Background Associations between the characteristics of the family environment, in particular poverty and family structure, and cognitive development are well established, yet little is known about the role of timing and accumulation of risk in early childhood. The aim of this paper is to assess the associations between income poverty, family instability and cognitive development in early childhood. In particular, it tests the relative role of family economic hardship compared with family instability in affecting cognitive functioning at the age of 5 years.
Methods The study draws on data from the UK Millennium Cohort, linking data collected in infancy, age 3, and age 5 years. Cognitive ability was directly assessed at age 5 years with the British Ability Scales. Using regression models we examine associations between persistent income poverty, family transitions, and children's cognitive ability, controlling for family demographics and housing conditions, as well as child characteristics.
Results The findings suggest that the experience of persistent economic hardship as well as very early poverty undermines cognitive functioning at 5 years of age. Family instability shows no significant association with cognitive functioning after controlling for family poverty, family demographics, housing and a set of control variables indicating child characteristics.
Conclusions Persistent poverty is a crucial risk factor undermining children's cognitive development—more so than family instability.
Statistics from Altmetric.com
- Child development
- cognitive problems
- early cognitive development
- family transitions
- longitudinal data analysis
- longitudinal follow-up study
Early cognitive development is a crucial indicator of developmental health, as it is associated with later educational and occupational attainment as well as health and wellbeing.1–7 What happens to children early in their lives is critical for their future development.8–10 A major risk factor undermining children's cognitive development is family poverty, in particular persistent poverty and adverse living conditions.8 ,11–14 In recent years family instability has become recognised as a further salient risk factor affecting children's development.15–21 Poverty and family instability are closely interlinked, as poverty affects families economically and socially, as well as on an emotional level. Economic hardship, for example, has been associated with a greater risk of relationship break-up.16 ,22 While the effects of both poverty and family structure on child development are well established, there is less knowledge about their relative impact on children's outcomes.23–25
In the following we will assess the relative role of family poverty and family instability on the cognitive functioning of young children. Poverty affects the amount and quality of material resources that are available to children, which in the following we will refer to as the poverty hypothesis. In addition, there is consistent evidence to suggest that children raised in stable two-parent families do better than those who experience multiple transitions in family structure, which has been referred to as the instability hypothesis.24 ,25 Because family break-up and the experience of poverty often co-occur,26 it is important to assess their combined as well as separate effects on children's outcomes. Evidence from previous research on the relationship between poverty, family structure and children's academic attainment has produced conflicting findings, with some arguing that poverty may explain much of the effect of family structure on children's educational achievement,27–30 whereas others have argued that family structure operates independently of family economic status in influencing children's outcomes.16 ,27 ,31 ,32 Differences in findings might be due to variations in the ages of the children studied, differences in assessments, or different operationalisations of family structure. In addition, most previous studies have focused on poverty and family structure as states and have not taken into account continuity and change in family circumstances.
An alternative explanation for the association between poverty, family instability and children's cognitive functioning is that all of these factors might be associated with each other due to their association with previous characteristics of the parent (such as mother's age and education).25 ,30 According to the selection hypothesis24 parents' own characteristics may affect their ability to maintain a stable income or a stable and committed partnership, and impact on the characteristics of their children, either through the environment in the home, through genetic transmission, or more likely the combination of both. We will thus control for the role of parental characteristics in our analysis. In addition, housing conditions have been identified as a potential risk factor shaping the cognitive attainment of young children,33–36 for example, due to overcrowding or lack of personal space. We thus assess the role of environmental influences on cognitive development by controlling for indicators of living conditions in our analysis.
Using a large nationally representative sample, the aim of this study is to disentangle the sometimes conflicting conclusions of previous studies by addressing the following questions: First, does persistent family poverty undermine children's cognitive functioning? Second, does family instability depress levels of cognitive functioning in children? Third, if both poverty and family instability affect cognitive functioning, which effect is larger? Fourth, can associations between poverty, family instability and cognitive functioning be explained through previous characteristics of the parent and/or current housing conditions. This study focuses on cognitive functioning at the age of 5 years, due to its proximity to school entry and the crucial role of early cognitive functioning on later achievement and health.37 All analyses control for characteristics of the child to take into account early individual difference factors, some likely to reflect biologically based influences, which have been shown in past studies to be associated with cognitive development.38–41 This study will be one of the first to assess the relative effects of persisting poverty and family status transitions on children's cognitive functioning at the age of 5 years in a general population sample.
The study draws on data collected for the Millennium Cohort Study (MCS), a survey of 18 819 babies born between September 2000 and January 2002 into 18 553 families living in the UK.42 The first sweep of the MCS was carried out during 2001 and 2002 when most babies were 9 months old. The sample design allowed for the disproportionate representation of families living in areas of child poverty. Due to disproportionate sampling, special weights have to be applied in analysing the data.43
Data were collected from parents via personal interview and self-completion questionnaires. In 2006, at the age of 5 years, 15 246 families took part in the survey and for 14 682 children we have complete data on the cognitive assessments. The following analyses are based on 8874 children and their mothers for whom we have complete data on all measures. In comparison with the original sample, the analytical sample contains relatively more socially privileged and better educated mothers and slightly more girls. Children in the analytical sample also had slightly higher cognitive test scores than children for whom we have no information on family income or family status at the three measurement points (mean 51.8 (SE 0.18) vs 48.8 (0.32) for the picture vocabulary subtest and mean 50.7 (SE 0.20) vs 49.4 (0.23) for the pattern construction subtest).
We used equivalised net household income (taking into account household size and composition) as our indicator of family poverty,26 ,44 identifying families with less than 60% of the national median income at each of the three measurement points. The dichotomised information was dummy-coded into a categorical variable with nine levels (table 1). The categorical dummy variable provides information about both the timing and the duration of income poverty.
The family transitions variable is derived from information about mothers' relationship status (married, cohabiting, single) at the three different measurement points. The 27 possible combinations were dummy-coded into a categorical variable with eight levels, reflecting the most common transition patterns (table 2). The categorical dummy variable provides information on stability and change in family structure during the first 5 years of the child's life.
At 5 years of age each child was directly assessed by specially trained interviewers using the British Ability Scales (second edition), a reliable measure of cognitive functioning with good external validity.45 ,46 Here we focus on two of the subscales: naming vocabulary and pattern construction, capturing core aspects of verbal and non-verbal skills. Age-related starting points, decision points and alternative stopping points were used to ensure that the motivation and self-esteem of the child were protected, that the testing focused on the most suitable items for the child, and that the assessment time was kept to a minimum.45 Test scores were T-standardised to a mean of 50 and a SD of 10.
Mother's age at birth of child (below ages 20, 20–29, 30–39, 40+ years).
Parental education: mother's or father's level of education, whichever was highest (none through postgraduate degree level).
Housing conditions at age 5 years of the child
Home ownership (yes/no).
Crowding (rooms/people in household).
Child gender (0=male, 1=female).
Child age at assessment (continuous, in months).
Child birth weight (continuous, in kg).
Child's ethnicity (white, mixed, Indian, Pakistani or Bangladeshi, Black, other).
Delay in gross and fine motor development at 9 months was assessed by parental reports using statements adapted from the Denver developmental screening test.47 Delay in the developmental milestones is defined by the infant not reaching a milestone that 90% of infants in that age group can pass, for example, only 88% of infants can move around the floor at 8 months but 92% can do this by 9 months.48 ,49 So an 8-month-old baby is not delayed if s/he cannot move around, but a 9 month or older infant who cannot move around the floor is identified as delayed on this milestone.
To test the associations between poverty, family transitions and cognitive ability we ran a series of regression models for naming vocabulary and pattern construction separately. Because cognitive ability was assessed on a continuous and normally distributed scale we used ordinary least squares regression.
Table 1 shows the prevalence of poverty experienced between the age of 9 months and 5 years as well as the associated means and 95% CI for children's cognitive ability scores. The majority of families (62.1%) were identified as not being poor at any of the three assessments, although approximately 13% of families experienced persisting poverty. There appears to be a poverty gradient in children's cognitive test scores, with those exposed to persistent poverty scoring approximately 5 to 7 points less in the naming vocabulary test than those who never experienced poverty. Verbal abilities appear to be more strongly affected by poverty than non-verbal skills.
Table 2 shows the prevalence of different family transitions and associated levels of cognitive ability. The majority of parents were stably married (56.6%), and approximately a tenth were either continuously cohabiting with the same partner (12.7%) or continuously single (7.8%). Just under a quarter of mothers who cohabited when their child was aged 9 months were married 4/5 years later (usually to the biological father). We also find that approximately 10% of the single mothers had entered marriage by 2006. We furthermore find significant minorities of mothers who either had exited a relationship, or experienced one or more other family transitions in the first 5 years of their child's life. Children growing up in stable two-parent families show higher levels of cognitive ability than those in stably cohabiting families or those who experienced a change in living arrangements. Children in stable single parent families score lowest in assessments of both verbal and non-verbal skills.
The regression results for naming vocabulary and pattern construction are shown in tables 3 and 4, respectively. Model 1 is the poverty baseline model including the poverty measure and controlling for child characteristics. Child characteristics on their own explain approximately 8% of the variance in naming vocabulary and approximately 7% of the variance in pattern construction, suggesting that child characteristics play a crucial role in shaping cognitive attainment by the age of 5 years. Poverty has a significant effect on children's cognitive functioning at the age of 5 years after controlling for child characteristics. Persisting poverty across the three time points has the greatest negative effect. Model 1 explains approximately 15% of the variation in naming vocabulary and 10% of pattern construction.
Model 2 is the family instability baseline model including the family transition variable and controlling for child characteristics. Being stably married is the baseline, compared with which each of the other family structures are significant risk factors for reduced levels of children's cognitive functioning at the age of 5 years, after controlling for child characteristics. Model 2 explains approximately 12% of the variation in naming vocabulary and 9% of pattern construction.
Model 3 includes both the poverty and family transition variables simultaneously, again controlling for child characteristics. Controls for family transitions had little impact on estimates of the effects of family poverty: all poverty variables remain significantly associated with cognitive functioning at the age of 5 years, except for transient experiences of poverty at the age of 3 years only (npn), which showed no significant risk effect on pattern construction. By contrast, controls for income poverty markedly reduced estimates of the effects of family transitions. Only a subset of the family transition experiences remain significantly associated with naming vocabulary, in particular stable cohabitation, moving from cohabitation into marriage, cohabitation to other transitions and stably single, all of which showed a negative association compared with being stably married. For pattern construction, only stable single parent family status continues to show a significant (negative) effect.
Model 4 adds information about maternal age and parental education, including both the poverty and family transition variables and controlling for child characteristics. After adding the demographic variables the experience of persisting poverty (ppp) remains a significant risk factor for verbal ability (naming vocabulary), as does cumulative (npp and ppn) and intermittent poverty (npn, pnp), as well as the early experience of poverty at the age of 9 months (pnn). For pattern construction persistent and cumulative poverty also show a significant negative effect. Family structure, by contrast, has no significant association with either verbal or non-verbal cognitive ability in this multivariate model. Model 4 explains approximately 19% of the variation in naming vocabulary and 11% in pattern construction. It seems that taking into account parental characteristics considerably reduces the poverty effect, and the effects of family instability on cognitive functioning at the age of 5 years appear to be attributable to previous parental characteristics.
Model 5 adds the indicators for current housing conditions to the model. In addition and above the influence of family poverty, family instability, parental characteristics and the child control variables there is a significant association with indicators of housing conditions at the age of 5 years, in particular overcrowding. Adding indicators of living circumstances reduces the association between poverty and cognitive functioning, although associations between persistent and cumulative poverty, as well as early poverty at the age of 9 months remain significant in addition and above the effects of the other variables included in the model. Associations between cognitive functioning and indicators for family transitions are non-significant in this multivariate model. There are, however, significant effects of previous parental characteristics, in particular parental education. Of the child characteristics, age, gender, ethnicity and gross motor delay were significantly associated with children's naming vocabulary, and age, gender, ethnicity, birth weight and gross motor delay remained significantly associated with pattern construction. Model 5 explains approximately 19% of the variation in naming vocabulary and 12% of pattern construction.
We have used a large, longitudinal dataset to establish the relative effects of poverty and family instability on children's cognitive ability in early childhood. The findings suggest a strong and significant negative effect of income poverty on cognitive functioning at the age of 5 years, whereby the experience of persistent and cumulative poverty, and notably also exposure to hardship during the first year of life, have a detrimental effect on cognitive functioning. The findings suggest a significant role of cumulative risk experiences depending on the duration of exposure to poverty, as well as sensitive periods during early life.10 ,12 The effect of poverty appears to be slightly stronger on verbal than on non-verbal skills, confirming previous findings.50 ,51
Family structure and family instability on the other hand, had no significant association with cognitive ability after controlling for child characteristics, family poverty and family demographics. Our findings are thus consistent with the poverty hypothesis, suggesting that poverty, and in particular the experience of persistent as well as early poverty, undermines children's cognitive functioning.16 ,27–29 In addition, we also found that some of the effects of poverty, and especially those of family instability, were attributable to previous parental attributes, such as mother's age and parental education, suggesting the potential role of selection effects.24 ,30 Another factor shaping the association between poverty and cognitive functioning is housing conditions, in particular crowding, which represent a significant risk factor undermining children's cognitive attainment.33–36 We furthermore found a significant role of child characteristics in shaping cognitive outcomes at the age of 5 years, suggesting a possible link to biologically based influences. Future research should disentangle in more detail the processes and mechanisms through which material and social disadvantage is transmitted, and pay special attention to questions regarding the role of poverty experienced during the first year of life, which might be especially detrimental for later functioning.
Study strengths and limitations
In interpreting the findings some strengths and limitations of the study have to be considered. First, the longitudinal nature of the present study has inevitably led to some attrition, raising concerns about selection bias. Only 78% of children from the baseline sample completed the cognitive assessments at 5 years of age. Of these, we only have complete data on income poverty and family transitions for 60%. The analytical sample was from relatively more privileged family backgrounds than the baseline sample, and there were significant differences in the levels of children's cognitive functioning at the age of 5 years. Our findings might thus underestimate the negative effect of poverty and disadvantage on cognitive functioning. Furthermore, we were only able to explain approximately a fifth of the variance in children's outcomes, and potential other influences on children's early cognitive skills (such as genetic as well as other influences reflecting more proximal aspects of the child's environment) were not assessed in our models, and future research has to delineate potential pathways and mediating processes in more detail.
The present study also has some advantages over existing work. First, the sample size resulted in high statistical power, and enabled us to identify heterogeneous family forms, differentiating between stable family arrangements and family transitions. Second, we could identify patterns of persistent poverty, but also take into account the timing of poverty experiences during the first 5 years in life. Third, we have direct assessment of cognitive capabilities measured at the age of 5 years. Fourth, the data are drawn from families who reside throughout the UK, which gives our findings a high degree of generalisability.
Our findings can help to close some gaps in the research literature, especially regarding the relative effects of family poverty and family instability on cognitive functioning during early childhood. We confirm the devastating negative effect of income poverty on children's early development, and show that family structure effects are spurious after controlling for child characteristics, poverty, parental education and mother's age. We hope that our findings contribute towards resolving previous uncertainties regarding these effects.
What is already known on this subject
While the negative effects of both poverty and family structure on child development are well established, there is less knowledge about their relative impact on children's cognitive functioning. Furthermore, previous evidence focused mostly on poverty and family structure as states and has not taken into account continuity and change in family circumstances.
What this study adds
This study is the first to assess the relative effects of persisting poverty and family status transitions on children's cognitive functioning at the age of 5 years using a large, longitudinal, general population sample. The study shows that early and persistent poverty undermines cognitive development, whereas family instability shows no significant association with cognitive functioning after controlling for family poverty and a set of control variables.
Those who carried out the original collection of the data bear no responsibility for its further analysis and interpretation.
Funding The analysis and writing of this article were supported by grants from the Nuffield Foundation and the Welcome Trust. BM is supported by the Medical Research Council. Data from the Cohort Studies were supplied by the ESRC Data Archive.
Competing interests None.
Ethics approval This study involves secondary analysis of publicly available data. Ethics approval for data collection was obtained from the South West and London Multi-Centre Research Ethics Committees in the UK.
Provenance and peer review Not commissioned; externally peer reviewed.
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.