2012. Vol.2, No.4, 394-400
Published Online October 2012 in SciRes (http://www.SciRP.org/journal/sm) http://dx.doi.org/10.4236/sm.2012.24052
Copyright © 2012 SciRes.
Does IQ Vary Systematically with All Measures of
Socioeconomic Status in a Cohort of Middle-Aged,
and Older, Men?
Shona J. Kelly1,2, Nicholas R. Burns3, Greta Bradman3, Gary Wittert2, Mark Daniel1,4
1Sansom Institute for Health Research, University of South Australia, Adelaide, Australia
2Freemason’s Foundation Centre for Men’s Health and the Discipline of Medicine, University of Adelaide,
3School of Psychology, University of Adelaide, Adelaide, Australia
4Department of Medicine, St. Vincent’s Hospital, The University of Melbourne, Parkville, Australia
Received July 6th, 2012; revised August 13th, 2012; accepted August 24th, 2012
Differences in IQ have been offered as an explanation for socioeconomic gradients in morbidity and mor-
tality. Previous research has largely relied on linking education and conscription testing data with later
life health. As this early life testing was used to determine a person’s academic path it is difficult to dis-
entangle the effects of IQ from education. This study used IQ and socioeconomic status (SES) data col-
lected concurrently in mid-life from men who did not experience IQ-test-driven career path direction in
early life. If IQ is associated with SES generally then multiple domains of IQ it will be associated with all
components of SES. In a subsample of men aged 35 - 80 (n = 287) from the Florey Adelaide Male Ageing
Study, we evaluated relationships between each of four domains of cognitive ability (IQ domains): fluid
(Gf); crystallised (Gc); visual/spatial (Gv) and processing speed (Gs). SES was measured as standardized
education, income, occupational prestige and deprivation score. Age-adjusted linear regression was used
to test each SES-z-score individually against each IQ domain. Then all four SES measures were included
in a single model for each IQ domain. This study found that a panel of standard IQ tests were positively
associated with attained education but not with income or area-level deprivation score. Two IQ abilities,
Gf and Gc, were also associated with occupational prestige score. These associations suggest that lesser
levels of health associated with lower socioeconomic status is not accounted for by a lesser innate ability
and that intervention may be possible.
Keywords: Intelligence; Social Class; Health Inequalities; Socioeconomic Status
Socioeconomic gradients in morbidity and mortality are well
established; each step in the socioeconomic hierarchy has poor-
er health, shorter life expectancy and earlier onset of disease
than the step above (Marmot, 2004). Although the term “socio-
economic” makes it clear that material deprivation is a major
contributor at the bottom of the social hierarchy, it is less clear
why, even in economically well-off populations, differences
can still be found through middle and into higher social posi-
tions. Differences in IQ have been offered as an explanation
(Batty, Deary, & Gottfredson, 2007) and the area of research
has been dubbed “cognitive epidemiology” (Deary & Batty,
2007). Intelligence quotient (IQ), is considered to index, albeit
imperfectly, innate cognitive ability.
If cognitive ability, as indexed by IQ, is a fundamental driver
of socioeconomic gradients in health and is not only acting via
education, then, at least within a gender, it should also be
graded across multiple measures of social position. Cognitive
epidemiologists support their hypothesis with the observation
that scores on IQ tests are worse the lower the social status
(Neisser et al., 1996) and it is presumed that people with low IQ
are unable to understand health education messages. A system-
atic review of scores from childhood and early adulthood found
that tests taken at either time are associated with greater later
mortality (Calvin et al., 2011). Proponents of the hypothesis
also note that low IQ is “clustered” with a set of measures of
low social standing (Deary, Batty, Pattie, & Gale, 2008). There
are several problems with this argument. First, higher IQ is
associated with greater educational attainment which suggests a
possible effect at the upper end of the socioeconomic spectrum
as well as the bottom. Also, many of the IQ tests used in the
previous research were used to stream children or young adults
into graded academic pathways (Pearce, Deary, Young, &
Parker, 2006: pp. 1-21; Hemmingsson, Essen, Melin, Allebeck,
& Lundberg, 2007: p. 1412) which suggests a clear link be-
tween educational attainment and social position, but makes it
difficult to separate the effects of IQ from education in produc-
ing socioeconomic gradients in health.
There is also a significant body of research demonstrating
that the IQ scores, as opposed to cognitive abilities, of the en-
tire population have increased steadily throughout the twentieth
century to the present (the Flynn effect (Flynn, 1987)) and the
tests have to be frequently “renormed” to keep the mean IQ
score for the population at 100. While the health of the entire
population has improved socioeconomic gradients in health
have not declined in any substantive way over this same period
of time (Queensland University of Technology & Australian
S. J. KELLY ET AL.
Institute of Health and Welfare, 2004). Also, education is not
the only pathway to high income and presumably, high social
Social position, socioeconomic status (SES), and social class,
are all terms used to describe the phenomenon that humans tend
to sort themselves into social hierarchies. Research within soci-
ology and population health has demonstrated that social posi-
tion results from a number of measurable aspects of modern
Western society including gender, educational attainment, in-
come, occupational prestige and area-level measures of aggre-
gate advantage/disadvantage, and these factors are strongly
correlated (Liberatos, Link, & Kelsey, 1988). The key finding
for health is that morbidity and mortality tend to be graded
across any measure of social position as do the risk factors for
disease (Marmot, 2004).
The IQ data used in cognitive epidemiology research largely
consists of childhood IQ measures (collected between ages 7
and 14) and from armed forces conscription-testing conducted
in early adulthood. A body of research has found that childhood
IQ measures are significantly associated with parental measures
of social position such as occupation (Batty & Deary, 2005;
Chandola, Deary, Blane, & Batty, 2006; Lawlor, Batty, Clark,
McIntyre, & Leon, 2008; Pearce et al., 2006), attained educa-
tion level (Chandola et al., 2006; Lawlor et al., 2006), child-
hood family income (Lawlor et al., 2006), area-level depriva-
tion (Hart et al., 2003); and also with attained education
(Chandola et al., 2006; Lager, Bremberg, & Vagero, 2009;
Lawlor et al., 2008) and attained occupational status (Chandola
et al., 2006; Hart et al., 2003). IQ measures from conscription
cohorts have also been shown to be associated with paternal
occupation (Batty, Deary, Tengstrom, & Rasmussen, 2008;
Batty, Gale, Tynelius, Deary, & Rasmussen, 2009; Hemmings-
son et al., 2007), attained education (Batty, Shipley, et al.,
2008), attained occupational status (Hemmingsson et al., 2007),
and mid-life income (Hemmingsson et al., 2007). As these
education and conscription tests were used to determine socio-
economic pathways providing access to more academi-
cally-focused schools or placing the respondent at officer posi-
tions in the armed forces (Hemmingsson et al., 2007; Pearce et
al., 2006) it is difficult to determine whether early life labelling
or actual ability was the driving factor for a person’s mid-life
IQ is, conceptually, more complex than just the tests used in
educational or conscription assessments. Contemporary IQ
theory has converged into a model that is sometimes called
CHC theory (Carroll, 1993, Horn & Cattell, 1966, Horn & Noll,
1997). This model comprises, depending on the level of analy-
sis, approximately eight broad abilities, some of which are:
fluid ability (Gf), the basic processes of reasoning; crystallized
ability (Gc), the breadth of knowledge, experience, learning,
and acculturation; visual ability (Gv), the perception and proc-
essing of visual form and spatial relationships; and cognitive
processing speed (Gs), the rapid cognitive processing of infor-
mation. Abilities considered to be less related to education (Gf,
Gv, Gs) are less likely to be used in the data sets used by cogni-
tive epidemiology research because much of this research util-
izes pre-existing data from educational and conscription testing.
Therefore, the relationship between SES and a wider array of
cognitive abilities has not been broadly tested. Those abilities
that are less linked to education would be a better test of the
cognitive epidemiology hypothesis that socioeconomic position
is tied to IQ.
One aspect of IQ, crystalized ability (Gc), is considered to be
stable over time and shows little change with aging. This can be
confirmed if the relationship between IQ and SES holds into
middle age. Deary and colleagues (Deary, Whiteman, Starr,
Whalley, & Fox, 2004) had 80-year old study participants
complete the same tests at age 11 and again at age 80. The sub-
stantial content of these tests was similar to the measures of
modern crystallised ability (Gc) tests. The researchers found a
moderate-to-high correlation between the two tests of 0.66
which demonstrates reasonable stability. The researchers sub-
sequently found that childhood IQ was related to mortality
before age 65 but not thereafter (Hart et al., 2005). In their re-
port the authors highlighted the problem with testing the elderly
where it is hypothesized that people with low IQ are less likely
to have survived into old age and are therefore not available for
testing. In another study of older adults researchers measured
verbal and numeric abilities (similar to Gc) at ages 56 and 78.
The researchers also found high correlations between the IQ
abilities between the two time points (all r > 0.73) (Deary, Al-
lerhand, & Der, 2009). They then analysed the impact of adding
IQ to statistical models using different SES measures to explain
health outcomes and found that IQ had the greatest effect in
reducing the size of the association for education and the least
effect for area-level deprivation. They also found that the mag-
nitude of the effect of IQ on reducing the association between
SES and health varied greatly by health outcome (Batty, Der,
Macintyre, & Deary, 2006). This does not support the idea of a
generalised role for IQ in explaining health inequalities.
In summary, childhood and early adulthood IQ have been
shown to be associated with mortality but much of the evidence
is derived from IQ testing used to direct people onto more- or
less-academic pathways making it difficult to determine
whether innate ability (IQ) or the economic advantage of edu-
cation is driving the associations. Evidence from a population
where education was not the only pathways to high social posi-
tion needs to be assessed. And, given that there is a modest
relationship between early and late life IQ, does the IQ-SES
relationship still hold in mid-life?
We examined the following questions in a group of mid-
dle-aged Australian men for whom education was not the only
pathway to socioeconomic success.
1) Does a relationship between IQ and SES exist in mid-
dle-aged and older men?
2) Is IQ associated with any other measures of SES than
Data were from a sub-study of the Florey Adelaide Male
Aging Study (FAMAS). FAMAS is a multidisciplinary popula-
tion-based cohort study examining the health and health be-
haviours of 1195 randomly selected men, 35 - 80 years, living
in the northwest regions of Adelaide, Australia (Martin et al.,
2007). The study was begun in 2002 and the participants are
periodically re-examined or sent updating questionnaires. Be-
tween Dec 2005 and Feb 2007 all participants were invited to
participate in a sub-study and 300 men (aged 37 - 83) volun-
teered to complete an extensive battery of tests assessing cogni-
tive abilities. The FAMAS cohort is representative of the male
population from which it was drawn (Martin et al., 2007) and
this sub-cohort did not differ for age, country of origin, marital
Copyright © 2012 SciRes. 395
S. J. KELLY ET AL.
status, employment status, or annual income. Compared with
the source population, the entire FAMAS cohort does have a
greater proportion of men with post-secondary qualifications, in
particular university qualifications (Martin et al., 2007) while
this sub-cohort had a slightly greater proportion with post-high
school, non-university qualifications such as trade qualifica-
tions (see Table 1).
This study was approved by the Human Research Ethics
Committee of the Royal Adelaide Hospital. All subjects gave
written informed consent.
Cognitive ability was measured with eleven tests selected to
measure four broad domains of ability; these were administered
during a single session. The test- battery consisted of:
1) Two measures of fluid ability (Gf): i) Comprehensive
Ability Battery—Inductive Reasoning; (Hakstian & Cattell,
1975); and ii) Standard Raven Progressive Matrices (Raven,
Raven, & Court, 1998). Gf involves reasoning and problem
solving often with novel stimuli and is linked to cognitive com-
plexity (Schrank, 2005).
2) Three measures of crystallised ability (Gc): i) the Spot the
Word Test; ii) Mill Hill Vocabulary Scale—Senior Form ((Ra-
ven et al., 1998); and iii) the Information subtest of the
WAIS-R (Australian Adaptation, (Wechsler, 1981)). Gc is
sometimes referred to as acculturated knowledge.
3) Three measures of visuospatial ability (Gv): i) Compre-
hensive Ability Battery—Flexibility of Closure; ii) Mental
Rotations Test (Vandenberg & Kuse, 1978); and iii) Space
Relations: Paper Folding. Gv is the ability for apprehension of
spatial forms, often involving their manipulation or rotation in
Descriptive Statistics of the Study Cohort by to age group (mean (SD)
Education n (%)
<= high school 63 (22.0%)
Trade/apprenticeship 91 (31.7%)
Diploma/certificate 81 (28.2%)
university 52 (18.1%)
Income in thousands of Austr alian dollars n (%)
<12/year 20 (7.0%)
12 - <20/year 43 (15.0%)
20 - <30/year 35 (12.2%)
30 - <40/year 40 (13.9%)
40 - <50/year 36 (12.5%)
50 - <60/year 28 (9.8%)
60 - <80/year 37 (12.9%)
80+ 48 (16.7%)
Index of Rela tive Soc io -economic Disadvantage
mean (SD) 961.6 (78.7)
ANU4 Occupation Score
mean (SD) 45.0 (22.5)
imagination (Schrank, 2005).
4). Three measures of speed of processing (Gs) were used: i)
WAIS-R Digit Symbol; ii) Woodcock-Johnson III—Visual
Matching; iii) Woodcock-Johnson Revised—Cross Out Subtest
(Schrank, 2005). Gs is a measure of speed of processing and
involves the ability to perform quickly and automatically with
either over-learned or novel stimuli (Schrank, 2005).
A measurement model was fitted as shown in Figure 1 using
MPlus v6.0 (Muthén & Muthén, 2010). All loadings and the
covariances of the four latent variables were statistically sig-
nificant and the fit of the model was excellent (χ²(38) = 58.1, p
= .019, CFI = .99, TLI = .98, SRMR = .036, RMSEA = .043,
CI90 = (.018, .064). An attempt to estimate a higher-order
g-factor was not successful with the higher-order g being iden-
tical with Gf, a common occurrence (Undheim & Gustafgsson,
1987). Standardised intelligence tests are age-normed; an indi-
vidual’s score is referenced to the average score obtained by
people of their age in a large population-representative samples.
For any age, the mean IQ is defined as 100 and the standard
deviation is 15. However, in this analysis we used latent vari-
able scores derived from raw scores from the tests, adjusted
only for age at time of testing on the basis that different abilities
have different age trajectories (Salthouse, 2004).
Socioeconomic status (SES) was measured with education,
income, occupational prestige and an area-level measure. Edu-
cation was assessed as the highest level attained: less than or
equal to high school (reference); trade/apprenticeship; certifi-
cate/diploma; university. Household income categories were (in
thousands of Australian dollars): up to 20 (reference); increas-
ing in increments of 20, up to 80-or-more. The occupational
title supplied by the respondent was coded to the ANU4 scale
(Jones 2001) developed for Australian Occupations. Area-level
SES was based on one of the Socio-Economic Indexes for Ar-
eas (SEIFA) scores, developed by the Australian Bureau of
Statistics, which are derived from postcode information (Aus-
tralian Bureau of Statistics, 2003). The Index of Relative So-
cio-Economic Disadvantage (SEIFA-dis) was used in this
analysis. For analysis all SES measures were converted to
z-scores with a mean of zero and a standard deviation of one.
Linear regression was used to test each SES-z-score indi-
vidually against each IQ domain (16 regressions). Then all four
SES measures were included in a single model for each IQ
domain. All regressions also included age as a continuous vari-
able. Stata /IC 11.0 was used in all analyses (Statacorp, College
Station, TX, USA). Results are presented as unstandardised
coefficients and 95% confidence intervals.
Table 1 describes 287 men who had complete data and were
included in the analysis. Approximately one-fifth each had a
high school or less education or a university education. Seven
percent reported an income of less than 12,000 Australian dol-
lars per year. As expected education was strongly correlated
with all the IQ abilities as were income and occupation (Table
2). Income had the lowest correlation with Gc while Gv and Gs
had low correlations with occupation. Area-level SES was cor-
related only with Gc.
In the age-adjusted models that examined the relationship
between each ability and each SES measure all models but one
Copyright © 2012 SciRes.
S. J. KELLY ET AL.
Copyright © 2012 SciRes. 397
e1 e2 e3 e4 e5
e6 e7 e8 e9 e10 e11
CAB.C MRT PF
VM CO DS
.73 .87 .84 .83 .83 .74 .63 .54 .83 .84 .87
Gf Gc Gv Gs
.66 .48 .72
CAB.I = Inductive Reasoning; SPM = Standard Progressive Matrices; MH = Mill Hill Vocabulary Test; STW = Spot-the-Word; GK = General
knowledge; MRT = Mental Rotation Test; PF = Paper Fodling; VM = Visual Matching; CO = Cross Put; Ds = Digit Symbol.
Measurement model for four latent variables showing standardised parameter estimates.
Selected Pearson’s Correlation Coefficients (p) between the Standar-
dized SES variables and the IQ ability factors (n = 287).
1 2 3 4
0.12 0.18 0.21
0.047 0.002 <0.001
0.39 0.37 0.24 0.11
<0.001 <0.001 <0.001 0.07
0.40 0.19 0.35 0.14
<0.001 0.002 <0.001 0.014
0.35 0.34 0.17 0.08
<0.001 <0.001 0.004 0.160
0.32 0.40 0.18 0.10
<0.001 <0.001 0.030 0.085
found a significant relationship between the ability and the SES
domains (Tabl e 3). Gv was not associated with income.
When all the SES and ability measures were included in the
same age-adjusted model education was statistically significant
for each IQ domain and occupational prestige was statistically
significant in the models for Gf and Gc (Table 3). Neither in-
come nor area-level disadvantage score was statistically sig-
nificantly associated with IQ in any of the multivariate models.
Main Findings of This Study
Education, in this cohort of men was not the only path to
higher income (and presumably a higher SES).We found that
age-adjusted latent-variable-ability scores were primarily asso-
ciated with higher education when all SES measures were in-
cluded in the regression model. Two IQ abilities, Gc and Gf,
were also associated with occupational prestige. These findings
do not support the view that socioeconomic gradients in health
can be explained by gradients in IQ.
What Is Alrea dy Known on Thi s Topi c
By using data from a population that were not streamed into
educational/career pathways based on their results from IQ tests
we were able to test whether there was any the relationship
between innate ability and SES attainment in life. We did not
observe an association between multiple domains of intelli-
gence test scores with all measures of socioeconomic status.
Our results suggest that previous research indicating IQ asso-
ciations with health may have been largely driven by the strong
relationship between IQ and education and not by a relationship
between low cognitive ability and low social position. The ma-
jority of the research body in this area has focused on life-
course pathways that might link earlier life circumstances with
later life health and thus is not strongly comparable to this piece
of work. The only other study to compare IQ and SES collected
concurrently in middle-age was the West of Scotland Twenty-
S. J. KELLY ET AL.
Linear Regression of each ability factor score with individual standardized SES measures (n = 287).
Latent ability variable association with each
individual SES variable adjusted for age
Latent ability variable association in full model with all
SES variables and adjusted for age
Standardi zed SES b 95% CI p b 95% CI p
Education 0.42 0.29, 0.54 <0.001 0.31 0.17, 0.45 <0.001
Income 0.19 0.03, 0.35 0.018 0.02 –0.13, 0.17 0.793
Occupational prestige 0.39 0.26, 0.51 <0.001 0.18 0.03, 0.33 0.018
Area-level SES 0.18 0.04, 0.31 0.010 0.10 –0.03, 0.23 0.135
b 95% CI p b 95% CI p
Education 1.43 1.03, 1.83 <0.001 1.01 0.55, 1.47 <0.001
Income 0.58 0.07, 1.10 0.025 –0.02 –0.52, 0.47 0.932
Occupational prestige 1.40 0.99, 1.82 <0.001 0.77 0.25, 1.26 0.002
Area-level SES 0.58 0.14, 1.02 0.010 0.31 –0.10, 0.73 0.141
Visual Spatial IQ
b 95% CI p b 95% CI p
education 0.50 0.33, 0.68 <0.001 0.41 0.21, 0.61 <0.001
income 0.16 –0.06, 0.37 0.156 –0.03 –0.25, 0.18 0.760
occupational prestige 0.41 0.23, 0.59 <0.001 0.16 –0.05, 0.38 0.146
area-level SES 0.20 0.01, 0.38 0.037 0.12 –0.06, 0.30 0.194
b 95% CI p b 95% CI p
Education 1.07 0.65, 1.50 <0.001 0.77 0.28, 1.26 0.002
Income 0.73 0.21, 1.24 0.006 0.28 –0.25, 0.80 0.299
Occupational prestige 1.03 0.60, 1.47 <0.001 0.44 –0.08, 0.96 0.094
Area-level SES 0.57 0.12, 1.01 0.012 0.33 –0.11, 0.77 0.141
07 Study (Ginty, Phillips, Der, Deary, & Carroll, 2011). In this
Scottish study, people in a manual occupation had significantly
lower IQ scores than those in a non-manual occupation which is
consistent with our findings for occupational prestige but the
study did not include attained education as in this study.
Our failure to identify socioeconomic gradients in IQ is per-
haps not surprising as the debate around IQ and health often
fails to distinguish between IQ (theorised as innate ability) and
cognitive performance (as used within the medical field) which
does not make any assumptions about innate ability. It has been
repeatedly shown that exposure to chronic stress such as severe
illness, extended care-giving, or incarceration reduces cognitive
function (e.g., (Lee, Kawachi, & Grodstein, 2004)). At any age,
lower social position is also associated with more daily hassles,
more worries, financial hardship, etc (Orpana, Lemyre, & Kelly,
2007). As these stressors increase in prevalence with decreasing
social position, this phenomenon has been suggested to account
for the decline in cognitive function at an earlier age that is seen
with lower social position (McEwen & Gianaros, 2010). Anec-
dotal reports suggest that declines in cognitive performance
often reverse when the stressors are removed.
What This Study Adds
From a policy perspective these findings are encouraging.
When cognitive ability is regarded as “innate” ability it is per-
ceived as immutable (i.e., impossible to improve). But our
findings and the Flynn effect suggests that low SES people are
not incapable of understanding health messages or making life-
style changes. Or, to phrase it in more empowering terms; the
inverse relationship between health-adverse behaviours and
SES is not accounted for by intellectual ability and there is
potential for change as demonstrated by the enormous reduction
in smoking prevalence over the past six decades
(http://www.cancercouncil.com.au/3190 1/reduce-ri sks/smking -re
duce-risks/tobacco-facts/ statistics- on-smoki ng-in-a ustralia/? pp=3
Strengths and Limitati o ns
A strength of this study is the use of multiple measures of
ability. We should have been able to identify any differences in
innate ability that could not be largely attributed to education
because three of the four abilities estimated in our good fitting
Copyright © 2012 SciRes.
S. J. KELLY ET AL.
measurement model (Gf, Gv, and Gs) are not considered to
depend on formal education. The two comparable studies from
older adults have relied primarily upon tests of vocabulary,
comprehension and mathematical skills (Deary, Allerhand, &
Der, 2009; Deary, Whiteman, Starr, Whalley, & Fox, 2004)
which are consistent with the Gc ability.
Another strength is the use of four measures of social posi-
tion. The SES measures in this study cohort were not highly
correlated with each other which is in contrast to much of the
literature from Europe and North America (eg, (Treiman, 1977)
(Blundell, Dearden, Meghir, & Sianesi, 1999)). The low corre-
lation likely reflects the unusual circumstances of Australian
men in this age group where education was not the only path to
a high-paying job. Indeed one-fifth of those with only high
school or less education were in the highest two income quar-
tiles and one-third of those with a university education were in
the lowest two income quartiles. Age is unlikely to have con-
founded the results as it was accounted for in statistical analy-
ses, and because most men would have been well established in
their careers or heading into retirement in this study population
and should have already reached their maximum educational
This analysis was limited by the possibility of a survivor bias
where the most intellectually disadvantaged individuals died
young and/or failed to participate. Also the participants in the
sub-cohort are self-selected and a greater proportion had terti-
ary education than reported in 2001 census data (Holden et al.,
2005) and a slightly greater proportion than seen in census data
also reported not being in the workforce. In spite of this higher
level of education we do have a good distribution of household
income levels with 22% of this sub-study group reporting that
they were in the less than 20,000 dollars per year category or
had only high school or less education level. Compared with
the large linkage datasets used in cognitive epidemiology re-
search our, relatively, small sample size, may have limited our
ability to identify subtle differences in ability. Unfortunately,
the sample size in a project such as this is limited by the practi-
cal realities of executing such an extensive test battery. Within
the data there is evidence of variability in ability scores with no
skewness or kurtosis which suggests no systematic bias in these
This study found that a panel of standard IQ tests were posi-
tively associated with attained education but not with income or
area-level deprivation score. Two IQ abilities, Gf and Gc, were
also associated with occupational prestige score. These associa-
tions suggest that lesser levels of health associated with lower
socioeconomic status is not accounted for by a lesser innate
ability and that intervention may be possible.
This work was supported by The FLOREY Foundation at the
University of Adelaide, The Premier’s Research Fund (South
Australia) and the National Health and Medical Research
Council of Australia (project number 627227).
Shona Kelly was funded, in part, by a Fellowship from the
South Australia (SA) Centre for Intergenerational Health (CIH),
a consortium of the University of South Australia, University of
Adelaide, Flinders University, & SA Health. SK acknowledges
the CIH and the South Australia Minister for Health.
The authors thank Dr. Andre Araujo for his comments on the
Australian Bureau of Statistics (2003). Census of population and hous-
ing. Belconnen, ACT: Central Office, Australian Bureau of Statistics.
Batty, G. D., & Deary, I. J. (2005). Education and mortality: A role for
intelligence? Journal of Epidemiology & Community Health, 59,
Batty, G. D., Deary, I. J., & Gottfredson, L. S. (2007). Premorbid (early
life) IQ and later mortality risk: A systematic review. Annals of Epi-
demiology, 17, 278-288. doi:10.1016/j.annepidem.2006.07.010
Batty, G. D., Deary, I. J., Tengstrom, A., & Rasmussen, F. (2008). IQ
in early adulthood and later risk of death by homicide: Cohort study
of 1 million men. British Journal of Psychiatry, 193, 461-465.
Batty, G. D., Der, G., Macintyre, S., & Deary, I. J. (2006). Does IQ
explain socioeconomic inequalities in health? Evidence from a pop-
ulation based cohort study in the west of Scotland. British Medical
Journal, 332, 580-584. doi:10.1136/bmj.38723.660637.AE
Batty, G. D., Gale, C. R., Tynelius, P., Deary, I. J., & Rasmussen, F.
(2009). IQ in early adulthood, socioeconomic position, and uninten-
tional injury mortality by middle age: A cohort study of more than 1
million Swedish men. American Journal of Epidemiology, 169, 606-
Batty, G. D., Shipley, M. J., Mortensen, L. H., Boyle, S. H., Barefoot, J.,
Gronbaek, M. et al. (2008). IQ in late adolescence/early adulthood,
risk factors in middle age and later all-cause mortality in men: The
Vietnam Experience Study. Journal of Epidemiology & Community
Health, 62, 522-531. doi:10.1136/jech.2007.064881
Blundell, R., Dearden, L., Meghir, C., & Sianesi, B. (1999). Human
capital investment: The returns from education and training to the in-
dividual, the firm and the economy. Fiscal Studies, 20, 1-23.
Calvin, C. M., Deary, I. J., Fenton, C., Roberts, B. A., Der, G., Leck-
enby, N., & Batty, G. D. (2011). Intelligence in youth and all-cause-
mortality: A systematic review with meta-analysis. International
Journal of Epidemiology, 40, 626-644. doi:10.1093/ije/dyq190
Chandola, T., Deary, I. J., Blane, D., & Batty, G. D. (2006). Childhood
IQ in relation to obesity and weight gain in adult life: The National
Child Development (1958) Study. International Journal of Obesity,
30, 1422-1432. doi:10.1038/sj.ijo.0803279
Deary, I. J., Allerhand, M., & Der, G. (2009). Smarter in middle age, a
cross-lagged panel analysis of reaction time and cognitive ability
over 13 years in The West of Scotland Twenty-07 Study. Psychology
and Aging, 24, 40-47. doi:10.1037/a0014442
Deary, I. J., & Batty, G. D. (2007). Cognitive epidemiology. Journal of
Epidemiology & Community Health, 61, 378-384.
Deary, I. J., Batty, G. D., Pattie, A., & Gale, C. R. (2008). More intel-
ligent, more dependable children live longer. A 55-year longitudinal
study of a representative sample of the Scottish nation. Psychological
Science, 19, 874-890. doi:10.1111/j.1467-9280.2008.02171.x
Deary, I. J., Whiteman, M. C., Starr, J. M., Whalley, L. J., & Fox, H. C.
(2004). The impact of childhood intelligence on later life: Following
up the Scottish Mental Surveys of 1932 and 1947. Journal of Per-
sonality & Social Psycholo g y , 86, 130-147.
Flynn, J. R. (1987). Massive IQ gains in 14 nations: What IQ tests
really measure. Psychological Bulletin, 101, 171-191.
Ginty, A. T., Phillips, A. C., Der, G., Deary, I. J., & Carroll, D. (2011).
Cognitive ability and simple reaction time predict cardiac reactivity
in the West of Scotland Twenty-07 Study. Psychophysiology, 48,
Hakstian, A. R., & Cattell, R. B. (1975). The comprehensive ability
battery. Champaign, IL: Institute for Personality and Ability Testing.
Hart, C. L., Taylor, M. D., Davey Smith, G., Walley, L. J., Starr, J. M.,
Hole, D. J., & Deary, I. J. (2003). Childhood IQ, social class, depri-
Copyright © 2012 SciRes. 399
S. J. KELLY ET AL.
Copyright © 2012 SciRes.
vation, and their relationships with mortality and morbidity risk in
later life: Prospective observational study linking the Scottish Mental
Survey of 1932 and the Midspan Studies. Psychosomatic Medicine,
65, 877-883. doi:10.1097/01.PSY.0000088584.82822.86
Hart, C. L., Taylor, M. D., Smith, G. D., Whalley, L. J., Starr, J. M.,
Hole, D. J., & Deary, I. J. (2005). Childhood IQ and all-cause mor-
tality before and after age 65: Prospective observational study linking
the Scottish mental survey 1932 and the midspan studies. British
Journal of Health Psychology, 1 0, 153-165.
Hemmingsson, T., Essen, J. v., Melin, B., Allebeck, P., & Lundberg, I.
(2007). The association between cognitive ability measured at ages
18-20 and coronary heart disease in middle age among men: A pro-
spective study using the Swedish 1969 conscription cohort. Social
Science & Medicine, 65, 1410-1419.
Holden, C. A., McLachlan, R. I., Pitts, M., Cumming, R., Wittert, G.,
Agius, P. A., & de Kretser, D. M. (2005). Men in Australia telephone
survey (MATeS): A national survey of the reproductive health and
concerns of middle-aged and older Australian men. Lancet, 366, 218-
Lager, A., Bremberg, S., &Vagero, D. (2009). The association of early
iq and education with mortality: 65 Year longitudinal study in Mal-
mo, Sweden. British Medical Journal, 339, B5282.
Lawlor, D. A., Batty, G. D., Clark, H., McIntyre, S., & Leon, D. A.
(2008). Association of childhood intelligence with risk of coronary
heart disease and stroke: Findings from the Aberdeen Children of the
1950s Cohort Study. European Journal of Epidemiology, 23, 695-
Lawlor, D. A., Najman, J. M., Batty, G. D., O’Callaghan, M. J., Wil-
liams, G. M., & Bor, W. (2006). Early life predictors of childhood
intelligence: Findings from the Mater-university Study of Pregnancy
and Its Outcomes. Paediatric and Perinatal Epidemiology, 20, 148-
Lee, S., Kawachi, I., & Grodstein, F. (2004). Does caregiving stress
affect cognitive function in older women? Journal of Nervous and
Mental Disease, 192, 51-57.
Liberatos, P., Link, B. G., & Kelsey, J. L. (1988). The measurement of
social class in epidemiology. Epidemiologic Reviews, 10, 87-121.
Marmot, M. (2004). Status syndrome: How your social standing di-
rectly affects your health and life expectancy. London: Bloomsbury.
Martin, S., Haren, M., Taylor, A., Middleton, S., Wittert, G., & FA-
MAS. (2007). Cohort profile: The Florey Adelaide Male Ageing
Study (FAMAS). International Journal of Epidemiology, 36, 302-
McEwen, B. S., & Gianaros, P. J. (2010). Central role of the brain in
stress and adaptation: Links to socioeconomic status, health and dis-
ease. Annals of the New York Academy of Sciences, 11 86, 190-222.
Muthén, L. K., & Muthén, B. O. (2010). Mplus User’s guide (6th ed.).
Los Angeles, CA: Muthén & Muthén.
Neisser, U., Boodoo, G., Bouchard Jr., T. J., Boykin, A. W., Brody, N.,
Ceci, S. J., & Urbina, S. (1996). Intelligence: Knowns and unknowns.
American Psychologist, 51, 77-101. doi:10.1037//0003-066X.51.2.77
Orpana, H. M., Lemyre, L., & Kelly, S. J. (2007). Do stressors explain
the association between income and changes in self-rated health? A
Longitudinal analysis of the national population health survey. In-
ternational Journal of Behavioral Medicine, 14, 40-47.
Pearce, M. S., Deary, I. J., Young, A. H., & Parker, L. (2006). Child-
hood IQ and deaths up to middle age: The Newcastle Thousand
Families Study. Public Health, 120, 1020-1026.
Queensland University of Technology, & Australian Institute of Health
and Welfare (2004). Health inequalities in Australia: Mortality
health inequalities monitoring series. Canberra: AIHW.
Raven, J., Raven, J. C., & Court, J. H. (1998). Manual for Raven’s
progressive matrices and vocabulary scales. Sections 1 - 7 with 3
research supplements. San Antonio, TX: Harcourt Assessment/
Salthouse, T. A. (2004). What and when of cognitive aging. Current
Directions in Psychological Science , 13, 140-144.
Schrank, F. (2005). Woodcock-Johnson III tests of cognitive abilities.
In D. P. Flanagan, & P. L. Harrison (Eds.), Contemporary intellec-
tual assessment: Theories, tests, and issues (2nd ed., pp. 371-401).
NY: Guilford Press.
Treiman, D. J. (1977). Occupational prestige in comparative perspec-
tive. New York: Academic Press.
Undheim, J. O., &Gustafgsson, J. E. (1987). The hierarchical organiza-
tion of cognitive abilities: Restoring general intelligence through the
use of linear structural relations (LISREL). Multivariate Behavioral
Research, 22, 149-171. doi:10.1207/s15327906mbr2202_2
Vandenberg, S. G., & Kuse, A. R. (1978). Mental rotations: A group
test of three-dimensional spatial visualisation. Perceptual and Motor
Skills, 47, 599-604. doi:10.2466/pms.19188.8.131.529
Wechsler, D. (1981). Wechsler adult intelligence test-revised. San
Antonio, TX: Psychological Corporation.