Parents’ basic skills and children’s test scores Augustin De Coulon, Elena Meschi and Anna Vignoles
Background The UK has a poor record in terms of the basic skills of its adult population Moser report: 20% (40%) of adults in England had severe literacy (numeracy) difficulties (DfEE, 1999). UK in bottom half of the OECD basic skills distribution (Leitch 2006). Having poor basic skills has clear consequences for the adults themselves, in terms of their economic circumstances (earnings, employment) Are there implications for the cognitive and non cognitive skill of the next generation? Should we target low skill adults?
Aims Investigate how the basic skills of adults (numeracy and literacy) relate to the early cognitive and non-cognitive skill of their children Test for an inter-generational relationship Focus on skills parents have when they raise their children not correlation of early ability 2004 British Cohort Study survey which provides basic skill assessments for adults and cognitive/non cognitive tests for children
Literature Literature on the inter-generational transmission of education and skill (e.g. Behrman and Rosenzweig, 2002; Black, Devereux and Salvanes 2005; Black et al. 2009; Carneiro et al, 2007) Literature on skill formation and early cognitive and non cognitive development (e.g. Blanden, Gregg and Macmillan, 2007; Brown, McIntosh and Taylor, 2009; Cunha and Heckman, 2008, Ermisch et al. 2002; Michael, 2004, 2005; Todd and Wolpin, 2007) Causal positive effect from parental education on child’s education Positive inter-generational correlation in early ability – difficult to separate genetic and environment effects Early years sensitive to parental investments in terms of cognitive and non cognitive skill development
Data Face to face survey of the British Cohort Study (1970 cohort) carried out in 2004 All respondents (CM) assessed in terms of their literacy and numeracy in adulthood. 9,665 cohort members in the core dataset 4,792 randomly selected into “Parent and Child” elements of the survey. 2,824 (59 per cent) has at least one child
Model • Dependent variables • Index of child cognitive skill age 3-6 • Measure of child non cognitive skill age 3-6 • Subscript c = child; subscript p = parents or family; • Sp cognitive/non cognitive skills of the child’s parents; • X set of exogenous child characteristics; • F family structure
Model • Measure extent to which parents’ skills in adulthood provide an additional explanation of child’s early cognitive and non cognitive skill • Adjust SE for clustering within households • Estimate by gender of both parent and child
Covariates • Sex • Age of child • Number of siblings • First born • Family structure • Parental education • Household income • Family receives state benefits • Occupation and employment of parent • Parenting, attitudes etc.
Identification strategy • Rely on early measure of parental IQ and ability • If we find relationship between parents basic skills and children’s cognitive outcomes is this causal? • Relationship may be due to unobserved factors correlated with both parents’ and children skills
Parental skill measures age 5 • British Ability Scales Edition 2 • Cognitive functioning using ability scales • Separates non verbal and spatial ability • Designed to cover wide age range and abilities • Found to be reliable, particularly as measures of general ability and verbal IQ (Cook, 1988)
Parental skill measures in adulthood • Series of multiple-choice questions, followed by a set of open-ended questions (Parsons and Bynner, 2005). • Multiple choice adapted from the Skills for Life National Adult Basic Skills Baseline Survey. Numeracy test devised by CDELL at Uni of Nottingham. • Distributions similar to those found in national data on assessments of UK adults, obtained from the Skills for Life Survey (see Williams et al., 2003)
Offspring’s measures of skill • Adapted British Ability Scales Second Edition (BAS II). • Naming Vocabulary • Early Number Concepts • Non-cognitive skill measured using the “Strength and Difficulties Questionnaire” (SDQ)
Measures used in the model • Synthetic indices of parental and child skill • Constructed using Principal Component Analysis (PCA) • Interpret these indices as measures of rank of each parent or child in terms of test scores • (Cawley et al. 1996)
Robustness • Separately for different tests (vocabulary, early number etc) • By gender of parent • By gender of child • By parental education • IV estimates
Conclusions • We measure parental skill in both childhood (IQ type measures) and in adulthood (cognitive skill measures) • Early and adult measures of parental skill remain highly significant when included in the model • Conditional on parents’ early ability, parent skill level in adulthood is a strong predictor of the child’s own skill level
Conclusion • We show, consistent with existing literature, a strong inter-generational correlation between parent ability in childhood and the early ability of their children • But knowing a parent’s cognitive skill in adulthood does provide additional information • Adult skill levels are easier to observe and can therefore be potentially used to target “at risk” children
Conclusion • Parental cognitive skill in adulthood, i.e. their literacy and numeracy skill, is significantly correlated with their children’s propensity to have emotional and behavioural difficulties. • Holds even when we control for the parent’s early cognitive and non-cognitive skill. • Can help us target children at risk of these emotional and behavioural problems.