IJE Advance Access originally published online on January 30, 2006
International Journal of Epidemiology 2006 35(3):658-663; doi:10.1093/ije/dyl011
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Life Course Epidemiology |
Height at age 18 years is a strong predictor of attained education later in life: cohort study of over 950 000 Swedish men
1 Department of Genetics and Pathology, Rudbeck laboratory, Uppsala University, Sweden
2 Department of Public Health Sciences, Karolinska Institute, Stockholm, Sweden
3 Present address: Department of Medical Epidemiology and Biostatistics, Karolinska Institute, Stockholm, Sweden
* Corresponding author. Child and Adolescent Public Health Epidemiology Group, Department of Public Health Sciences, Karolinska Institute, Norrbacka, SE-17176 Stockholm, Sweden. E-mail: finn.rasmussen{at}ki.se
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Background Adult body height has been related to socioeconomic position in cross-sectional studies. Intelligence, shared family factors, and non-familial circumstances may contribute to associations between height and attained education, but their relative importance has been difficult to resolve.
Methods A nation-wide record-linkage cohort study of over 950 000 Swedish men born 1950–75 followed with respect to attained education for up to 27 years after measurement of height at age 18 (baseline). The association between height and attained education in later life was investigated by logistic regression modelling with adjustment for age, geography, parental socioeconomic position, and cognitive ability. Shared family factors were accounted for in analyses of full-brother-pairs using conditional logistic regression.
Results The odds ratio (OR) for attaining higher education 7–27 years after baseline was 1.10 [95% confidence interval (95% CI) 1.09–1.10] in fully adjusted models per 5 cm increase in height. Men taller than 194 cm were two to three times more likely to obtain a higher education as compared with men shorter than 165 cm. The association remained within brother-pairs, OR 1.08 (95% CI 1.07–1.10), suggesting that non-familial factors contribute to the association between height and education attainment. A significant interaction (P < 0.0001) was found between year of birth, height, and attained education, showing slightly weaker associations among later birth cohorts.
Conclusions The strong positive association between height and educational achievement remaining after adjustment for year of birth, parental socioeconomic position, other shared family factors, and cognitive ability may reflect educational discrimination based on height although residual confounding cannot be ruled out.
Keywords Adult height, education, intelligence, socioeconomic factors, heritability, discrimination
Accepted 10 January 2006
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Body height in early adulthood is influenced by both genetic and environmental factors acting during prenatal growth, childhood, and puberty. Since a long time, socioeconomic factors have been known to be related to adult body height.1–4 Starvation and infectious diseases influence growth, and historically such factors probably caused most of the height differences between the poorer and richer strata of the population. The increasing wealth in many societies has led to dramatic reduction in starvation and infectious disease.5 Still the accumulated results of social and material disadvantage during foetal life and childhood may be responsible for a substantial part of the height differences between various groups in society.6
Both family7 and twin-studies show that height is a highly genetic trait in modern Western societies, with heritability (proportion of variance attributable to additive genetic effects) close to 0.90 in males and 0.75 in females.8 However, it does not follow from this that height differences between social groups are determined by genetic factors. It is somewhat puzzling that socioeconomic disparities in height seem to be persisting in Western societies in spite of greatly improved nutrition and lower occurrence of severe infectious diseases.9 It is of note that the literature is not unanimous and some studies have reported decreasing social disparities in height.10
Previous research on associations between height and socioeconomic position has generally focused on the influence of social and environmental factors on growth and adult height. However, in the light of the very high heritability of height in modern societies, we argue that it is relevant to investigate how height may influence future social outcomes. One of the most important indicators of social position is educational level. We have been able to study how height in young adulthood predicts attainment of high (university level) education in later life in a very large cohort of Swedish men born 1950–75. In our analyses we have been able to adjust for parental socioeconomic position, shared family environmental factors, and logical cognitive ability of cohort members.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Information on this study sample was obtained by linkages of the following Swedish national registers: the Multi-Generation Register (MGR), the Military Service Conscription Register (MSCR), and the Population and Housing Censuses (PHC) of 1970–2000. From the MGR all non-adopted males born in Sweden between 1950 and 1975 with information on their parents (n = 1 373 122) were identified. Of these men, 1 189 190 (86.6%) had a record in the MSCR. After removal of outliers (inclusion criteria weight 40–250 kg, height 130–225 cm) and individuals with missing conscription data (height, logical cognitive test score, date of conscription, or conscription centre), the number of men included were 1 080 189. The Swedish military service conscription examination is required by law and performed at age 18. Foreign citizenship, or severe chronic medical condition, or a handicap documented in a medical certificate, are the only reasons accepted for non-participation. The logical cognitive test has been described elsewhere.11 Information on the educational achievement up to the year 2000 was available for 1 044 462 of the included conscripts surviving and remaining in the country to that year. Maternal and paternal education, and occupation were used as indicators of socioeconomic position of the conscripts. Lack of information on socioeconomic position reduced the set of individuals with complete information to 951 219, constituting the material on which the analyses are performed. Association between height at conscription and probability of reaching at least 1 year of university level education (called higher education) by year 2000 was investigated by stratified analyses of higher education by birth-year and height in 5 cm intervals. Estimation of effect sizes and adjustments for possible confounding were done by logistic regression (LOGISTIC procedure in SAS version 8.2). Possible confounding from year of birth, conscription test centre, logical cognitive test score, and parental social position was investigated. Two measures of parental social position were used (i) the highest socioeconomic index (SEI) of either parent by the time of the conscription (eight categories: unskilled workers, skilled workers, farmers and self employed, lower level civil servant, middle level civil servant, higher civil servant, leading or free academic position, and other); and (ii) duration of maternal and paternal education (six categories: <9 years of primary school, 9 years of primary school, <3 years of secondary school, 3 years of secondary school, <3 years of higher education, and 3 or more years in higher education). The highest recorded level from PHC 1970–2000 was used for each parent. Within brother-pair analyses were performed by conditional logistic regression using the PHREG procedure in SAS version 8.2. Among all first-born full brothers, pairs discordant for higher education at year 2000 (n = 46 199 pairs) were selected for analysis. The association between height and being the higher educated brother within each pair was then estimated.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The mean height of eligible individuals (n = 951 219) at conscription was 179.2 cm (standard deviation = 6.4 cm). A total of 306 424 conscripts (32%) had obtained at least 1 year of higher education by the year of 2000.
The probability of achieving higher education later in life increases linearly with height. This can be clearly seen by plotting the fraction of men with higher education against height at conscription (in 5 cm categories) and birth-year (in 5 year birth-cohorts) (Figure 1). Strong relations between height and higher education are found in all birth-strata. For example, among men born 1950–55 the fraction with higher education by year 2000 is 15% for men shorter than 165 cm as compared with 45% for men taller than194 cm. Inverse relations of similar strength was observed when the fraction of men having only primary schooling at year 2000 was analysed against height at conscription in 5 cm categories (data not shown).
|
Confounding from secular trends, geographic variation, parental education and SEI, and logical cognitive performance of index subjects was investigated using logistic regression (Table 1). In the unadjusted model (A), the OR for higher education is 3.5 between the tallest and the shortest category. The trend is linear with an increase in OR of 21% for each 5 cm a conscript is taller. Adjustment for birth-year and conscription test centre had practically no effect on the association (Model B), ruling out confounding due to secular or geographic variation.
|
Parental socioeconomic confounding was studied by including into the model the educational level and SEI of parents at the time of conscription (Model C). This resulted in a slight attenuation of the association between height and higher education, with the OR (95% CI) per 5 cm increase in height going from 1.21 (1.20–1.21) to 1.16 (1.15–1.16). Thus, a minor part of the height association can be explained by disparities in family socioeconomic background in this cohort.
To assess whether intelligence mediates the association between height and education, we adjusted for logical cognitive performance, which is a measure of IQ obtained at conscription examination (Model D). This attenuated the association further, with a decrease in OR per 5 cm increase in height from 1.16 to 1.10 (1.09–1.10). This reduction in OR is of similar size as that observed after adjustment for parental socioeconomic position. Thus, the relation between logical cognitive performance and height does to a significant degree contribute to the height—higher education association.
The important question of whether the association between height and education has diminished over time was tested by performing separate analyses for each 5-year birth-strata. The association is less pronounced in later birth-years (Table 1), and the interaction term between height and year of birth is significant in all adjusted models.
We then tested if inter-generational transmission of height effects on education is evident by estimating the probability of higher education as a function of height among biological fathers. Our dataset contains information on height at conscription for men born between 1950 and 1975. For a subset of conscripts born during the later years information was available on their fathers height (n = 14 321). The association of height in these fathers and higher education by year 2000 among their sons was examined. Sons of tall fathers had increased chances of reaching higher education (Table 2). For each 5 cm increase in height in the fathers, the OR increases with 15%; birth year and conscript centre adjusted OR = 1.15 (1.11–1.19). After adjustment for parental education and SEI the effect of father's height was reduced by approximately one-half [OR per 5 cm increase in height being 1.07 (1.03–1.10). Further attenuation was evident when logical cognitive performance of the son was included in the model (OR = 1.03 (0.99–1.08)].
|
A powerful test to analyse if the association between height and higher education solely relies on familial factors is to examine if the association holds also within brother-pairs. All available pairs of first-born and second-born non-twin biological sons of each mother—father combination, in total n = 168 701 pairs, were identified. Pairs in which the brothers were discordant for higher education were then selected (n = 46 199 pairs) for analysis. The OR of being the higher educated brother was 1.14 (1.12–1.16) per 5 cm difference in height in both the unadjusted model and in the model adjusting for year of birth and test centre (Table 3). The OR was attenuated to 1.08 (1.07–1.10) upon further adjustment for logical cognitive performance.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Although relationships between body size and social variables have been well documented previously we believe the results of the present investigation provide new insights in a number of ways. The strength of the association is remarkable, with a 2- to 3-fold difference in the probability of achieving higher education among tall (>194 cm), compared with short (<165 cm) men. A previous study on height stereotypes has suggested that being short would be more of a liability to less favourable social position than being tall would be an asset.12 The present results do not support this finding for educational achievement in the examined population of Swedish men. Instead a marked linearity of the effect is evident, with a steady increase in higher education with increasing height at age 18 years.
A number of possible explanations to the observed association between height and higher education could be assessed by inclusion of covariates into the analysis. Both height and education increases somewhat with year of birth. Because educational achievement is assessed at the same year (2000) for all cohort members, year of birth is by design strongly related to the number of years each individual has been monitored for the attainment of higher education. Geographic differences in height are small, but in education somewhat larger. Inclusion of birth-year and geographic information into the model left the estimates materially unchanged, showing that secular trends or geographic variation are unlikely causes of the association. Height is related to other measures of body size, which in turn are likely to be related to social position and education (but not the focus of this study). We tested the effects of including body weight into the model. A marked augmentation of the association was observed (data not shown), indicating that height is likely to be the prime dimension of body size responsible for the positive relation between size and education.
Socioeconomic factors of raising environment, as reflected by the socioeconomic position of parents during childhood, are well known to be related to both growth and educational achievement.1,13 Detailed adjustments for the parental social position are, therefore, important. By including the highest reported occupational level of either mother or father together with the educational level of both the mother and the father into the model the possible confounding from these socioeconomic aspects of the upbringing were investigated. When socioeconomic level during childhood were controlled the association became somewhat weakened with a drop in OR per 5 cm from 1.21 to 1.16. We conclude that a minor fraction of the association between height and education can be explained by differences in parental socioeconomic position.
Graduation marks from secondary school has during the covered study time been an important criteria for entering a higher education in Sweden. Cognitive abilities influence school results, which in turn influence graduation marks from primary and secondary schools. Height is related to measures of intelligence.14–16 Since head and brain size are positively correlated to intelligence17–19 and height in turn is correlated to head size,20 it might be speculated that height is related to likelihood of higher education, through some biological pathway acting via brain size and intelligence. To test whether intelligence contributes to the association between height and higher education, logical cognitive ability by the time of conscription was entered into the model. After this adjustment the association was reduced, indicating that a minor part of the positive effect of height for the chances to achieve higher educational later in life can be explained by increased cognitive ability among taller men by the age of 18. Since cognitive ability is the product of both the genetic potential of an individual as well as training by schooling, variation in schooling up to the age of 18 in itself clearly must affect cognitive ability testing results. One twin study has previously suggested that the correlation between height and education is not due to overlapping genes behind both traits.4 A recent study on Norwegian male twins found on the other hand a significant proportion (0.35) of the correlation between height and intelligence to be due to overlapping genetic factors.21 The mechanisms underlying the influence of cognitive ability on the association between height and education remain to be clarified.
The association between height and higher education diminishes somewhat with year of birth. This is in agreement with a diminishing role over time of influences of poverty and social disadvantage on foetal and childhood growth such as under-nutrition and infectious disease. Still, strong associations remain also for men born in the 1970s.
Since height has a clear genetic basis in modern society, tall individuals generally have tall parents. The probability of higher education among men coming from tall as apposed to short families should, therefore, be increased. This was investigated by using height of biological father as an indicator of ‘familial height’. Strong associations between height of the father and own achievement of higher education were evident. After adjustment for parental education and SEI, the strength of the association was reduced by one half, showing that most of the transmission of effects of height on education is abolished when socioeconomic factors were taken into account. When cognitive performance of the son was also included into the model the effect of paternal height was no longer significant. The bulk of the effect of having a tall father can thus be explained by socioeconomic and cognitive factors.
There is little reason to believe that starvation or disease constitutes the mechanism by which socioeconomic background influences the relation between height and education in the investigated population. A general view is instead that such influence comes from accumulated results of social and material conditions during foetal life and childhood.6 It should be mentioned that even if we have used the most detailed information on socioeconomic position available, there might still be residual confounding from socioeconomic variation between families not captured by the variables included in the ordinary logistic regression models. Conditional logistic regression was, therefore, used to test the association within biological full-brother-pairs discordant for higher education. The results show that the taller of the two brothers also tended to be the one with the higher education. This still holds true after adjustment for birth-order (birth year) and logical cognitive performance. We interpret this result as evidence for non-familial mechanisms contributing to the association between height and education.
An alternative pathway through which socioeconomic background influences the relation between height and education may be culturally founded co-variation of height and socioeconomic position over generations. Tallness is clearly of social advantage for males in most populations.12,13 Tall men are, for example, preferred as dates22 and employees.12 Of special interest here are findings showing that prejudices about competence among children are influenced by height. Taller boys are perceived as more competent than shorter boys of the same age in school environments from preschool onwards.12
Even after secular, geographic, socioeconomic, cognitive, and general familial factors have been accounted for, height remained positively associated with higher education. Various types of inequalities exist in society. Discrimination owing to sex, ethnic origin, weight (obesity), and age are well-known examples. This study suggests that height may be an additional member on this list although residual confounding cannot be ruled out.
KEY MESSAGES
|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
1 Meyer HE, Selmer R. Income, educational level and body height. Ann Hum Biol 1999;26:219–27.[CrossRef][Web of Science][Medline]
2 Walker M, Shaper AG, Wannamethee G. Height and social class in middle-aged British men. J Epidemiol Community Health 1988;42:299–303.
3 Judge TA, Cable DM. The effect of physical height on workplace success and income: preliminary test of a theoretical model. J Appl Psychol 2004;89:428–41.[CrossRef][Web of Science][Medline]
4 Silventoinen K, Kaprio J, Lahelma E. Genetic and environmental contributions to the association between body height and educational attainment: a study of adult Finnish twins. Behav Genet 2000;30:477–85.[CrossRef][Web of Science][Medline]
5 Padez C. Stature and stature distribution in Portuguese male adults 1904–1998: the role of environmental factors. Am J Human Biol 2002;14:39–49.[CrossRef]
6 Kuh D, Wadsworth M. Parental height: childhood environment and subsequent adult height in a national birth cohort. Int J Epidemiol 1989;18:663–68.
7 Brown WM, Beck SR, Lange EM, Davis CC, Kay CM, Langefeld CD et al. Age-stratified heritability estimation in the Framingham Heart Study families. BMC Genet 2003;4 (Suppl 1):S32.
8 Silventoinen K, Sammalisto S, Perola M, Boomsma DI, Cornes BK, Davis C et al. Heritability of adult body height: a comparative study of twin cohorts in eight countries. Twin Res 2003;6:399–408.[CrossRef][Web of Science][Medline]
9 Cavelaars AE, Kunst AE, Geurts JJ, Crialesi R, Grotvedt L, Helmert U et al. Persistent variations in average height between countries and between socio-economic groups: an overview of 10 European countries. Ann Hum Biol 2000;27:407–21.[CrossRef][Web of Science][Medline]
10 Peck AM, Vagero DH. Adult body height and childhood socioeconomic group in the Swedish population. J Epidemiol Community Health 1987;41:333–37.
11 Gunnell D, Harrison G, Rasmussen F, Fouskakis D, Tynelius P. Associations between premorbid intellectual performance, early-life exposures and early-onset schizophrenia. Cohort study. Br J Psychiatry 2002;181:298–305.
12 Jackson LA, Ervin KS. Height stereotypes of women and men: the liabilities of shortness for both sexes. J Soc Psychol 2001;132:433–45.
13 Silventoinen K. Determinants of variation in adult body height. J Biosoc Sci 2003;35:263–85.[CrossRef][Web of Science][Medline]
14 Humphreys LG, Davey TC, Park RK. Longitudinal correlation analysis of standing height and intelligence. Child Dev 1985;56:1465–78.[CrossRef][Medline]
15 Abbott RD, White LR, Ross GW, Petrovitch H, Masaki KH, Snowdon DA et al. Height as a marker of childhood development and late-life cognitive function: the Honolulu-Asia Aging Study. Pediatrics 1998;102:602–09.
16 Teasdale TW, Owen DR, Sorensen TI. Intelligence and educational level in adult males at the extremes of stature. Hum Biol 1991;63:19–30.[Medline]
17 Andreasen NC, Flaum M, Swayze V 2nd, O'Leary DS, Alliger R, Cohen G et al. Intelligence and brain structure in normal individuals. Am J Psychiatry 1993;150:130–34.
18 Paradiso S, Andreasen NC, O'Leary DS, Arndt S, Robinson RG. Cerebellar size and cognition: correlations with IQ, verbal memory and motor dexterity. Neuropsychiatry Neuropsychol Behav Neurol 1997;10:1–8.[Medline]
19 Tisserand DJ, Bosma H, Van Boxtel MP, Jolles J. Head size and cognitive ability in nondemented older adults are related. Neurology 2001;56:969–71.
20 Bushby KM, Cole T, Matthews JN, Goodship JA. Centiles for adult head circumference. Arch Dis Child 1992;67:1286–87.
21 Sundet JM, Tambs K, Harris JR, Magnus P, Torjussen TM. Resolving the genetic and environmental sources of the correlation between height and intelligence: a study of nearly 2600 Norwegian male twin pairs. Twin Res Hum Genet 2005;8:307–11.[Medline]
22 Hensley WE. Height as a basis for interpersonal attraction. Adolescence 1994;29:469–74.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  B. Galobardes, J. Lynch, and G. Davey Smith Measuring socioeconomic position in health research Br. Med. Bull., February 6, 2007; (2007) ldm001v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Vagero and B. Modin Commentary: The associations between height, cognition, and education and their relevance for health studies Int. J. Epidemiol., June 1, 2006; 35(3): 663 - 664. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||