once social background characteristics were controlled for, a measure of average weekly ELC duration was added to the model to test whether the association still held once differences in the number of hours the child attended their ELC provider was taken into account.

Finally, to test whether associations between the ELC measures and child outcomes differed for children from different backgrounds, 'interaction effects' were fitted to the models. This allowed us to test whether a relationship between, for example, ELC quality and children's social development varied according the level of household income. Where an interaction effect was found to be significant, two separate models were fitted: one for children in the wealthiest 40% of households and one for the remaining 60%. ^[55]

5.4. Weekly ELC duration and child outcomes at age 5

The bivariate analysis showed no statistically significant associations between the average weekly number of hours children attended their main ELC provider and their adjustment to primary school (as reported by their parent or carer).

Some associations were found between the average weekly number of hours children attended their main ELC provider and both vocabulary and problem solving ability. ^[56] On both measures, children who were attending their main ELC provider for between 12.5 and 16 hours per week appeared to do less well compared with their peers who attended their main ELC provider for either a lower or a higher number of hours per week. The association between duration of ELC attendance and vocabulary was no longer evident when children's vocabulary score at age 3 was controlled for. This suggests that the group of children who attended ELC for between 12.5 and 16 hours per week already had poorer vocabulary than their peers at the time they were aged 3 - i.e. before they started pre-school. ^[57] The association between duration of ELC and problem solving scores at age 5 was still evident after controlling for problem solving scores at age 3 but no longer held once differences in social background were taken into account. This indicates that the association found between ELC duration and problem solving ability was driven by differences in the children's social background rather than by how many hours they spent in ELC.

The bivariate analysis also showed an association between weekly duration of ELC attendance and children's level of social and behavioural difficulties. Bivariate associations were found in relation to the measures of total difficulties, hyperactivity and peer problems. However, once age 3 outcomes and differences in social background were taken into account, the associations with total difficulties and peer problems were no longer statistically significant. In contrast, the association between weekly ELC duration and the likelihood of children exhibiting above average levels of hyperactivity at age 5 remained even after controlling for levels of hyperactivity reported at age 3 and for differences in social background. Compared with those who attended for 12.5 hours or less, children who attended ELC for 30 hours per week or more were slightly more likely to be in the group with above average levels of hyperactivity at age 5. ^[58] This association remained statistically significant even when controlling for differences in the quality grades achieved by the ELC setting. ^[59] Of course, it is not possible to conclude that the long hours of ELC use are causing the higher risk of displaying above average levels of hyperactivity at age 5. For example, children who already displayed raised levels of difficulties at age 3 may have been more likely to subsequently spend longer hours in ELC because other forms of childcare were not suitable - hence, their difficulties might explain their long hours of ELC attendance, rather than the other way around. However, further analysis showed no differences in the association between duration of ELC and the likelihood of displaying above average levels of hyperactivity at age 5 according to the level of difficulties reported at age 3. ^[60] Thus, the association does not appear to be explained by children with higher levels of difficulties at age 3 being more likely to use longer hours of ELC - that is, irrespective of whether they had higher or lower levels of hyperactivity difficulties at age 3, children who experienced long durations of ELC were more likely to exhibit above average levels of hyperactivity at age 5.

As previously noted, this analysis looked specifically at time the child spent at their main ELC provider, and the findings therefore cannot be used to draw conclusions about use of formal childcare per se. Even so, it is worth bearing in mind that, as illustrated in section 3.3.2, among children in BC2, time spent at their main ELC provider made up a substantial proportion of the time spent in formal care - over two thirds (70%) of all the formal childcare received, and an even higher proportion (74%) of group-based formal care. Also, preliminary additional analysis (results not shown) suggested that there was an association between time spent in any group-based care and children's likelihood of exhibiting raised levels of hyperactivity at age 5 - although the research required to fully understand this association (between levels of hyperactivity and the combination of ELC and childcare being used, the time spent with different ELC and childcare providers, and overall time in group care) is beyond the scope of this report. Furthermore, the finding is in line with other research which has suggested that spending long hours in non-parental care may negatively affect children's outcomes. For example, previous GUS research found that children who attended non-parental childcare for more than 40 hours per week at the age of 3 had higher levels of social, emotional and behavioural difficulties than children who spent less (or no) time in non-parental care (Bradshaw and Wasoff, 2009).

Interestingly, the association between weekly ELC duration and the likelihood of exhibiting above average levels of hyperactivity at age 5 seemed to differ according to household income. ^[61] In fact, separate analysis of children in the wealthiest 40% of households showed no statistically significant association between the number of hours children spent in ELC and the likelihood of displaying raised levels of difficulties at age 5. In contrast, among children in lower and middle income households (outside the wealthiest 40%), children who spent 30 hours or more in ELC per week were more likely than those who spent 12.5 hours or less to display above average levels of hyperactivity at age 5. ^[62]

The analysis showed no differences in outcomes between children who spent between 12.5 and 30 hours in ELC and those who spent 12.5 hours or less.

5.5. ELC quality and child outcomes at age 5

No statistically significant associations were observed between the quality of the ELC setting - as measured by the Care Inspectorate - and children's subsequent adjustment to primary school.

Earlier analysis of BC1 data found an association between attending an ELC setting with a high care and support grade and having slightly better vocabulary ability at age 5 (Bradshaw et al., 2014). This association was statistically significant, meaning it was unlikely to have occurred due to chance alone, but the average difference in vocabulary observed was very small. The analysis carried out for this report - using BC2 data - found no statistically significant associations between any of the ELC quality measures considered and either problem solving or vocabulary ability. ^[63] To explain the difference in results would require further analysis which is beyond the scope of this report.

The bivariate analysis showed a number of associations between ELC quality measures and children's social and behavioural outcomes at age 5. Initial associations were found between three of the Care Inspectorate measures of ELC provider quality - care and support, staffing, and consistently high scores across all four themes (grading mix) - and the level of total difficulties reported at age 5. However, associations between total difficulties and care and support and staffing grade were very weak and once difficulties at age 3 and differences in social background were accounted for, these associations were no longer statistically significant.

The analysis also suggested that the quality of staffing was associated with the level of peer problems reported for children at age 5 - this association was statistically significant after controlling for the level of peer problems reported at age 3 and differences in social background. The multivariable analysis suggested that as the staffing grade increased, the likelihood of children displaying above average levels of peer problems decreased, although the effect size was small. In other words, attending an ELC setting with higher quality staffing grades appeared to be associated with a very small decrease in the likelihood of children exhibiting above average levels of peer problems at age 5. This association also held when accounting for differences in the number of hours children attended their ELC provider. This is perhaps not surprising given the limited variation in the amount of time children spent at their main ELC provider (i.e. most children attended for between 12.5 and 16 hours per week). Nonetheless, this seems to suggest that the association between staffing quality and children's level of peer problems was evident irrespective of the number of hours spent in ELC.

The bivariate analysis also showed associations between the overall, cross-theme level of quality of the ELC setting - that is, whether a provider achieved 'very high' or 'excellent' grades across all four quality themes - and the levels of peer problems and pro-social behaviour reported for children at age 5. Both these associations were statistically significant also after taking into account age 3 scores and differences in social background. This analysis suggested that those who attended an ELC provider which achieved 'very good' or 'excellent' grades across all four of the Care Inspectorate's quality themes were less likely to exhibit above average levels of peer problems at age 5. ^[64] Similarly, children who attended an ELC provider with high quality grades across the four themes were less likely to display below average levels of pro-social behaviour at age 5 when taking into account their level of pro-social behaviour at age 3, compared with children who attended an ELC provider that did not achieve such consistently high quality grades. ^[65] As above, this association remained statistically significant when accounting for differences in the number of hours children attended ELC. The analysis found no statistically significant differences in these relationships according to household income.