early cognitive development and parental education

Infant and Child DevelopmentInf. Child Dev. 8: 49–62 (1999)

Early Cognitive Development andParental Education

Elizabeth Robertsa, Marc H. Bornsteinb, Alan M. Slatera,*and Jacqueline Barretta

a University of Exeter, Exeter, UKb National Institute of Child Health, USA

Relations between cognitive development in infancy and earlychildhood, and parental education were examined. Previous re-search has found little association between measures of theparenting environment, including parental education and socio-economic status (SES), and cognitive development in infants andchildren under 2 years of age. However, the earlier studies maynot have reliably measured individual differences in cognitiveabilities, thus, there is uncertainty as to what age elements in theparental environment affect cognitive development. Seventy-sixinfants were tested on a range of cognitive tasks at 3-monthintervals between the ages of 9 and 18 months. Information onparental education (a component of SES) was collected. Seventy-one of the children returned at 27 months and completed theBayley Scales of Infant Development, Mental Scale, which wasused as an outcome measure for the earlier tasks. The findingspresent a clear indication that cognitive development in earlychildhood is affected by the parenting environment, at leastfrom as early as 12 months. Copyright © 1999 John Wiley &Sons, Ltd.

Key words: cognitive development; early childhood; infancy;parental education; prediction

Several years ago it was established that childhood intelligence is moderatelypredictable from infancy, as shown by various visual information processingmeasures, especially habituation and novelty responsiveness (Bornstein andSigman, 1986; McCall and Carriger, 1993; Slater, 1995; Bornstein et al., 1997).However, it has also been argued that these measures are no more predictivethan variables that make up the infant’s background, in particular parentalsocio-economic status (McCall, 1977). Socio-economic status (SES) is normallymeasured using information about family occupation, education and income(Hernandez, 1997). The question of whether and when environmental condi-tions influence child competencies is a central one for studies of infant status,individual differences and their development. It is known that parental SESrelates to parents’ beliefs and behaviours (Ninio, 1988). The present study

* Correspondence to: Psychology Department, Washington Singer Laboratories, University of Exeter,Exeter EX4 4QG, UK. Tel.: +44 1392 264595; fax: +44 1392 264623; e-mail: [email protected]

CCC 1522–7227/99/010049-14$17.50Copyright © 1999 John Wiley & Sons, Ltd.

Received 22 May 1998Accepted 22 October 1998

E. Roberts et al.50

investigated relations between infant performance at different ages on differenttasks and childhood cognitive status, as well as the role of parental education inthe prediction of childhood intelligence.

Several studies have set out to assess the contribution of SES to childhoodintelligence. Rubin and Balow (1979) conducted a large-scale study with over1000 participants to determine relations among infant development at 8 months,as measured on the Bayley Scales MDI, SES measured on the 10 point Socio-eco-nomic Index, and later performance on several intelligence and achievementmeasures at school in children aged 4–9 years. The 8-month Bayley Scale scoreswere lowly, but significantly, correlated with SES, r=0.05, although in view ofthe large number of participants, this result might not be very meaningful. SESwas a stronger predictor of the outcome variables taken at 4–9 years, rs=0.29–0.43, compared with the correlations with the Bayley Scales, rs=0.11–0.46.Kopp and Vaughn (1982) found SES correlated significantly, r=0.32 (n=71),with Bayley MDI scores obtained by 2-year-olds. Similarly, Cohen and Parmelee(1983) found significant correlations of 0.25 between years of maternal educationand cognitive measures from the Stanford–Binet at 2 years, increasing to 0.46 atthe age of 5 years. Rose and Wallace (1985) found a significant correlationbetween parental education and visual recognition memory in preterms aged 12months, r=0.30. Parental education also emerged as a significant predictor ofBayley Scales MDI at 24 months, r=0.44, in that study and remained so for laterages.

McCall (1972, cited in McCall et al., 1973) reported an association betweenhigher parental education and more rapid habituation, and Rose et al. (1978)found that infants aged approximately 12 months from families with higherparental education had significantly higher novelty scores on cross-modaltransfer tasks. In a longitudinal study, in which participants were tested on arange of variables from age 4 months, Kagan et al. (1978), p. 1016, found that ‘noinfant variable predicted IQ as well as the social class of the child’s family’.Social class was measured using parental occupation and education and wasfound to be the major correlate at age 10 years of the WISC-R, r=0.55 for boys,0.50 for girls (total n=68).

Mayes and Bornstein (1995) observed that none of these studies examined thecontribution of maternal education to standard measures of habituation perfor-mance and novelty recovery in infancy obtained using an infant-control proce-dure, which is the habituation technique most commonly used currently. Theystudied 94 term infants aged 3 months, using such a technique, and found noassociation between maternal education and habituation measures. They pointout, however, that other measures of infant functioning, or conversely, othermeasures of parental influence, may be associated with infant measures at thisage.

It appears that socio-economic status, as measured by such variables aseducation, occupation and income, relates to childhood IQ scores from the ageof approximately 2 years, as demonstrated by the studies cited above. Studies inwhich parental measures are correlated with standardised infancy measures,such as the Bayley Scales or the Gesell Developmental Schedules, do not showany association. The Bayley Mental Scale for children under 2 years is mainlycomposed of items that measure perceptual motor skills, rather than mentaldevelopment as such, and so the cognitive developments that are taking placemay not actually be measured. Studies that have attempted to measure theeffects of various environmental factors in infancy and early childhood mayequally conclude with non-significant results because they use measures that donot tap mental development.

Copyright © 1999 John Wiley & Sons, Ltd. Inf. Child Dev. 8: 49–62 (1999)

Early Cognitive Development and Parental Education 51

It is possible that environmental variables that are associated with SES do notexert an effect on early cognitive development, regardless of the infant measuresused, until after infancy. If this is so, an effect will not be seen whatever themeasures used. Alternatively, cognitive measures may become more reliableand valid with age. It is possible, therefore, that the failure of those studies thathave tested infants younger than 2 years and found little association with SEShave not actually measured cognitive abilities in these infants, or have notreliably measured individual differences in cognitive abilities. It is the currenthypothesis that the measures usually used to test mental development ininfancy do not assess the same cognitive abilities as the IQ measures used at 2years and later. If this is the case, then there is no reason to suppose that theywould show an association with cognitive outcome measures. The cognitive andlinguistic measures described below have been developed in order to addressthis issue, and it is proposed that these measures will show an association withlater IQ measures, and will also therefore significantly correlate with measuresthat are themselves correlated with SES.

Many of the studies described above have used measures of SES. However,the components of SES can often be misleading as a measure of variablesaffecting the home environment. Occupation can be a misleading variable as jobtitle alone may be meaningless without more detailed information. Mother’soccupation can also be an irrelevant variable when the mother has given upwork (sometimes temporarily) after the birth of the child. For the same reasonmother’s income may also be unreliable. Maternal education is often chosen torepresent SES in developmental studies, as it is highly correlated with SES as awhole (r=0.69 in Bradley et al., 1989, r=0.60 in Rose et al., 1991).

In the present study, parental education (i.e. the mean of mother’s andfather’s education) was chosen instead of SES for several reasons. Level ofeducation is a reliable variable, as it is accurately reported. It was decided toinclude father’s, as well as mother’s, education in this study, as some of thefathers of children in the study were equally or more responsible for day-to-daychildcare than the mothers in situations where the mother was the only wageearner or where neither parent was employed. In addition, Hauser (1994), p.1544, claims that ‘there is no reliable evidence that the effect of mother’seducational attainment dominates that of father’s educational attainment’. Whileparental education has been found to correlate highly with SES, the authorswould not wish to claim that the two measures are interchangeable. A reviewerof this paper has pointed out that the degree of overlap between the two will bevariable and will, in part, depend upon the population being examined.

The specific aims of the current study therefore were, first, to establishwhether cognitive performance at 27 months, as measured by the Bayley MDI,can be predicted from a range of cognitive and linguistic tasks administeredbetween 9 and 18 months, and, second, to examine the associations of bothmaternal and paternal education with performance on those tasks. If perfor-mance on the Bayley Scales at 27 months can be predicted by tasks given at ages9, 12, 15 and 18 months, it is possible that those early tasks are measuringcognitive ability (an alternative possibility, of course, which affects manycorrelational analyses, is that significant relationships may result from theintervention of a ‘third variable’). If parental education also correlates withperformance on the early tasks, then the variables associated with parentaleducation may be having an effect on cognitive ability at these ages.


E. Roberts et al.52

Table 1. Testing schedule

Girls Boys nGroup Age of testing (months)

13 13A 269, 12, 15 and 189 and 15 13 11 24B

14 12C 2612 and 18

METHOD

Sample

A sample of 76 fullterm children (40 girls and 36 boys) participated in the study.The main sources of recruitment were parent–toddler groups, baby clinics andnewspaper articles describing the study. All the children were born within 3weeks of expected delivery date, and were healthy and with no known medicalproblems. Their mean weight at birth was 3360 g. Fifty-six of the children werefirst-borns. The mean age of the mothers was 30.4 years (range 20–40), and ofthe fathers was 32.2 (range 21–46). The children were tested in three groups tocontrol for possible effects of repeated testing. The testing schedule is given inTable 1. Each testing session took place within 97 days of the child’s appropri-ate ‘birthday’. Seventy-one of the children (36 girls and 35 boys) returned fortesting on the Bayley Scales (second edition) Mental Scale (Bayley, 1993) as closeto their 27-month ‘birthday’ as possible.

Parental measures

When each of the children was 15 months old, parents were asked to completean Infant Background Questionnaire. The questionnaire was in two main parts.Part one requested details relating to the baby including birth weight, method ofdelivery, and behaviour as a young baby. Part two asked for informationconcerning the parents. Each parent was asked to indicate their educationallevel: this ranged from leaving school at 16 years (the minimum legal age toleave school) to postgraduate qualifications. For each parent the data used werethe number of years’ education completed, based on a starting age of 5 years.These data are given in Table 2.

Infant measures

Two experimenters were present at all of the testing sessions, which werevideotaped. The second experimenter scored the tasks during the sessions. Both

Table 2. Parental education

MothersNumber of years’ education Fathers

%Number%Number

54.7 4011–12 54.04113–14 16 21.3 6 8.1

815–16 21.61610.616.212Over 16 13.310

Totals 75 100.0 74 100.0



experimenters independently re-scored the tasks immediately after the sessionsby reference to the videotape. The experimenters’ scores were then comparedwith each other and with the notes made during the testing session. If agree-ment was not reached at this point, the video was re-examined and discussed.Where agreement could not be reached, the data were excluded from theanalyses.

9 months(1) A two-step means-end task was presented, in which an attractive toy was

placed on a support cloth out of reach of the infant, and then covered withanother cloth. The infant had to pull the support cloth, then remove the coverin order to reach the toy. There were three trials. A score was calculated basedon the baby’s intentional behaviour at each stage of the task, as described byWillatts (1992, personal communication), and averaged over the three trials.Performance on this problem-solving task improves with age in the period 6–10months and has been found to correlate with 3 year IQ (Willatts, 1997).

(2) Performance on search tasks is known to change and improve with age, asinfants’ understanding of objects develops (e.g. Baillargeon, 1993). An ‘A not B’search task was presented, in which an attractive toy was hidden in one of twoboxes attached to a board (location A). The baby was allowed to search for thetoy; after two successful searches at this location the hiding place was changedto the second location (B). A maximum of five trials at each location wasallowed. For each of the two hiding locations, A and B, scoring was based onhow many attempts were needed for two successive successful searches. If thechild searched correctly on the first and second trials, the maximum of 8 pointswas awarded. For each extra trial that was required for two successive correctsearches to be achieved, 2 points were deducted, giving the scoring systemoutlined below. If the child scored correctly on single trials, rather than on twotrials in a row, one point was scored (in total, not for each trial).

ScoreCorrect trials

1 and 2 862 and 3

3 and 4 424 and 51Single trial(s)

The same scoring system was applied to both the A and B trials, giving apossible maximum score of 16.

12 monthsAn understanding of causality and of causal relations is known to undergo

considerable change during infancy and early childhood (Bauer and Mandler,1989; Cohen et al., 1998) and at 12 months and subsequent ages measures ofcausal understanding were presented in the form of enabling relations tasks.

(1) Enabling relations tasks: four 2-step events, characterised by causal con-nections, were modelled individually for the baby. The materials were thenpresented for imitation. The events included: (i) make a rocking horse byattaching a horse to a magnetised base, then push it to make it rock; (ii) makea rattle by pushing a peg into a container and shake it; (iii) put a baby elephant


E. Roberts et al.54

into a hopper on the back of a large elephant and push down to make the babyappear down the trunk; (iv) put a car body onto a battery-powered wheeledbase, then add the driver to make the car go. Measures of performance onindividual tasks were taken by scoring the number of actions produced in thecorrect order, as outlined by Bauer and Mandler (1992). For each task, the childscored 3 points if both actions were produced in the right order, 2 if both wereproduced but in the wrong order, and 1 if only one action was produced. Amean score was also taken for all four tasks.

(2) A search task was presented, in which three boxes attached to a boardwere used as hiding places. The toy to be hidden was enclosed in the experi-menter’s hand, the hand placed under one of the cloths covering the boxes, thenopened to reveal that the toy was still there, closed again and hidden under asecond cloth. Six trials were presented, with the combination of visited andhiding places changing on each trial. Each trial was scored as follows:

Correct first searchScore 4Correct second search, where first search was at the ‘visited’ box3

2 Correct second search, where first search was at the non-visited boxCorrect third search1No search at all0

15 months and 18 monthsDifferent enabling relations and search tasks were used at these ages from

those used at 12 months. Also at 15 and 18 months, the children were given acategorisation task, described below. Categorisation skills are fundamentalcognitive abilities that are known to improve across infancy and early childhood(Riccuiti, 1965; Mandler, 1997).

(1) Enabling relations tasks: four causal events were presented, as at 12months. However, at these ages they involved three steps, rather than two.These tasks included: (1) Make a rattle. Place a peg in one half of a barrel, put thetwo halves together, then shake the barrel; (2) Make a gong. Place a bar on twoupright supports, hang the gong on the bar, bang it with a hammer; (3) Make aduck ‘swim’. Place a plastic lily pad into a hole in a square apparatus, insert theduck into a slot in the apparatus, push and pull the handle to make the duck‘swim’; and (4) Get the ball through the tube. Put a ball in the top of a perspextube, push in a plunger to make the ball drop down the tube, pull out a peg tolet the ball fall out of the bottom of the tube. For each task, a score was givenas follows:

Three actions produced, two pairs of actions in the right order6Score5 Three actions produced, one pair of actions

Two actions produced, one pair of actions4Three actions produced, no pairs3

2 Two actions produced, no pairsOne action produced1

A ‘pair of actions’ constitutes two actions that follow on one from the other. Intask 1, for example, placing the peg in the barrel and putting the two halves of



the barrel together is regarded as one pair, and putting the two halves of thebarrel together and shaking the rattle is another pair.

(2) Categorisation task: two sets of four items (pandas and cars or bricks andbabies) were presented randomly to the child on a table. The child was simplyasked to play with the toys, while the experimenter sat with her hands cupped,each hand equi-distant from the child. This encouraged sorting of the toys intotwo distinct sets. Behaviour with the items was observed and recorded, andrated according to the level of sorting behaviour exhibited.

1Score Single category touching2 Single category grouping3 Serial touching of two kinds of objects

Two category grouping4

(3) Search task: this was similar to the search task presented at 12 months,except that there were only two hiding places. A minimum of four trials waspresented. In the first two trials, the procedure was as for the 12 month searchtask. This condition was referred to as ‘object present’. In the second two trials,the toy was hidden at the first location and an empty hand shown to the childbefore being placed under the second cloth. This condition was referred to as‘object absent’. If the child searched at the same location in either the ‘objectpresent’ or ‘object absent’ condition (i.e. incorrectly on the first trial andcorrectly on the second trial), the first trial of each condition was repeated inorder to establish whether the child had a position bias or whether learning hadoccurred. A scoring system was devised that covered all possible search pat-terns. The trials were treated as two sets of two conditions, i.e. ‘object absent’and ‘object present’. Two points were awarded for each successful search ontrials 1 and 2. Where trials were repeated, either 1 point was subtracted forposition bias or 1 added for learning, depending on the outcome of the trial. Therange of scores for each condition was therefore as follows:

ScoreTrial

2 1 (repeated)1

CExample A C — 2+2=4C CI 2+1=3Example BI — 2−1=1Example C CC II 2−1=1Example D

IExample E I — 0+0=0

C, correct; I, Incorrect

Measures of language development

Measures of children’s language ability were taken at 12, 15 and 18 monthsusing the parent report MacArthur Communicative Development Inventory(CDI), Infant and Toddler Scales (Fenson et al., 1991). These give a measure ofboth word comprehension and production at 12 and 15 months (Infant Scale),


E. Roberts et al.56

Table 3. Summary of design

9 12Age at testing 15 18 27(months)

A and C A and B A and CGroups tested A, B and CA and Bn=50 n=52 n=50 n=52 n=71

Enabling Bayley Scales:EnablingMeans-end EnablingTasksrelations relations relations

A not B Search Search Search Mental ScaleLanguage Categorisation Categorisation

Language Language

and production only at 18 months (Toddler Scale). The authors of the CDIconducted a study to examine the differences between results obtained fromparents with different levels of education. No significant differences werefound on the production and comprehension scores for either the Infant orthe Toddler Scale (Fenson et al., 1991). Bates (1993), commenting on the CDI,says that ‘At this point, we are convinced that parental report is the best wayto obtain a global estimate of language comprehension for children in thefirst and second years of life . . . (but) at the same time the CDI has seriouslimitations.’ (p. 226). She acknowledges that ‘many readers are still very skep-tical of parental report’ (p. 227). Comprehension at 15 and 18 months wastherefore also estimated in the laboratory testing sessions by presentation ofcolourful pictures of everyday items (e.g. sock, phone, dog, car), which thechildren were asked to identify. Production of language was measured at 18months by noting down, from the video recording of the test session, all thechild’s meaningful utterances.

Outcome measure at 27 months

The children were given the Bayley Scales of Infant Development (secondedition) Mental Scale (Bayley, 1993); the score is expressed as a Mental Devel-opment Index (MDI). Table 3 summarises the design of the study.

RESULTS

All analyses were initially conducted separately for groups A, B and C. Noeffects of repeated testing were found, which would be indicated by signifi-cantly higher scores for children in group A at the 12, 15 or 18 month testingsessions: all across-group comparisons, using t-tests, were non-significant. Allanalyses, therefore, refer to the whole sample. Table 4 gives the mean scoresand S.D.s obtained on each task.

The correlations described below and presented in Tables 5 and 6 need tobe treated with caution. Like others reporting across-age correlations (re-viewed by McCall and Carriger, 1993; Slater 1995; Bornstein et al., 1997), theauthors have not made statistical corrections for the number of tests run.However, only one of the significant correlations is in the opposite directionto that expected: this is between Parental Education and CDI Phrases at 15months (Table 5) and is commented on below.



Table 4. Mean scores and S.D.s for 9–18 month measures

Mean scoreTask(S.D.)

9 Months (n=50)Means-end (support and search) (n=47) 11.4 (5.0)

Two sets of data removed for fussing, and one for problems with the video camera9.5 (6.1)AB Search (n=49)

One set of data was removed as inter-rate agreement was not reached

12 Months (n=52)Enabling relations: Horse (n=52) 1.4 (1.2)

2.1 (1.1)Enabling relations: Rattle (n=52)1.5 (1.0)Enabling relations: Elephant (n=48)

Four sets of data were removed as inter-rater agreement was not reachedEnabling relations: Car (n=47) 1.1 (0.9)

This task was not available for the first five babies tested1.6 (0.7)Enabling relations: Mean (n=52)

Invisible displacements search: (n=52) 7.6 (2.6)4.5 (4.0)CDI production (n=71)

CDI comprehension (n=71) 57 (38)71 completed CDIs returned by parents

15 Months (n=50)One set of data was removed from the analysis of the 15 month results, as the baby

was illEnabling relations: Rattle (n=49) 3.6 (2.3)Enabling relations: Gong (n=49) 2.2 (1.6)Enabling relations: Duck (n=44) 3.7 (2.1)

Equipment broken for five test sessionsEnabling relations: Ball (n=48) 3.3 (2.0)

Equipment broken for one test session3.2 (1.2)Enabling relations: Mean (n=49)1.9 (1.5)Categorisation (n=49)

Search (n=47) 5.4 (2.1)Inter-rater agreement was not reached on two sets of data

4.7 (2.9)Language comprehension (lab. measure) (n=47)Two sets of data were removed for children fussing

110 (61.4)CDI comprehension (n=70)70 CDIs returned completed by parents

14 (13.1)CDI production (n=70)18.9 (5.6)CDI phrases (n=70)

18 Months (n=52)4.6 (2.2)Enabling relations: Rattle (n=51)

Enabling relations: Gong (n=52) 3.6 (2.1)Enabling relations: Duck (n=48) 5.0 (1.5)

Equipment broken for four test sessions4.0 (2.0)Enabling relations: Ball (n=51)

Equipment broken for one session4.3 (1.3)Enabling relations: Mean (n=52)2.5 (1.3)Categorisation (n=52)5.8 (2.4)Search (n=52)

Language comprehension (lab. measure) (n=51) 14.6 (10.3)Language production (lab. measure) (n=52) 9 (9.8)

67 (84)CDI production (n=71)72 completed CDIs returned: one discarded as an outlier


E. Roberts et al.58

Correlations between 9, 12, 15 and 18 month tasks, and outcome measure

None of the tasks presented at 9 months correlated significantly with the27-month Bayley MDI, but 12, 15 and 18 month tasks predicted scores on theBayley Scales at 27 months at increasingly high levels (see Table 5). At 12months, the search task and one of the enabling relations tasks (Horse) werepredictive. At 15 months, the enabling relations ‘Rattle’ task, together with theenabling relations mean score, were predictive. By 18 months, all the tasksexcept one enabling relations task and the search task correlated significantlywith the Bayley MDI at 27 months.

Correlations between 9, 12, 15 and 18 month tasks and parental education

There was less of an age-related effect with these correlations. Of the earliertasks, the 9 month means-end task, the 12 month enabling relations ‘Horse’ taskand the enabling relations mean score correlated significantly with parentaleducation. One of the 15 month enabling relations tasks (Rattle), together withthe mean, showed positive and significant correlations. However, the CDImeasure, comprehension of phrases, showed a negative correlation (−0.3,pB0.01): this could well be attributable to inaccuracies in parental report. At 18months, two enabling relations tasks and the mean were significantly correlatedwith parental education, but measures of language comprehension and produc-tion, together with two of the enabling relations tasks, the search task and thecategorisation task, were not significant.

Partial correlations: cognitive measures

The partial correlations presented in column 3 of Table 5 show the effect ofparental education on the correlations between infant measures and 27-monthMDI.

At 12 months, the Enabling Relations ‘Horse’ task correlated significantly withboth the MDI and parental education. However, when parental education waspartialled out, the association between the task and the MDI was positive, butno longer significant. On the other hand, the correlation between the Search taskand the MDI remained significant when parental education was partialled out.

At 15 months, the mean score of Enabling Relations tasks, together with oneof the component tasks, ‘Rattle’, showed significant correlations with the MDI.However, these items were also significantly correlated with parental education,and when education was partialled out, correlations were lower and non-significant.

Among the 18 month measures, several of the Enabling Relations taskspredicted MDI scores, as did the Categorisation measure. The mean EnablingRelations score, ‘Gong’ and ‘Duck’ also correlated significantly with parentaleducation; for all three of these variables, however, the association with MDIremained significant after partialling out parental education.

Partial correlations: language measures

None of the language measures at 12 months was significantly correlated witheither the MDI or parental education. At 15 months, CDI Phrases (a measure ofcomprehension) was negatively correlated with parental education. Whenparental education was partialled out, the correlation with MDI became positiveand significant. At 18 months, none of the language measures was correlated



Table 5. Correlations between 9–18 month tasks, 27-month MDI and parental education

Correlations withTask Correlations Correlations with27-month Bayleywith parental27-month Bayley

MDI MDI (with parentaleducationeducation partialledout)

9 Months (n=43–46)0.19 0.38**Means-end (support and 0.09

search)0.11AB search 0.02 0.06

12 Months (n=45–50)0.31* 0.170.28*Enabling relations: Horse0.13 0.02Enabling relations: Rattle 0.110.03 −0.09−0.04Enabling relations: Elephant

0.17Enabling relations: Car 0.15 0.110.24* 0.11Enabling relations: Mean 0.220.06 0.24*0.24*Invisible displacements search

0.09CDI production (n=66) −0.05 0.170.12CDI comprehension −0.17 0.21

15 Months (n=40–45)0.27* 0.41** 0.08Enabling relations: Rattle0.13 0.16 0.03Enabling relations: Gong

0.24 0.130.16Enabling relations: Duck0.26*Enabling relations: Ball 0.080.17

0.29* 0.47**Enabling relations: Mean of 0.13all four tasks

−0.09 0.23 −0.20Categorisation−0.04 −0.07Search −0.06

0.04 0.23 0.24Language comprehension(lab. measure)

0.10 −0.19 0.26*CDI comprehension (n=65)0.19 −0.08 0.23*CDI production

−0.30** 0.36**0.14CDI phrases

18 Months (n=45–49)0.21 0.21 0.28*Enabling relations: Rattle

0.30* 0.44***0.49***Enabling relations: Gong0.25*Enabling relations: Duck 0.27* 0.42**

0.21 0.11Enabling relations: Ball 0.24*Enabling relations: Mean of 0.36** 0.51***0.54***all four tasks

0.24* 0.17 0.22Categorisation0.02 0.160.12Search

0.55** 0.23 0.52***Language comprehension(lab. measure)

0.40** 0.15Language production (lab. 0.38**measure)

0.35** 0.13 0.34**CDI production (n=66)

* pB0.05; ** pB0.01; *** pB0.001.

with parental education, but all were with the MDI, and the correlationsremained significant when parental education was partialled out.

To summarise the partial correlations, where an association between an earliertask and the outcome measure existed, the prediction was maintained for 7 outof 12 of the tasks when parental education was partialled out. Of the five


E. Roberts et al.60

Table 6. Correlations between parental education and 27-monthMDI scores

Parental measure Correlation

0.46***Mother’s education (n=71)0.45***Father’s education (n=69)0.50***Joint education (n=69)

Pearson’s one-tailed, *** pB0.001.

correlations that attenuated to non-significance, four were with enabling rela-tions and one was with the categorisation task.

Table 6 shows that both father’s and mother’s education appear to have aboutthe same association with children’s performance on the Bayley Scales MDI at27 months.

DISCUSSION

Earlier research has demonstrated a clear relation between cognitive abilities inchildhood and parental education, but it is unclear at what point in earlydevelopment the parenting environment affects cognitive development. Recentfindings have suggested that early information processing (habituation andrecovery to novelty performance) at 3 months is uncorrelated with a measure ofthe parenting environment, maternal education (Mayes and Bornstein, 1995).The current study demonstrates that a similar measure of the parenting environ-ment correlates significantly with several different measures of infant cognitiveprocessing from 9 months. Thus, the present findings extend previous researchin suggesting that the parenting environment may have an increasing effect onchildren’s cognitive development across the first 2 years.

It is clear that many aspects of infant development are affected by environ-mental factors, and it is therefore important that studies of development identifymeasures of infants that are uncorrelated with measures of the parentingenvironment (such as education). The data presented in Table 5 demonstraterelations between infant cognitive development and 27-month MDI with theeffects of parental education removed. But the findings also present a clearindication that infant cognitive development is affected by the parentingenvironment.

ACKNOWLEDGEMENTS

The authors would like to thank the infants and their parents who participatedin this research. The research was supported by Research Grant R000232967from the Economic and Social Research Council to AS, and by a grant from theNational Institute of Health.

REFERENCES

Baillargeon, R. (1993). The object concept revisited: new directions in the investigation ofinfants’ physical knowledge. In C.E. Granrud (ed.), Visual Perception and Cognition inInfancy. Hove: Erlbaum.



Bates, E. (1993). Comprehension and production in early language development. In E.S.Savage-Rumbaugh, J. Murphy, R.A. Sevcik, K.E. Brakke, S.L. Williams and D.M.Rumbaugh (eds), Language comprehension in ape and child. Monographs of the Societyfor Research in Child Development, Serial No. 223, vol. 58, pp. 222–242.

Bayley, N. (1993). Bayley Scales of Infant Development, 2nd edn., San Antonio, TX: ThePsychological Corporation.

Bauer, P.J. and Mandler, J.M. (1989). One thing follows another: effects of temporalstructure on 1–2-year-olds’ recall of events. Developmental Psychology, 25, 197–206.

Bauer, P.J. and Mandler, J.M. (1992). Putting the horse before the cart: the use of temporalorder in recall of events by 1-year-old children. Developmental Psychology, 28, 441–452.

Bornstein, M.H. and Sigman, M.D. (1986). Continuity in mental development frominfancy. Child Development, 57, 251–274.

Bornstein, M.H., Slater, A.M., Brown, E.M., Roberts, E. and Barrett, J. (1997). Stability ofmental development from infancy to later childhood: three ‘Waves’ of research. In G.Bremner, A. Slater and G. Butterworth (eds), Infant Development: Recent Advances. Hove,UK: Psychology Press.

Bradley, R.H., Caldwell, B.M., Rock, S.L. and Ramey, C.L. et al. (1989). Home environ-ment and cognitive development in the first 3 years of life. A collaborative studyinvolving six sites and three ethnic groups in North America. Developmental Psychology,25, 217–235.

Cohen, L.B., Amsel, G., Redford, M.A. and Casasola, M. (1998). The development ofinfant causal perception. In A. Slater (ed.), Perceptual Development: Visual, Auditory, andSpeech Perception in Infancy. Hove: Psychology Press.

Cohen, S.E. and Parmelee, A.H. (1983). Prediction of 5-year Stanford–Binet scores inpreterm infants. Child Development, 54, 1242–1253.

Fenson, L., Dale, P.S., Reznick, J.S., Thal, D., Bates, E., Hartung, J.P., Pethick, S. andReilly, J.S. (1991). Technical Manual for the MacArthur Communicative Development Inven-tories. San Diego: San Diego State University.

Hauser, R.M. (1994). Measuring socio-economic status in studies of child development.Child Development, 65, 1541–1545.

Hernandez, D.J. (1997). Child development and the social demography of childhood.Child Development, 68, 149–169.

Kagan, J., Lapidus, D.R. and Moore, M. (1978). Infant antecedents of cognitive function-ing: a longitudinal study. Child Development, 49, 1005–1023.

Kopp, C.B. and Vaughn, B.E. (1982). Sustained attention during exploratory manipulationas a predictor of cognitive competence in preterm infants. Child Development, 53,174–182.

McCall, R.B. (1977). Childhood IQs as predictors of adult educational and occupationalstatus. Science, 197, 482–483.

McCall, R.B. and Carriger M.S. (1993). A meta-analysis of infant habituation andrecognition memory performance as predictors of later IQ. Child Development, 64,57–79.

McCall, R.B., Hogarty, P., Hamilton, J. and Vincent, J. (1973). Habituation rate and theinfant’s response to visual discrepancies. Child Development, 44, 280–287.

Mandler, J.M. (1997). Development of categorisation: perceptual and conceptual cate-gories. In G. Bremner, A. Slater and G. Butterworth (eds), Infant Development: RecentAdvances. Hove: Psychology Press.

Mayes, L.C. and Bornstein, M.H. (1995). Infant information-processing performance andmaternal education. Early Development and Parenting, 4, 91–96.

Ninio, A. (1988). The effects of cultural background, sex, and parenthood on beliefs aboutthe timetable for cognitive development in infancy. Merrill-Palmer Quarterly, 34, 369–388.

Riccuiti, H.N. (1965). Object grouping and selective ordering in infants 12–24 months.Merrill-Palmer Quarterly, 11, 129–148.

Rose, S.A., Feldman, J.F., Wallace, I.F. and McCarton, C. (1991). Information processing at1 year: relation to birth status and developmental outcome during the first 5 years.Developmental Psychology, 27, 723–737.

Rose, S.A. and Wallace, I.F. (1985). Cross-modal and intramodal transfer as predictors ofmental development in full-term and preterm infants. Developmental Psychology, 21,949–962.


E. Roberts et al.62

Rose, S.A., Gottfried, A.W. and Bridger, W.H. (1978). Cross-modal transfer in infants:relationship to prematurity and socio-economic background. Developmental Psychology,14, 643–652.

Rubin, R.A. and Balow, B. (1979). Measures of infant development and socio-economicstatus as predictors of later intelligence and school achievement. Developmental Psychol-ogy, 15, 225–227.

Slater, A.M. (1995). Individual differences in infancy and later IQ. Journal of ChildPsychology and Psychiatry, 36, 69–112.

Willatts, P. (1997). Beyond the ‘couch potato’ infant: How infants use their knowledge toregulate action, solve problems, and achieve goals. In G. Bremner, A. Slater and G.Butterworth (eds), Infant Development: Recent Advances. Hove: Psychology Press.


early cognitive development and parental education

Documents