Article Text

Download PDFPDF

What role for the home learning environment and parenting in reducing the socioeconomic gradient in child development? Findings from the Millennium Cohort Study
  1. Y Kelly1,
  2. A Sacker1,
  3. E Del Bono1,
  4. M Francesconi1,
  5. M Marmot2
  1. 1Institute for Social and Economic Research (ISER), University of Essex, Colchester, UK
  2. 2Epidemiology and Public Health, University College London, London, UK
  1. Correspondence to Professor Y Kelly, Institute for Social and Economic Research (ISER), University of Essex, Wivenhoe Park, Colchester CO4 3SQ, UK; ykelly{at}essex.ac.uk

Abstract

Background Early child health and development (ECD) is important for health in later life. Objectives were to (1) examine the extent of socioeconomic inequality in markers of ECD at ages 3 and 5 years; (2) examine whether the ECD–income gap widens between these ages; (3) assess the contribution of the home learning environment, family routines and psychosocial environment to observed inequalities in ECD.

Methods Data on socioemotional difficulties, and tests of cognitive ability in 3-year-old (n=15 382) and 5-year-old (n=15 042) children from the UK Millennium Cohort Study were used.

Results Children in the highest income group were less likely to have socioemotional difficulties compared with those in the lowest income group at 3 and 5 years (2.4% vs 16.4% and 2.0% vs 15.9%, respectively) and had higher mean scores: age 3 'school readiness' 114 versus 99; verbal ability 54 versus 48, and age 5: verbal ability 60 versus 51, non-verbal ability 58 versus 54 and spatial ability 54 versus 48 (all p<0.001). The income gap in verbal ability scores widened between ages 3 and 5 (Wald test, p=0.04). Statistical adjustment for markers of home learning, family routines and psychosocial environments did more to explain the income gap in socioemotional difficulties than in cognitive test scores.

Conclusion Our results suggest that relationships between family income and markers of ECD are amenable to change. The role of home learning, family routines and psychosocial environmental factors are potentially important in closing income gaps in ECD.

View Full Text

Statistics from Altmetric.com

Introduction

Early child health and development (ECD) is important for health in later life.1 The social gradient in markers of child development have been documented in the UK and elsewhere.2,,5 It follows that if we do something about the social gradient in child development this might impact on later social gradients in health.1

A vast array of environmental factors including parenting styles and activities and the parent–child relationship influence ECD,6,,8 and in turn, ECD at school entry predicts later educational attainment.4 9 Government funded initiatives aimed at improving the lives of young children emphasise the importance of what parents do, the home learning environment and the warmth of relationships in fostering good developmental outcomes in young children.10 11

What is already known on this topic

  • Early child health and development (ECD) is important for health in later life and social gradients in ECD are evident.

  • Numerous environmental factors including parenting styles and activities and the parent–child relationship influence ECD.

  • Government funded initiatives emphasise the importance parental activities, the home learning environment and warm relationships in fostering good developmental outcomes.

What this study adds

  • In the UK, there are strong socioeconomic inequalities in ECD with income gaps in socioemotional difficulties and cognitive ability throughout the preschool period.

  • Income gaps in cognitive test scores widen during the preschool years.

  • Markers of home learning, family routines and psychosocial influences explained income gaps in socioemotional difficulties better than cognitive test scores.

Prior studies have documented social gradients in markers of child development in school age children,4 and at single time points in early childhood.2 3 5 However, in contemporary UK settings we do not know the magnitude of social inequalities in ECD across the preschool years, nor do we know whether inequality gaps remain constant throughout the preschool period or widen over time. This paper adds to current knowledge by: (1) examining the extent of inequality in markers of ECD, according to income, at two time points – ages 3 and 5 years; (2) examining whether the ECD–income gap widens between ages 3 and 5 years; and (3) assessing the contribution of the home learning environment, family routines and psychosocial environment in explaining observed inequalities in ECD.

Methods

The Millennium Cohort Study

The Millennium Cohort Study (MCS) is a nationally representative longitudinal study of infants born in the UK. The sample was drawn from births in the UK between September 2000 and January 2002. The survey design, recruitment process and fieldwork have been described in detail elsewhere (http://www.cls.ioe.ac.uk/studies.asp?section=0001000200010010).12 The first three sweeps of the survey involved home visits by interviewers when cohort members were aged 9 months, 3 years and 5 years. During structured interviews at home visits, questions were asked about socioeconomic circumstances, demographic characteristics, home learning, family routines and psychosocial environment. At ages 3 and 5, cognitive assessments were carried out by trained interviewers and questions were asked about the cohort members' socioemotional difficulties.

Ethics approval for the MCS was obtained from the relevant ethics committees and parents gave informed consent before interviews took place, and separate written consent for cognitive assessments.

Socioemotional difficulties

When cohort members were approximately 3 and 5 years old, parents were asked to complete the Strengths and Difficulties Questionnaire (SDQ). At age 3 an age appropriate adapted version of the SDQ was used, and at age 5 the age 4–15 years version was employed (http://www.sdqinfo.org). Briefly, the SDQ is a validated tool which has been shown to compare favourably with other measures for identifying hyperactivity and attention problems.13 14 The SDQ asks questions about five domains of behaviour, namely: conduct problems, hyperactivity, emotional symptoms, peer problems and pro-social behaviour. Scores from the conduct problems, hyperactivity, emotional symptoms and peer problems subscales are summed to construct a total difficulties score. Clinically relevant cut-points for problem behaviours were determined to be the scores of the top 10% of all MCS children with SDQ data at ages 315 and 516 years, and were ≥17 and ≥15, respectively.

Cognitive ability assessments

Cognitive ability was assessed using widely validated, age appropriate tests. At age 3 the tests were: the Naming Vocabulary subscale from the British Ability Scale (BAS) and the Bracken School Readiness Assessment (BSRA). The BAS Naming Vocabulary assesses verbal ability/expressive language. During this test children are asked to name items pictured in a booklet.17 The BSRA measures basic concept development and the readiness of the child for formal education – the higher the score the more 'school ready' a child is considered to be. During the test children are shown a set of colour pictures that contain six subtests to assess basic concepts such as colours, letters, numbers/counting, sizes, comparisons and shapes.18 Mean age standardised values for the BSRA composite score are reported. At 5 years in addition to the BAS Verbal Ability subscale, two other BAS subscales were administered, namely: Picture Similarities which assesses non-verbal/problem solving ability, during which the child is asked to place a picture card against the most similar in concept among a set of four other pictures; and Pattern Construction which assesses spatial ability and consists of a set of timed tasks for the child, copying and constructing patterns with coloured tiles and cubes. These assessments use age related starting points and alternative stopping points to protect the motivation and self-esteem of the child.17 Mean age standardised t score values for BAS subscales are reported. The BSRA and BAS have been shown to be predictive of later child cognitive performance.18-20

Socioeconomic and demographic markers

Family income was categorised in to five broadly similar bands across survey sweeps. Demographic markers were whether the child was first born, whether the household language was English or another language, and the mother's age at the time of birth.

Home learning, routines and psychosocial environment

Variables were categorised into three theoretically informed overlapping domains of the home environment, which were learning, routines and psychosocial environmental factors. Markers of the home learning environment from infancy were: parental basic skills difficulties – this variable was a composite measure based on responses to questions to parents on ability to read a children's book, fill in forms and check change in a shop; at age 3 questions were asked about the frequency of learning activities: someone reads stories to the child, visits to the library, help with alphabet, numbers/counting, learning songs, poems and rhymes, and does drawing and painting; and at age 5 questions were asked about the frequency of: someone reads to the child, help with reading, writing and numbers, telling stories to the child, visits to the library, musical activities and draws, paints or makes things. Indicators of family routines at ages 3 and 5 were whether the child had regular bedtimes and mealtimes. Markers of the psychosocial environment at age 3 were: maternal psychological distress (K6 questionnaire21), parent–child relationship (Pianta scale22), discipline strategies – this was a composite score of seven items, α=0.64 (How often do you do the following when child is 'naughty': Ignore, Smack, Shout, Send to bedroom/naughty chair, Take away treats, Tell off, Bribe), nine parent–child items from the Home Observation for Measurement of the Environment Inventory, α=0.60 (mother's voice conveys positive feeling; mother converses with child at least twice; mother answers child's questions or requests verbally; mother spontaneously praises child's qualities or behaviour twice during the visit; mother caresses, kisses or cuddles child at least once during the visit; mother introduces interviewer to the child; mother scolds (shouts) or makes derogatory comments to child more than once during the visit; mother uses physical restraint, grabs or pinches child during the visit; mother slaps or spanks the child during the visit – positively phrased items were reverse scored),23 whether the mother felt she was a competent parent, whether the family had lots of rules and whether these rules were enforced; and at age 5 were: maternal psychological distress (K6 questionnaire), discipline strategies (the same items as age 3), whether the mother felt she was a competent parent and whether the mother felt close to the child.

Data analysis

Behavioural and cognitive outcomes are known to be moderated by multiple births.24 Therefore, we analysed data for singleton infants with data on family income. We examine two MCS samples based on those whose mothers (1) participated in sweep 2 (age 3) of the survey (n=15 382), and (2) participated in sweep 3 (age 5) of the survey (n=15 042).

For sample 1 (age 3), socioemotional behaviour data were available for n=14 218, school readiness data for n=13 651 and verbal ability data for n=14 373. Results are presented for cohort members with complete data for explanatory factors of interest; this reduced the sample for socioemotional behaviour to n=11 562 (75.2%), school readiness to n=10 930 (71.1%) and verbal ability to n=11 467 (74.5%).

For sample 2 (age 5), socioemotional behaviour data were available for n=14 395, verbal ability data for n=14 764, non-verbal ability data for n=14 756 and spatial ability data for n=14 707. Complete data for explanatory factors of interest reduced the sample for socioemotional behaviour to n=13 603 (90.4%), verbal ability to n=13 537 (90.0%), non-verbal ability to n=13 533 (90.0%) and spatial ability to n=13 488 (89.7%).

The distribution of explanatory variables in full sweep 2 and 3 samples compared with complete case samples was found to be similar (see online appendix 1). Therefore, analyses are based on the cases with complete data on relevant variables using Stata v 11.0. The SVY command was used throughout to take account of the clustered sample design and the unequal probability of being sampled. Hence, all confidence intervals and p values account for clustering and all proportions, means and regression coefficients are weighted using sweep relevant weights. These weights allow for non-response at all sweeps.

Multivariate regression models were used to investigate the importance of demographic characteristics, home learning, family routines and psychosocial environment for socioemotional difficulties (logistic regression) and cognitive ability scores (linear regression) in children according to family income. All models adjust for gender. Socioemotional difficulties models additionally adjust for age at time of home visit, but cognitive outcome models do not as individual scores are age standardised. Model A adjusts for demographic characteristics; model B additionally adjusts for markers of the home learning environment and family routines; and model C additionally adjusts for psychosocial environment.

The percentage reduction in the income gradient before and after full adjustment was calculated from the log odds for band 5 (poorest) versus band 1 (richest) in the socioemotional difficulties models and mean scores for band 5 versus band 1 in the cognitive test score models. Cross model hypotheses were assessed based on methods for comparing regression coefficients between models suggested by Clogg et al,25 and implemented in Stata by the suest command.

To assess the policy relevance of our models, we estimated the percentage change in the prevalence of socioemotional difficulties predicted by the fully adjusted model after randomly reallocating (1) 50% and (2) 100% of the children from the 'read to less than weekly' group to the 'read to daily' group.

Results

Markers of home learning, family routines and psychosocial environments were socially patterned, with the highest income families more likely to have favourable profiles compared with lower income families (see online appendix 2). Developmental outcomes were associated with home learning activities, markers of family routines and psychosocial environment (see online appendix 3), and these associations were independent of family income (data not shown).

There were strongly graded relationships between family income and developmental markers (tables 1 and 2). At ages 3 and 5 years, crude prevalences show that children from the lowest income families were approximately seven and eight times, respectively, more likely to have socioemotional difficulties compared with children from the highest income families. Patterns of association between family income and a continuous measure of socioemotional difficulties were similar to those using the dichotomised score (data not shown). Multivariate models showed that the likelihood of socioemotional difficulties in income bands 2–5 were reduced on adjustment for demographic, home learning, family routines and psychosocial environment. After statistical adjustment for demographic, home learning and family routines, the likelihood of socioemotional difficulties remained at ages 3 and 5 (model B ORs 3.75 and 4.44, respectively). At ages 3 and 5 years, there was an approximate 50% overall reduction (model C; age 3 OR 2.81, age 5 OR 3.03) in the income gradient in fully adjusted models (Wald test, p<0.0001) (tables 1 and 2).

Table 1

Odds of socioemotional difficulties data and regression coefficients for cognitive test scores at age 3 by family income

Table 2

Odds of socioemotional difficulties data and regression coefficients for cognitive test scores at age 5 by family income

Children from the highest income families had substantially higher cognitive test scores compared with their counterparts from the lowest income band. For verbal ability, which was the only cognitive test with data available at both age points, the income gap widened between 3 and 5 years of age (crude difference, Wald test, p=0.04; fully adjusted difference, Wald test, p=0.01). On adjustment for demographic, home learning and family routines, differences across income bands in cognitive test scores remained (model B verbal ability coefficients age 3 −3.67, and age 5 −5.49) Adjustment for demographic, home learning, family routines and psychosocial environmental factors (model C) reduced the size of the income gradient in cognitive test scores (Wald test, p<0.0001). Similar patterns were seen for school readiness at age 3 and non-verbal and spatial ability test scores at age 5 (tables 1 and 2). The reduction in income gradient for verbal ability was greater for age 3 test scores (49%) compared with those at age 5 (38%) (tables 1 and 2, model C).

For our policy relevant analysis when (1) 50% and (2) 100% of the sample were randomly reallocated from the 'read to less than weekly' group to the 'read to daily' group, holding all else constant, the estimated proportion of children with socioemotional difficulties dropped by 10% and 20%, respectively.

Discussion

There are strongly graded relationships between family income and markers of child development at ages 3 and 5 years. And for verbal ability, the only cognitive test for which we had data available at both age points, the income gap appeared to widen with increasing age. On statistical adjustment for demographic, home learning, family routines and psychosocial environmental factors, there was a 50% reduction in the income gradient for socioemotional difficulties, and between 27% and 49% reductions in cognitive test score gaps. For verbal ability we found that statistical models 'explained' more of the income gradient at 3 years compared with 5 years, perhaps reflecting shifts in the amount of time children spend in the home, that is, with the transition to school environments.

Our findings are supported by other studies that have shown the importance of parenting activities across income groups.2 3 5 26 We found that for socioemotional difficulties, statistical adjustment for psychosocial environmental markers had additional explanatory power over and above adjustment for markers of home learning and family routines. This is perhaps not surprising as indicators of home learning and family routines likely tap the transactional element between child and environment inherent with, for example reading a story together, and having routines around bed and meal times,8 27 and it has been reported that such activities along with favourable psychosocial environments are most beneficial in families with secure bonds between parent and child.26 In contrast, for cognitive test scores, statistical adjustment for psychosocial environmental factors had relatively conservative effects on estimated relationships, particularly when children were age 5. This might be because cognitive development is less sensitive to psychosocial aspects of the environment such as discipline strategies. Or it may be because we had data on fewer markers of the psychosocial environment at age 5. Alternatively, it might be because markers of home learning, family routines and psychosocial environment tap into the same portion of the family milieu that fosters cognitive development.

Our findings from a large nationally representative sample of 3- and 5-year-old children are consistent with those of other studies.3 5 A strength of this study was that we examined data on objective measures, collected by trained observers, of cognitive ability in children. On the other hand, data on socioemotional difficulties were only available from a parent report and it has been shown elsewhere that multi-informant measures are more reliable for clinical identification of problem behaviours.28 However, the SDQ is a validated tool, and importantly we determined age-appropriate norms in the current study by using the large MCS cohort data15 16 rather than norms from a different age range. The cut-points use the same >90th percentile cut-off criterion for clinical relevance as used in the original norms.14 In common with a previous US study,3 our statistical models left a substantial portion of the income gap unexplained. But socioeconomic and family environment variables used in models lack precision as they are surrogates for a myriad of ill-defined socioenvironmental factors and thus their importance is underestimated. Implicit in the work of some researchers4 is that there is an underlying genetic explanation for income inequalities in ECD. However, genetic factors that influence socioemotional and cognitive development have not been well characterised, nor have their frequencies across socioeconomic groups been established.29

The malleability of ECD30 31 has proven fruitful for policy and several intervention programmes aimed at improving developmental outcomes in the under-5s have demonstrated benefits in aspects of ECD3 10 11 32 and have been shown to be cost-beneficial in the long term.32 These interventions along with welfare reforms typically focus on the parent, or the child, or both parent and child, and it appears that a range of approaches are useful.3 33 There is room for developing policy aimed at closing the inequality gap in child development, and to do this programmes need to be more effective in improving developmental outcomes in disadvantaged children compared with their advantaged peers. For example, in the current context, a simple counterfactual argument suggests that if half or all of the 5-year-old children who were read to less than daily were instead read to on a daily basis there would be corresponding 10% and 20% reductions in the proportion of 5 year olds with socioemotional difficulties.

Our study used cross-sectional data, and future work should consider a longitudinal view of the impact of relationships between income inequalities, home learning and psychosocial environments on ECD. Longitudinal analyses will also help to reveal the direction of causality in the complex sets of processes involved in social inequalities in ECD.

Acknowledgments

The authors would like to thank the Millennium Cohort Study families for their time and cooperation, as well as the Millennium Cohort Study team at the Institute of Education. The Millennium Cohort Study is funded by ESRC grants to Professor Heather Joshi (study director).

Appendices

Appendix 1

Distribution of demographic characteristics, home learning, family routines and psychosocial environment for full and analysed samples

Appendix 2

Distribution of demographic characteristics, home learning, family routines and psychosocial environment by family income for cohort members with socioemotional difficulty and/or cognitive test data

Appendix 3

Socioemotional difficulties (%) and cognitive test scores (means) by demographic characteristics, home learning, family routine and psychosocial environment

References

View Abstract

Supplementary materials

Footnotes

  • Funding This work was supported by a grant from the Economic and Social Research Council RES-596-28-0001. The funders had no role in the interpretation of these data or in the writing of this paper.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.