IJE Advance Access originally published online on September 23, 2008
International Journal of Epidemiology 2009 38(1):173-181; doi:10.1093/ije/dyn201
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Intelligence in girls and their subsequent smoking behaviour as mothers: the 1958 National Child Development Study and the 1970 British Cohort Study
1MRC Epidemiology Resource Centre, University of Southampton, Southampton General Hospital, Southampton, UK.
2MRC Centre for Cognitive Ageing and Cognitive Epidemiology, Department of Psychology, University of Edinburgh, Edinburgh, UK.
3Department of Quantitative Social Science, Institute of Education, University of London, London, UK.
4MRC Social & Public Health Sciences Unit, University of Glasgow, Glasgow, UK.
* Corresponding author. MRC Epidemiology Resource Centre, University of Southampton, Southampton General Hospital, Southampton, SO16 6YD, UK. E-mail: crg{at}mrc.soton.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Background Exposure to tobacco smoke either in utero or postnatally can have substantial adverse effects on child health, yet many women continue to smoke during pregnancy and after the birth. How women's intelligence in childhood affects their smoking behaviour as mothers is unclear.
Methods The participants were from two British national birth cohorts: 3325 women aged 33 years from the 1958 National Child Development Study and 1971 women aged 34 years from the 1970 British Cohort Study. We used structural equation modelling to examine the direct and indirect effects of intelligence measured at age 10–11 years, parental and current social class, educational attainment and age at first birth on smoking during pregnancy and current smoking status.
Results Forty per cent of women in the 1958 cohort smoked during pregnancy, compared with 28% of those from the 1970 cohort. In both cohorts, women with lower IQ in childhood were more likely as adults to smoke during pregnancy and to be a smoker currently. Structural equation modelling showed that the effects of childhood IQ on smoking behaviour were indirect, as they were statistically mediated by educational attainment and age at first birth. There was some effect of educational attainment and age at first birth on smoking behaviour over and above the effect of intelligence.
Conclusion Childhood intelligence influenced women's smoking behaviour as mothers primarily through its contributions to educational attainment and age at first birth.
Keywords intelligence, smoking, pregnancy, age at first birth, women
Accepted 27 August 2008
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Smoking during pregnancy increases the risk of obstetrical complications and can have substantial adverse effects on offspring, including fetal or infant death, growth restriction and a higher risk of respiratory and other illness in infancy and childhood.1 Exposure to second-hand tobacco smoke after birth is associated with similar adverse effects on child health.2
Although pregnancy can trigger some women to stop smoking,3,4 many women continue the habit,5 and of those who manage to stop for at least part of their pregnancies, many start smoking again after their children are born. One factor that might help to explain this is intelligence, as measured by standard psychometric tests. Findings that children who scored higher on tests of intelligence are less likely subsequently to start to smoke6,7 and more likely, once started, to give up8 suggest that intelligence is an important determinant of health behaviours and, perhaps, health literacy—the ability to understand and act on health information.9 Intelligence is known to influence other aspects of women's behaviour regarding their children's health, such as whether and how long to breastfeed,10,11 but there has been no investigation of the effect of prior intelligence on mothers’ smoking behaviour. Smoking in pregnancy is linked with lower educational attainment, disadvantaged socioeconomic status and younger age at first birth,4,12–17 all of which are strongly associated with lower prior intelligence.18–20 If intelligence does influence mothers’ smoking behaviour, some or all of its effect may be statistically mediated by these factors, though it is also possible that they exert mediating effects of their own.
We carried out path analyses (using structural equation modelling) of data from two UK national birth cohorts, the 1958 National Child Development Survey and the 1970 British Cohort Study. We investigated the influence of intelligence at age 10–11 years, educational attainment, social class (both parental and current) and age at first birth on mothers’ smoking behaviour. Given that the girls in each cohort grew up during periods in which the strength of the health messages about smoking directed at women differed as the evidence accumulated,21 we explored whether the magnitude of these predictors of smoking differed between the two cohorts.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The National Child Development Study22 (1958 cohort) and the 1970 British Cohort Study23 (1970 cohort) each comprise over 17 000 live births in Great Britain between March 3 and 9, 1958 and April 5 and 11, 1970, respectively.
Mental ability of the 1958 cohort was assessed at school when the children were aged 11 years using a general ability test, devised by the National Foundation for Educational Research in England and Wales.24 The test consisted of 40 verbal and 40 non-verbal items and was administered by teachers. Scores from this test correlate strongly with scores on a test of verbal ability used to select 11-year-old children for secondary school (r = 0.93),24 suggesting a high degree of validity. Mental ability of the 1970 cohort was assessed at school when the children were aged 10 years using a modified version of the British Ability Scales.25 These scales consisted of four tests: word definitions and word similarities which were used to measure verbal ability and recall of digits and matrices which were used to measure non-verbal ability.
We carried out a principal axis factor analysis of the positively correlated scores from the four tests used in the 1970 cohort in order to test for the presence of a general cognitive ability factor.26 Examination of the scree slope suggested the presence of a single component. The first unrotated factor accounted for 56% of the total variance among the four tests. The loading of each of the tests on the first unrotated factor was 0.46 for matrices, 0.40 for digit recall, 0.79 for word definitions and 0.79 for word similarities. We saved scores for each participant based on the first unrotated factor from the factor analysis.
Detailed data on pregnancy histories, including information on smoking during pregnancy and current smoking status, were collected from women in the 1958 cohort in 1991 when they were aged 33 years and from women in the 1970 cohort when they were aged 30 in 1999–2000 and again at age 34 years in 2004–05. The questionnaires used are available electronically from the UK Data Archive (http://www.data-archive.ac.uk).27–29
Data on parental social class were collected when the 1958 cohort were aged 11 years and the 1970 cohort were aged 10 years. Parental social class was derived from father's occupation or mother's occupation if no father was present and information on mother's occupation was available.30 Data on current occupational social class and highest academic qualifications were obtained for the 1958 cohort when they were aged 33 years and for the 1970 cohort when they were aged 34 years. If no data were available on current social class, due, for example, to women not being employed while caring for their family, information on social class from the previous follow-up in adulthood was used instead.
In total, 15 606 members of the 1958 cohort were eligible to take part in a follow-up survey in 1991 when they were aged 33 years (members who had died, emigrated or refused further follow-up were excluded from the target sample).31 Of 11 413 people interviewed, 5799 were women of whom 4424 (76%) had given birth. Of these, 3325 (82%) had data on smoking in pregnancy, current smoking status and childhood IQ. In total, 13 197 members of the 1970 cohort were eligible to take part in the 34-year follow-up (members who had died, emigrated, refused further follow-up or who had not taken part in any survey since the age of 10 years were excluded from the target sample).32 Of 9665 (73%) who agreed to be interviewed, 5039 were women of whom 2810 (56%) had given birth. Of these, 1971 (70%) had data on smoking in pregnancy, current smoking status and childhood IQ. Details of changes in the populations and samples of the two cohorts over time have been published.33
Women from the 1958 and 1970 cohorts who did not participate in the follow-ups at ages 33 and 34, respectively, had a slightly lower mean childhood IQ score than those who took part: in the 1958 cohort mean (SD) IQ score was 97.2 (14.6) vs 103 (14.2), while in the 1970 cohort, mean (SD) IQ was 97.3 (15.2) vs 102.1 (14.5), both P < 0.001. Some cohort members were missing data on parental social class in childhood or current social class.
Statistical analysis
We used ANOVA, t-test and the chi-square test to examine the characteristics of the participants. For the analysis of IQ score according to parental or current social class, we used estimated means and standard deviations by maximum likelihood under the assumption that missing social class data are missing at random. For ease of interpretation when describing the cognitive ability scores of the two cohorts in relation to participants’ characteristics, we transformed them into IQ equivalents, giving the scores in each cohort a mean of 100 and SD of 15. We used Cohen's formula to calculate effect sizes (d) for the differences in IQ according to smoking in pregnancy or current smoking status, whereby the difference in mean scores between two groups is divided by the pooled standard deviation.34
We used structural equation modelling as implemented in Mplus 4.2135 to measure the relations among the variables. As many of the variables were ordered categorical, we used probit regressions based on robust weighted least squares estimation. Because some of the ordered categorical variables functioned as both independent and dependent variables in our conceptual model, the theta parameterization was necessary. This method allows analysis of cases with missing data under the assumption that the data are missing at random, which means that ‘missingness’ is permissible even when it is related to covariates or outcomes, so long as covariate status does not determine presence or absence of data.36,37 Smoking in pregnancy and smoking status at age 33 (1958 cohort) or 34 years (1970 cohort) were the outcome variables. We hypothesized that the effects of childhood intelligence on smoking behaviour would be direct and also mediated via educational attainment and age at first pregnancy. Parental social class was predicted to influence childhood intelligence and to affect pregnancy, educational attainment and adult social class. Reciprocal effects of educational qualifications and adult social class on age at first pregnancy were of course possible, but the model was not identified if both sets of effects were included, and the effects we elected to model dominated in preliminary tests. The direction of this path was not a central part of the model and reversal of it would not affect the results.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Table 1 shows the characteristics of the participants in relation to childhood IQ score. In both cohorts mean IQ scores were higher in women from a more advantaged social background, whether in childhood or adulthood, in those with higher academic attainments and in those who first gave birth aged 25 or over.
|
Table 2 shows the distribution of smoking outcomes in the two cohorts and how these outcomes varied according to IQ in childhood. In total, 1315 (39.5%) of the women from the 1958 cohort reported that they had smoked during one or more of their pregnancies, compared with 560 (28.4%) of the women from the 1970 cohort. In the 1958 cohort, 29.1% of the women were current smokers compared with 32.6% of the women from the 1970 cohort. In both the 1958 and the 1970 cohorts the proportion of current smokers was considerably higher among women who had smoked in pregnancy: 73% and 78%, respectively. Women who had smoked in pregnancy had a lower mean IQ in childhood compared with those who had not: a difference of 5.3 points in the 1958 cohort and 2.7 points in the 1970 cohort (Table 3). Effect sizes for this relation in the two cohorts were both medium in size: Cohen's d = 0.37 and 0.20, respectively. There was a more pronounced difference between the cohorts as regards mean difference in IQ according to current smoking status. While in the 1958 cohort, women who were current or ex-smokers had a childhood IQ that was 4.1 points lower than non-smokers, in the 1970 cohort, there was a 1.2 point difference, and the effect size was correspondingly smaller.
|
|
Figure 1 shows path diagrams from structural equation modelling predicting (a) smoking in pregnancy and (b) current smoking status in the two cohorts.
|
In both cohorts, there was no direct effect of childhood intelligence on either smoking in pregnancy (Figure 1a) or smoking status (Figure 1b). The effects of intelligence were statistically mediated by age at first birth and educational attainment. For the most part, parental social class also had no significant direct effect on smoking behaviour, though it also influenced both smoking in pregnancy and smoking status indirectly via age at first birth and educational attainment. However, in the 1970 cohort only, parental social class did have some additional direct influence on the likelihood of smoking in pregnancy: the odds of smoking in pregnancy were slightly increased in women from more disadvantaged backgrounds in childhood. Adult social class had no direct effect on smoking behaviour in the 1958 cohort, but in the 1970 cohort, the odds of smoking in pregnancy and of still being a smoker were increased in women from lower occupational social classes. Older age at first birth and greater educational attainment were associated with lower odds of having smoked during pregnancy and of still being a smoker in both cohorts. In both cohorts, age at first birth and educational attainment appeared to have had small mediating effects of their own on smoking behaviour, as the proportions of variance in the outcome variables explained increased when these factors were added to the model as predictors of the outcomes.
There was some evidence of differences between the two cohorts in the strength of some of the observed associations. Parental social class had a stronger effect on age at first birth in the 1970 cohort than in the 1958 cohort. Similarly, age at first birth had a greater influence on educational attainment in the 1970 cohort than in the 1958 cohort. Educational attainment had a stronger influence on smoking behaviour in the 1958 cohort than in the 1970 cohort.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
In these analyses of women from two national birth cohorts, we found that women who smoked during pregnancy, or who continued to smoke subsequently, had a lower mean IQ in childhood. Path analysis, however, showed that intelligence in childhood had only indirect effects on women's smoking behaviour through its effects on age at first pregnancy and educational attainment. In both cohorts, older age at first pregnancy and greater educational attainment were associated with a lower likelihood of smoking while pregnant and of being a current smoker. Intelligence in childhood strongly predicted both of these variables in their own rights. Because of the possibility of self-selection into education on the basis of intelligence, these variables can only meet the conditions for mediational (intermediary causal) roles recommended by Baron and Kenny38 if they also served to increase our ability to predict the smoking behaviours of interest. Though the effects were relatively small, both variables did serve this function and thus should probably be considered to have played mediating roles in these analyses.
The results document the additional insight provided by path analysis over multiple regression. Though the additional variance in smoking behaviours explained by the direct contributions of educational attainment and age at first birth was small, these direct contributions are important in understanding how intelligence influenced smoking behaviour. Though intelligence strongly predicted educational attainment, it did not do so perfectly at all. Moreover, there is evidence that the genetic and environmental influences involved in educational attainment vary with level of intelligence.39 Thus, it is probably inappropriate either in general or in these data to think of educational attainment only as a proxy for intelligence. Rather, it appears that intelligence in childhood functions partly as a spur to the development of other behaviours such as pursuit of education that in turn both place people in social contexts and influence their individual perspectives in ways relevant to smoking behaviours.
Although some previous studies using data from the USA have reported associations between women's intelligence and their smoking behaviour as mothers, their focus was primarily the confounding role played by mothers’ intelligence in the often-demonstrated association between maternal smoking in pregnancy and offspring's cognitive or neuropsychological function.12,15,16 No previous study has examined the interrelations between intelligence in girls, intermediary factors and their later smoking behaviour as mothers. Path analysis using structural equation modelling allowed us to explore the effect of intelligence in 10- and 11-year olds on their subsequent smoking behaviour during and after pregnancy and examine potential mediating factors more explicitly than is possible in regression. Though we obtained evidence for some mediating effects, further analyses are needed to determine the underlying mechanisms.
The link between educational attainment and smoking behaviour is well established,4,12–14 Our results suggest that, though childhood intelligence is important in explaining this association, education itself also plays some direct role. There are several possible reasons for this. Women with greater educational attainment may have more knowledge and appreciation of the potential health consequences of smoking for themselves and their child. They are also less likely to come from a social environment where smoking is a common mechanism for coping with stress,13 and they may have better access to cessation treatments.40 On the other hand, because prior intelligence is so strongly predictive of educational outcomes,18 education may be one of the primary means through which the effects of intelligence on health behaviours are expressed. Indeed, it is possible to conceive of both higher intelligence and greater educational attainment as providing indicators of the reasoning and learning skills needed to understand and act on health messages about smoking.9 That the educational attainment of the women in our study had a strong direct influence on their smoking behaviour, independently of their IQ scores at age 10 or 11, might be because it is acting in part as a marker of cognitive abilities that have developed since that age, possibly as a result of their involvement in education. Impulsivity,41 peer pressure42 and emotional regulation43 all influence smoking behaviour. It is possible that the cognitive abilities relevant for decision making in this context may be captured more accurately by level of educational attainment than by IQ score at age 10 or 11 years.
Although educational attainment was associated with smoking behaviour in both cohorts, it had a slightly stronger effect in the 1958 cohort than in the 1970 cohort. One explanation for this may be that when health information about smoking in pregnancy was less widely available, as was the case when the 1958 cohort were growing up,21 knowledge and appreciation of the health consequences of smoking may have been more concentrated among those with higher educational attainment. It was not until the 1970s that women started to become a major focus of health education campaigns and adverts warning against smoking were targeted specifically at pregnant women.21
The other intermediary through which intelligence in childhood influenced subsequent smoking behaviour was age at first birth. In both cohorts, higher intelligence was associated with later age at first birth. In the only previous study to examine the relation between intelligence, education and age at first birth using structural equation modelling,20 IQ had a direct influence on timing of first birth. Young women who are selected into early motherhood by lower intelligence may thereby be set on a pathway to socio-economic disadvantage, which increases the likelihood of persistent smoking.44
Our finding in this study that the influence of intelligence in childhood on girls’ subsequent smoking behaviour as mothers was only indirect is consistent with results from a previous study that used structural equation modelling to examine the relation between childhood IQ and later smoking behaviour in a population of men and women. The effect of intelligence measured at age 11 years on cigarette consumption in these older men and women acted only indirectly through level of deprivation in adulthood. This study did not address the question of whether adult deprivation exerted its own mediating effect on cigarette consumption.45
The strengths of this study are its size, resulting in high statistical power and the representativeness of the sample, resulting in a high degree of generalizability for British women born around the same time. There are also some limitations. First, there has inevitably been some attrition from both cohorts over time. The women who took part in the follow-ups at age 33 or 34 years did gain higher IQ scores in childhood than those who did not participate, though the size of the differences was modest (0.3 of a SD in both cohorts). Unless the relation between childhood mental ability and smoking behaviour is different in non-participants to that found in the present analyses, little bias will have been introduced. Second, our study only includes those women who had experienced their first pregnancy by age 33 in the case of the 1958 cohort or by age 34 in the case of the 1970 cohort. With the trend in the UK towards later age at first pregnancy, particularly among more advantaged socioeconomic groups,46,47 women at the higher end of the IQ distribution may perhaps be under-represented particularly among the 1970 cohort members in our study. Finally, the data on adult social class and educational attainment relate to occupation and academic or vocational qualifications at the time of follow-up, some years after the majority of women had given birth for the first time. As a result, in the current analyses we have only been able to examine the influence of age at first pregnancy on the educational qualifications or occupational social class achieved by the time of follow-up, rather than exploring how these factors measured prior to motherhood might affect age at first pregnancy.
In this study of women from two national birth cohorts, born 12 years apart, there were indications that the prevalence of smoking during pregnancy has fallen over time. In both cohorts, older age at first pregnancy and greater educational attainment were associated with a lower likelihood of smoking while pregnant and of being a current smoker. Higher childhood intelligence influenced these women's smoking behaviour as mothers indirectly by helping to determine age at first birth and educational attainment, but it had no direct effect. There was some effect of educational attainment and age at first pregnancy on smoking behaviour over and above the effect of intelligence.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
We thank the UK Data Archive, University of Essex, for providing the data. The original data creators, depositors or copyright holders, the funding agencies and the UK Data Archive bear no responsibility for the analyses and interpretation presented here. D.B. is a Wellcome Trust Fellow. W.J. is an RCUK Fellow. The MRC and the University of Edinburgh provide core funding for the MRC Centre for Cognitive Ageing and Cognitive Epidemiology which supported the preparation of this manuscript.
Conflict of interest: None declared.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
1 BMA Board of Science and Education and Tobacco Control Resource Centre. Smoking and reproductive life. In: The impact of smoking on sexual, reproductive and child health (2004) London: British Medical Association.
2 DiFranza JR, Aligne CA, Weitzman M. Prenatal and postnatal environmental tobacco smoke exposure and children's health. Pediatrics (2004) 113:1007–15.
3 Brenner H, Mielck A. The role of childbirth in smoking cessation. Prev Med (1993) 22:225–36.[CrossRef][Web of Science][Medline]
4 Graham H, Der G. Patterns and predictors of tobacco consumption among women. Health Educ Res (1999) 14:611–18.
5 Acharya G, Jauniaux E, Sathia L, Griffin M, Morgan H. Evaluation of the impact of current antismoking advice in the UK on women with planned pregnancies. J Obstet Gynaecol (2002) 22:498–500.[CrossRef][Medline]
6 Batty GD, Deary IJ, Schoon I, Gale CR. Mental ability across childhood and risk factors for premature mortality in adult life: the 1970 British Cohort Study. J Epidemiol Community Health (2007) 61:997–1003.
7 Martin LT, Fitzmaurice GM, Kindlon DJ, Buka SL. Cognitive performance in childhood and early adult illness: a prospective cohort study. J Epidemiol Community Health (2004) 58:674–79.
8 Taylor MD, Hart CL, Davey Smith G, et al. Childhood mental ability and smoking cessation in adulthood: prospective observational study linking the Scottish Mental Survey 1932 and the Midspan studies. J Epidemiol Community Health (2003) 57:464–65.
9 Davis TC, Wolf MS. Health literacy: implications for family medicine. Fam Med (2004) 36:595–98.[Web of Science][Medline]
10 Der G, Batty GD, Deary IJ. Effect of breast feeding on intelligence in children: prospective study, sibling pairs analysis, and meta-analysis. Br Med J (2006) 333:945.
11 Silva PA, Buckfield P, Spears GF. Some maternal and child developmental characteristics associated with breast feeding: a report from the Dunedin Multidisciplinary Child Development Study. Aust Paediatr J (1978) 14:265–68.[Web of Science][Medline]
12 Baghurst PA, Tong SL, Woodward A, McMichael AJ. Effects of maternal smoking upon neuropsychological development in early childhood: importance of taking account of social and environmental factors. Paediatr Perinat Epidemiol (1992) 6:403–15.[Medline]
13 Graham H. When Life's a Drag: Women, Smoking and Disadvantage (1993) London: HMSO.
14 Owen L, Penn G. Smoking and Pregnancy: a Survey of Knowledge, Attitudes and Behaviour, 1992–9 (1999) London: Health Development Agency.
15 Batty GD, Der G, Deary IJ. Effect of maternal smoking during pregnancy on offspring's cognitive ability: empirical evidence for complete confounding in the US National Longitudinal Survey of Youth. Pediatrics (2006) 118:943–50.
16 Breslau N, Paneth N, Lucia VC, Papp E, Paneth-Pollak R. Maternal smoking during pregnancy and offspring IQ. Int J Epidemiol (2005) 34:1053.
17 Cnattingius S, Akre O, Lambe M, Ockene J, Granath F. Will an adverse pregnancy outcome influence the risk of continued smoking in the next pregnancy? Am J Obstet Gynecol (2006) 195:1680–86.[CrossRef][Web of Science][Medline]
18 Deary IJ, Strand S, Smith P, Fernandes C. Intelligence and educational achievement. Intelligence (2007) 35:13–21.[CrossRef][Web of Science]
19 Deary IJ, Taylor MD, Hart CL, et al. Intergenerational social mobility and mid-life status attainment: influences of childhood intelligence, childhood social factors, and education. Intelligence (2005) 33:455–72.[CrossRef][Web of Science]
20 Neiss M, Rowe DC, Rodgers JL. Does education mediate the relationship between IQ and age of first birth? A behavioural genetic analysis. J Biosoc Sci (2002) 34:259–75.[Web of Science][Medline]
21 Berridge V, Loughlin K. Smoking and the new health education in Britain 1950s-1970s. Am J Public Health (2005) 95:956–64.
22 Power C, Elliott J. Cohort profile: 1958 British birth cohort (National Child Development Study). Int J Epidemiol (2006) 35:34–41.
23 Elliott J, Shepherd P. Cohort profile: 1970 British Birth Cohort (BCS70). Int J Epidemiol (2006) 35:836–43.
24 Douglas J. The Home and the School. A Study of Ability and Attainment in Primary School (1964) London: MacGibbon and Kee.
25 Elliott C, Murray D, Pearson L. British Ability Scales (1978) Windsor: National Foundation for Educational Research.
26 Carroll J. Human Cognitive Abilities: A Survey of Factor-analytical Studies (2006) New York: Cambridge University Press.
27 University of London. Institute of Education. Centre for Longitudinal Studies. National Child Development Study: Sweep 5, 1991 [computer file]. (2007) City University. Social Statistics Research Unit [original data producers]. Colchester, Essex: UK Data Archive [distributor] SN: 5567.
28 University of London. Institute of Education. Centre for Longitudinal Studies. 1970 British Cohort Study: Thirty-Four-Year Follow-up, 2004–2005 (2007) [computer file]. Colchester, Essex: UK Data Archive [distributor] SN: 5585.
29 University of London. Institute of Education. Centre for Longitudinal Studies. 1970 British Cohort Study: Twenty-Nine-Year Follow-up, 1999–2000 (2007) [computer file]. Joint Centre for Longitudinal Research [original data producers]. Colchester, Essex: UK Data Archive [distributor] SN: 5558.
30 Office of Population Censuses and Surveys. Classification of Occupations 1980 (1980) London: HMSO.
31 Prescott-Clarke P, Moon N, Manley D. National Child Development Survey Fifth Sweep - Report on Survey Methodology (1993) London: Social Statistics Research Unit, City University.
32 Simmonds N, Fuller E, Lessof C, Foudouli V. 1970 British Cohort Study 2004-5 Survey Technical Report (2007) London: Centre for Longitudinal Studies, Institute of Education.
33 Plewis I, Calderwood L, Hawkes D, Nathan G. edn. In: The National Child Development Survey and the 1970 British Cohort Study Technical Report: Changes in the NCDS and the BCS70 Populations and Samples Over Time (2004) 1st. London: Centre for Longitudinal Studies.
34 Cohen J. Statistical Power Analysis for the Behavioral Sciences (1977) New York: Academic Press.
35 Muthen JM, Muthen BO. Mplus User's Guide (2006) 4th. Los Angeles: Muthen & Muthen.
36 Little RJ, Rubin DB. Statistical Analysis with Missing Data (2002) 2nd. New York: John Wiley.
37 Collins LM, Schafer JL, Kam CM. A comparison of inclusive and restrictive strategies in modern missing data procedures. Psychological Methods (2001) 6:330–51.[Web of Science][Medline]
38 Baron RM, Kenny DA. The moderator mediator variable distinction in social psychological-research - conceptual, strategic, and statistical considerations. J Pers Soc Psychol (1986) 51:1173–82.[CrossRef][Web of Science][Medline]
39 Johnson W, Deary IJ, Iacono WG. Genetic and environmental selection processes underlying educational attainment: something for everyone but everything for none. Psych Methods (2008) in press.
40 Fernandez E, Schiaffino A, Borrell C, et al. Social class, education, and smoking cessation: long-term follow-up of patients treated at a smoking cessation unit. Nicotine Tob Res (2006) 8:29–36.[Abstract]
41 Grano N, Virtanen M, Vahtera J, Elovainio M, Kivimaki M. Impulsivity as a predictor of smoking and alcohol consumption. Pers Individ Dif (2004) 37:1693–700.
42 Michell L, West P. Peer pressure to smoke: the meaning depends on the method. Health Educ Res (1996) 11:39–49.
43 Kassel JD, Evatt DP, Greenstein JE, Wardle MC, Yates MC, Veilleux JC. The acute effects of nicotine on positive and negative affect in adolescent smokers. J Abnorm Psychol (2007) 116:543–53.[CrossRef][Web of Science][Medline]
44 Graham H, Francis B, Inskip HM, Harman J. Socioeconomic lifecourse influences on women's smoking status in early adulthood. J Epidemiol Community Health (2006) 60:228–33.
45 Taylor MD, Hart CL, Smith GD, et al. Childhood IQ and social factors on smoking behaviour, lung function and smoking-related outcomes in adulthood: linking the Scottish Mental Survey 1932 and the Midspan studies. Br J Health Psychol (2005) 10:399–410.[CrossRef][Web of Science][Medline]
46 Buxton J, Clarke L, Grundy E, Marshall CE. The long shadow of childhood: associations between parental social class, own social class, educational attainment and timing of first birth; results from the ONS Longitudinal Study. Popul Trends (2005) 121:17–26.[Medline]
47 Ferri E, Smith K. Partnership and parenthood. In: Changing Britain, Changing Lives. Three Generations at the Turn of the Century—Ferri E, Bynner J, Wadsworth M, eds. (2003) London: Institute of Education Press.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Power, B. J. M. H. Jefferis, and O. Manor Childhood Cognition and Risk Factors for Cardiovascular Disease in Midadulthood: The 1958 British Birth Cohort Study Am J Public Health, January 1, 2010; 100(1): 129 - 136. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||