Content uploaded by Madhukar Dama
Author content
All content in this area was uploaded by Madhukar Dama on Jun 13, 2015
Content may be subject to copyright.
1 23
Naturwissenschaften
The Science of Nature
ISSN 0028-1042
Volume 100
Number 6
Naturwissenschaften (2013)
100:559-569
DOI 10.1007/s00114-013-1052-3
Cognitive ability correlates positively
with son birth and predicts cross-cultural
variation of the offspring sex ratio
Madhukar Shivajirao Dama
1 23
Your article is protected by copyright and
all rights are held exclusively by Springer-
Verlag Berlin Heidelberg. This e-offprint is
for personal use only and shall not be self-
archived in electronic repositories. If you wish
to self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.
ORIGINAL PAPER
Cognitive ability correlates positively with son birth and predicts
cross-cultural variation of the offspring sex ratio
Madhukar Shivajirao Dama
Received: 2 September 2012 /Revised: 18 April 2013 /Accepted: 19 April 2013 /Published online: 9 May 2013
#Springer-Verlag Berlin Heidelberg 2013
Abstract Human populations show remarkable variation in
the sex ratio at birth which is believed to be related to the
parental condition. In the present study, the global variation
of sex ratio at birth (SRB, proportion of male offspring
born) was analyzed with respect to indirect measure of
condition, the intelligence quotient (IQ). IQ correlates
strongly with lifespan across nations, which makes it a good
indicator of health of the large populations. Relation be-
tween three standard measures of average national IQ and
SRB was studied using multiple linear regression models.
Average national IQ was positively correlated with SRB (r=
0.54 to 0.57, p<0.001). Further, IQ emerged as a powerful
predictor of SRB after controlling for the effects of all the
known covariates like fertility, maternal age, polygyny prev-
alence, wealth, son preference, latitude, low birth weight,
and neonatal mortality in the regression models. These re-
sults suggest that the striking variation of offspring sex ratio
across nations could be caused in part by the difference in
general condition of populations.
Keywords Cognitive ability .Offspring sex ratio .Sex ratio
adjustment
Introduction
Sex ratio at birth (SRB) is conventionally defined as the
number of boys being born per 100 girls and averages 105
to 107 for human populations. Genetics of sex determina-
tion imply that SRB should be theoretically equal to 100 in
the absence of extra-genetic influences. However, SRB is
slightly male biased (a median value of 105.9 is widely used
as the baseline for calculating deviations in the sex ratio)
and show marked variation between populations (Hesketh
and Xing 2006), with a consistent decline recorded post-
world war II in many industrialized states (Moller 1996;
Davis et al. 1998; Parazzini et al. 1998), though concurrent
increase was also recorded in some countries (Parazzini et al.
1998). Several physiological, socioeconomic, and environ-
mental factors correlate with variation in population SRB
values (Teitelbaum 1970;James1987). Overall, it is observed
that higher birth order, older parental age, low or high mater-
nal weight, exposure to toxins, and stressful events is associ-
ated with female biased SRB (Hardy 2002). However, the
biological basis of regulation of SRB in response to so many
extra-genetic factors is still obscure.
Sex hormone level alterations and differential embryonic
survival in response to extra-genetic stimuli are proposed as
likely mediators of SRB variation at individual level (James
1986; Boklage 2005). James noted that women treated with
gonadotropins for induction of ovulation tend to produce
excess of female offspring and proposed the hormonal hy-
pothesis of sex ratio at birth that suggests high levels of
gonadotropins around conception may select for female
offspring (James 1980). The hormonal hypothesis has been
substantiated by numerous subsequent reports of son-
favored births under higher testosterone/or estrogen levels
in humans as well as other mammals (James 1986; Grant
and Irwin 2005;Grant2007; Grant et al. 2008). Where
hormonal hypothesis has been able to explain sex ratio
adjustment around conception, extensive work by Catalano
Communicated by: Sven Thatje
Electronic supplementary material The online version of this article
(doi:10.1007/s00114-013-1052-3) contains supplementary material,
which is available to authorized users.
M. S. Dama (*)
S/O: Shivajirao Dama, Atpost: Rajeshwar, Tq: Basavakalyan, Dist,
Bidar, Karnataka, India 585331
e-mail: madhukar262@gmail.com
Present Address:
M. S. Dama
Institute of Wildlife Veterinary Research, KVAFSU, Doddaluvara,
Thorenoor post, Somavarpet Taluk,
Kodagu District, Karnataka, India 571232
Naturwissenschaften (2013) 100:559–569
DOI 10.1007/s00114-013-1052-3
Author's personal copy
has suggested that women exposed to a variety of stressors
during pregnancy have a tendency to selectively abort male
fetus (Bruckner and Catalano 2007; Catalano et al. 2009)
which could reduce population SRB during large scale stress-
ful events like earthquake, economic crisis etc. (Catalano
2003; Catalano et al. 2005,2006).
Sons require higher nourishment from mothers compared
to daughters starting from earliest stages of gestation, and
factors that adversely affect maternal condition could cause
excess of male fetal mortalities (Cagnacci et al. 2003). This
sexual dimorphism in maternal investment requirement
could skew the offspring sex based on maternal condition
without conscious efforts. Although corroborated by labo-
ratory and wild animal studies, evidence for influence of
parental condition on offspring sex has been controversial
for humans (Cameron 2004). The contrasting results are
believed to be arising due to use of disparate indices (which
may not truly reflect parental investment ability) for mea-
suring parental condition like social status, education, envi-
ronmental calamities, resource availability, etc. (Cronk
2007). Hence, using lineal measures of parental condition,
this problem could be minimized, and comparable results
may be obtained.
Brain is the most demanding organ in the body, and
cognitive development requires a very high metabolic bud-
get (which amounts to 87 % and 27 % of the total metabolic
budget in infants and adults, respectively) (Falkner and
Tanner 1986). Cognitive development will be hampered in
individuals who fail to meet these high energetic demands.
Human populations show marked differences in general
intelligence, which is attributed to a variety of genetic and
environmental factors. Individual variation in cognition
arises due to differences in prenatal development, nutrition,
home and family environment, schooling, and genetic com-
position (Sameroff et al. 1987). This is reflected by a strong
positive correlation of intelligence and better health out-
comes at various stages of life (Arden et al. 2009a). Pro-
posed mechanistic pathways that could contribute to this
correlation are (1) intelligence and health could both be
influenced by common genetic and environmental factors,
(2) intelligence could influence health, and (3) health could
influence intelligence (Arden et al. 2009a). Putting in sim-
pler words, higher cognitive abilities enhance individuals'
care of their own health because it represents learning,
reasoning, and problem-solving skills that would ultimately
improve the overall physical condition. Lynn (2006) and
Rushton (1995) proposed that with migration away from
origin, ancient humans were exposed to colder environ-
ments which provided selective pressures for higher intelli-
gence, as colder climates posed novel problems that
required higher cognitive abilities to solve (Rushton 1995;
Lynn 2006). Though controversial, these lines of thinking
are supported by strong correlation of IQ with winter high
temperature (r=−0.76) and winter low temperature (r=−0.66)
(Templer and Arikawa 2006).
It is plausible that cross-cultural differences in cognitive
abilities are a reflection of inability of certain populations to
meet the metabolic demands of developing brain (Eppig et al.
2010; Lynn 2006). Similarly, humans with lower metabolic
resources (and lower cognitive ability), could be expected to
fail more often to meet the higher physiological costs of
producing a male offspring. I explored whether striking
cross-national variation of SRB could be measured in terms
of cognitive ability (∼a measure of parental condition) using
global population data.
Methods
Sample
The sample consisted of 109 nations for which data on all
the variables (dependent, independent, and control vari-
ables) was available (Supplementary File 1). Average SRB
for the sample was 1.05 (±0.02) compared to 1.07 (±0.02)
for the world average, suggesting that the sample was rep-
resentative of the world's SRB value. Univariate analysis
was conducted for the countries for which data was available
for dependent as well as predictor variable (Table 1).
Dependent variable
The dependent variable was the ratio of male births for
each female birth for the year 2009, as estimated by
United States Central Intelligence Agency (United States
Central Intelligence Agency 2011). A ratio of 1 indicates
equal number of male and female births, whereas a ratio
above or below 1 means more male or female births,
respectively. There are differences between the CIA esti-
mates and numbers reported by the Census offices of
Switzerland, Sweden, Norway, Ireland, India, and Japan.
However, these differences are minor, and CIA data are
widely used and accepted by cross-cultural researchers
(Mace et al. 2003;Barber2004;Navara2009).
Independent variable—measure of parental condition
Cognitive ability of populations was measured in terms of
the national average intelligence quotient (IQ), originally
explained by Lynn and Vanhanen (2006). The data were
obtained by direct measurement for 113 nations and esti-
mated for 79 more nations by averaging the known IQs of
neighboring states. Estimated IQs were validated by compar-
ing them to actual measurements of IQ in the same nations
(Lynn and Vanhanen 2006). Two independent studies have
further validated these IQs by showing a positive, near-perfect
560 Naturwissenschaften (2013) 100:559–569
Author's personal copy
correlation with many other measures of cognitive ability and
educational attainments (Lynn and Mikk 2009;Rindermann
2007). Recently, Wicherts et al. (2010) proposed revised IQ
values for 17 sub-Saharan African nations (calculated using
modified selection criteria) that they argue as more accurate
than earlier calculations. To avoid bias, I will use three differ-
ent IQ data sets in the present analysis: data of Lynn and
Vanhanen (2006) (LVE; mean=84.1, median = 84.5, s.d. =
11.8); collected data only Lynn and Vanhanen (2006)(LVCD;
mean= 86.4, median=87.0, s.d. =12.0); and data with revi-
sions of Lynn and Vanhanen (2006) from Wicherts et al.
(2010) (WEAM; mean=85.0, median= 85.0, s.d.= 11.1).
Control variables
In regression analysis, to avoid the confounding effect of the
correlation of SRB and IQ, I controlled for the effects of
factors that have been correlated with population SRB by
previous studies. These factors are fertility, maternal age,
polygyny intensity, wealth, son preference, latitude, and
prenatal nutritional status.
Barber (2004) tested the coital rate hypothesis of sex ratio
at birth (that predicts more female conceptions from
intercourse around the time of ovulation) proposed by James
(1971) for 148 countries and showed that SRB correlated
with the intensity of polygyny (r=−0.41), and total fertility
(r=−0.60). Further, using multiple regression analysis; he
showed that SRB correlated positively with wealth and
mother's age. These variables will be included as control
variables in our analysis. While maternal age was known to
influence offspring sex as early as in the year 1950 (Lowe
and McKeown 1950; Macmahon and Pugh 1953), influence
of parental wealth on offspring sex was further substantiated
by recent studies (Cameron and Dalerum 2009). Navara
(2009) predicted that latitudinal variation of climatic factors
may create the difference in resource availability and con-
sumer density between temperate and tropical regions,
which could lead to cross-national variation of SRB (Navara
2009). She examined SRB in relation to latitude and
associated climatic variables (LAT) as well as with socio-
economic status (SS). Across 202 countries, SRB was pos-
itively correlated with LAT, but not with SS showing that
tropical populations produce more daughters compared to
temperate and subarctic populations. The author offered
little explanation of why this trend exists and suggested that
the latitudinal variation of SRB may represent the vestiges
Table 1 Intercorrelation of dependent, predictor, and control variables
SRB Fertility Latitude logGDP Maternal age Polygyny
prevalence
Son
preference
Low birth
weight
Neonatal
mortality rate
IQ-LVE 0.57
***
−0.72
***
0.68
***
0.65
***
0.29
***
−0.54
***
0.19
*
−0.45
***
−0.65
***
192 192 192 192 172 119 119 109 109
IQ-LVCD 0.54
***
−0.75
***
0.67
***
0.71
***
0.42
***
−0.37
**
0.23 −0.32
*
−0.79
***
113 113 113 113 101 60 60 51 51
IQ-WEAM 0.54
***
−0.66
***
0.66
***
0.59
***
0.29
***
−0.52
***
0.15 −0.46
***
−0.59
***
192 192 192 192 172 119 119 109 109
SRB −0.37
***
0.36
***
0.24
***
0.03 −0.36
***
0.36
***
−0.21
*
−0.33
**
227 227 223 189 119 119 109 109
Fertility −0.42
***
−0.74
***
−0.18
*
.61
***
−0.04 .50
***
.81
***
227 223 189 119 119 109 109
Latitude 0.40
***
0.20
**
−0.19
*
0.29
**
−0.28
***
−0.33
***
226 192 119 119 109 109
logGDP 0.34
***
−0.33
***
0.14 −0.40
***
−0.71
***
189 119 119 109 109
Maternal age 0.21
*
0.26
**
0.10 0.06
120 120 109 109
Polygyny prevalence 0.14 0.41
***
0.61
***
121 109 109
Son preference 0.23
*
0.05
109 109
Low birth weight 0.56
***
109
Values in the first row of each column are Pearson correlation coefficient (r; two-tailed level of significance *p<0.05, **p<0.01, ***p<0.001) with
sample size in lower row
Naturwissenschaften (2013) 100:559–569 561
Author's personal copy
of prior selection based on factors such as day length and
ambient temperature, which are believed to influence male
fertility and SRB (Rojansky et al. 1992; Helle et al. 2008).
Looking at the consistent link between high resource avail-
ability and male-biased sex ratios in non-human mammals
(Cameron 2004), Mathews et al. (2008) tested a large cohort
of British Women and found that 56 % of women in the
highest third of preconceptional energy intake bore boys,
compared with 45 % in the lowest third.
Polygyny was defined as men having multiple wives
simultaneously. Countries were coded as 0=generally not
accepted/polygyny is not legal in a country, 0.5=accepted
by part of the population/polygyny is only legal for some
people, or 1= generally accepted/polygyny is legal in a
country. Indicator of polygyny prevalence was obtained
from Gender, Institutions and Development Database (United
Nations Development 2009).
Total fertility rate represents the number of children that
would be born to a woman if she were to live to the end of
her childbearing years and bear children in accordance with
current age-specific fertility rates (mean=2.92, median=
2.38, s.d.=1.54). Fertility estimates (2008) were taken from
World Bank (World Bank 2011).
Maternal age was calculated as mode estimated as the
center point of the 5-year age block with the highest fertility
in a country (Barber 2004; United Nations Development
2009) (mean=25.98, median=27.00, s.d.=3.02).
Gross domestic product per capita based on purchasing
power parity (GDP, 2009) is used as a measure of wealth.
GDP data was taken from the Central Intelligence Agency,
World Factbook (United States Central Intelligence Agency
2011). Values were log transformed for normality (mean =
3.59, median=3.63, s.d.=0.51).
Son preference is prevalent in many of the Asian coun-
tries and correlates positively with SRB (Hesketh and
Xing 2006). Son preference (missing women) describes
the difference between the number of women that should
be alive (assuming no son preference) and the actual
number of women in a country. Hence, it is essential to
control for the intensity of son preference in regression anal-
ysis. Countries were assigned values between 0 (no women
are missing) and 1 to indicate the intensity of son preference.
Values for prevalence of son preference were obtained from
Gender, Institutions and Development Database (United Na-
tions Development 2009) (mean= 0.14, median= 0.00, s.d.=
0.24).
Latitude values for nations were obtained from the Central
Intelligence Agency (CIA) World Factbook (United States
Central Intelligence Agency 2011), and numerical values were
used irrespective of direction (mean= 20.19, median= 18.00,
s.d.= 13.26).
To account for prenatal nutritional status, low birth
weight (LBW) and neonatal mortality rates (NNMR) that
are proven indicators of prenatal nutrition status are includ-
ed as covariates (Organization). NNMR is the number of
deaths during the first 28 completed days of life per 1,000
live births in a given year (World Health Organization
2011), whereas LBW is the percent of live born babies that
weigh less than 2,500 g for a given time period (World
Health Organization 2011). At the population level, NNMR
and LBW represent multifaceted public-health problems
that include long-term maternal malnutrition, ill health, hard
work and poor health care in pregnancy.
Statistical analysis
Beta coefficients and standard errors were estimated using
OLS regression models. Sex ratio at birth was entered as
dependent variable, and each of IQ values were entered
along with control variables as predictors. The predictor
variables in this study show a statistically significant
intercorrelation (Table 1). The interpretation of parameter
estimates can be problematic when predictor variables are
correlated (multicollinearity) as it causes standard errors to
become inflated (Shieh and Fouladi 2003). A general rule of
thumb is that variance inflation values (VIFs) greater than 5
are a cause for concern (O'Brien 2007). Linear regression
models showed minor indication of collinearity as some
VIFs were >4. However, when following such rules of
thumb, it is advised to consider the correlation among
predictor variables during interpretation of regression
outcomes. To further confirm that OLS models are unaf-
fected by multicollinearity, ridge regression model were
constructed for each IQ data (Fox 1991). Ridge regres-
sion artificially reduces correlation coefficient of each
pair of variables by incorporating a ridge parameter to
the diagonal of a correlation matrix of highly collinear
independent variables, leading to reduced error variance
of estimators. Based on this principle, ridge regression
effectively overcomes the multicollinearity problem (Price
1977). It should be noted that polygyny prevalence was on
ordinal scale, and some researchers prefer to use nonparamet-
ric statistics with ordinal variables. However, parametric sta-
tistics were used in the present study, as both methods produce
similar results (Anderson 1961). Continent of origin was
included as a nested variable in all the regression models to
make each country an independent data point and control for
continental variation that may influence the dependent vari-
able. Statistical analyses were conducted using STATA v. 11,
SPSS v. 16, and STATISTICA v. 10.
Results
Intercorrelations among the dependent, predictor, and con-
trol variables are presented in Table 1. Sex ratio at birth
562 Naturwissenschaften (2013) 100:559–569
Author's personal copy
andIQcorrelatedatr=0.57 (n=192, p< 0.001) using
LVE, r=0.54 (n=113, p<0.001) using LVCD and r=0.54 (n
=192, p<0.001) using WEAM (Fig. 1). It can be seen that
except maternal age, all the variables showed highly
statistically significant correlation with SRB. Fertility, polyg-
yny prevalence, LBW and NNMR correlated negatively with
SRB. Latitude, logGDP, and son preference correlated
positively with SRB.
OLS regression model revealed a positive relation
between IQ-LVE and SRB after controlling for known
confounding factors (Table 2). Other statistically signifi-
cant variables were maternal age and son preference.
SRB declined with maternal age and increased with son
preference prevalence. I was concerned about inflated
standard errors due to slightly larger value of variance
inflation factor (VIF) for fertility. To address this con-
cern, ridge regression model was constructed to over-
come the problem of multicollinearity. Results of ridge
regression were similar to OLS except that maternal age
became statistically non-significant. Hence, multicollinearity
was not a serious concern for present analysis. Virtually
similar results were obtained for IQ-WEAM and IQ-LVCD
data sets (Tables 3and 4).
Fig. 1 Sex ratio at birth by intelligence coefficient (IQ). Sex ratios were
averaged for four quartiles of the average national IQ-WEAM scores
Table 2 Multiple regression analysis predicting sex ratio at birth by IQ–LVE after accounting for fertility, logGDP, maternal age, polygyny
prevalence, son preference, latitude, low birth weight and neonatal mortality rate
OLS regression model
Unstandardized Coefficients Std. Error Beta t Sig. VIF
(Constant) 0.997 0.036 27.613 0.000
IQ–LVE 0.001 0.000 0.500 4.091 0.000 2.927
Fertility −0.002 0.002 −0.151 −0.975 0.332 4.709
Latitude 0.000 0.000 0.065 0.693 0.490 1.715
logGDP 0.002 0.005 0.034 0.306 0.760 2.434
Maternal age −0.001 0.001 −0.169 −2.158 0.033 1.201
Polygyny prevalence −0.008 0.006 −0.133 −1.243 0.217 2.250
Son preference 0.026 0.008 0.271 3.234 0.002 1.376
Low birth weight 0.000 0.000 −0.008 −0.082 0.934 1.656
Neonatal mortality rate 0.000 0.000 0.175 1.247 0.215 3.871
Continent −0.002 0.002 −0.135 −1.368 0.174 1.915
Ridge regression model (λ= 0.1)
Unstandardized Coefficients Std.Err. Beta t Sig.
(Constant) 1.006 0.032 31.040 0.000
IQ–LVE 0.001 0.000 0.407 3.749 0.000
Fertility −0.002 0.002 −0.131 −1.028 0.306
Latitude 0.000 0.000 0.102 1.175 0.243
logGDP 0.001 0.005 0.029 0.293 0.770
Maternal age −0.001 0.001 −0.140 −1.843 0.068
Polygyny prevalence −0.007 0.006 −0.117 −1.190 0.237
Son preference 0.024 0.008 0.251 3.152 0.002
Low birth weight 0.000 0.000 −0.004 −0.044 0.965
Neonatal mortality rate 0.000 0.000 0.107 0.900 0.370
Continent −0.002 0.002 −0.091 −0.997 0.321
OLS model accounted for 50.1% of the variance in sex ratio at birth, F10, 98 = 9.821, p <0.001. Ridge regression model accounted for 46.36% of
the variance in sex ratio at birth, F10, 98 =8.47, p< 0.001. VIF = variance inflation factor
Naturwissenschaften (2013) 100:559–569 563
Author's personal copy
Discussion
Statistical analysis revealed that the cross-cultural variation
in birth sex ratio could be predicted by average national IQ
scores, after keeping all the known confounding variables
constant. Consistent with the predictions, as average nation-
al IQ increased (hence condition), son births increased. All
the statistical analyses showed that average national IQ was
a predictor of natal sex ratio, whether using either of two
datasets of (Lynn and Vanhanen 2006) or data of Wicherts et
al. (2010). The correlation between average national IQ and
SRB was higher than that between SRB and any other
variable which was previously shown to correlate with birth
sex ratio (Table 1).
Fertility rate, latitude, wealth, polygyny prevalence, birth
weight, and neonatal mortality rate correlated with SRB, but
the correlation was not statistically significant when the
effects of average national IQ were removed (Table 5).
However, average national IQ correlated with SRB even
after controlling for any (Table 5) or all of the control vari-
ables (r=0.363, p<0.001). Interestingly, the partial correla-
tion of SRB and average national IQ was stronger than the
value for SRB and son preference (r= 0.301, p< 0.002) after
removing the effects of all the independent variables. This
suggests that average national IQ difference could explain
cross-cultural variation of sex ratio at birth, even better than
son preference, which invariably results in higher SRB. Sex
ratio at birth was negatively correlated with fertility rate,
polygyny prevalence, low birth weight, and neonatal mor-
tality rate, which shows that humans in underdeveloped
countries tend to produce more daughters, which may reflect
their inability to meet the higher physiological costs of
producing sons. A positive correlation of wealth and SRB
further supports this deduction.
Multiple regression shows that of average national IQ,
fertility rate, maternal age, polygyny prevalence, logGDP, son
preference prevalence, latitude, low body weight, and neonatal
mortality rate, average national IQ is the best predictor of natal
Table 3 Multiple regression analysis predicting sex ratio at birth by IQ (WEAM) after accounting for fertility, logGDP, maternal age, polygyny
prevalence, son preference, latitude, low birth weight and neonatal mortality rate
OLS regression model
Unstandardized Coefficients Std. Error Beta t Sig. VIF
Constant 1.004 0.037 26.898 0.000
IQ–WEAM 0.001 0.000 0.385 3.545 0.001 2.234
Fertility −0.003 0.002 −0.200 −1.286 0.202 4.590
Latitude 0.000 0.000 0.106 1.142 0.256 1.638
logGDP 0.003 0.005 0.070 0.613 0.541 2.447
Maternal age −0.001 0.001 −0.167 −2.098 0.039 1.201
Polygyny Prevalence −0.007 0.006 −0.130 −1.189 0.237 2.264
Son preference 0.027 0.008 0.289 3.406 0.001 1.359
Low birth weight 0.000 0.000 0.006 0.063 0.950 1.689
Neonatal mortality rate 0.000 0.000 0.171 1.193 0.236 3.871
Continent −0.002 0.002 −0.090 −0.909 0.366 1.852
Ridge regression model (λ= 0.1)
Unstandardized Coefficients Std.Err. Beta t Sig.
Constant 1.009 0.033 30.122 0.000
IQ–WEAM 0.001 0.000 0.334 3.362 0.001
Fertility −0.002 0.002 −0.159 −1.247 0.215
Latitude 0.000 0.000 0.129 1.485 0.141
logGDP 0.003 0.005 0.057 0.564 0.574
Maternal age −0.001 0.001 −0.140 −1.823 0.071
Polygyny Prevalence −0.007 0.006 −0.116 −1.164 0.247
Son preference 0.025 0.008 0.264 3.289 0.001
Low birth weight 0.000 0.000 0.008 0.086 0.932
Neonatal mortality rate 0.000 0.000 0.100 0.832 0.407
Continent −0.001 0.002 −0.062 −0.676 0.501
OLS model accounted for 48.20% of the variance in sex ratio at birth, F10, 98 = 9.107, p<0.001. Ridge regression model accounted for 45.01% of
the variance in sex ratio at birth, F10, 98 =8.02, p< 0.001. VIF = variance inflation factor
564 Naturwissenschaften (2013) 100:559–569
Author's personal copy
sex ratio by a large margin (refer to beta coefficients in Tables 2,
3,and4). The effects of fertility rate, polygyny prevalence,
logGDP, latitude, maternal age, low body weight, and neonatal
mortality rate are not significant, whereas son preference have
discrete predictive power beyond intelligence. Although the
effects of variables other than IQ and son preference are not
revealed by the regression model, it cannot be said that these
factors are not involved. An indirect link exists between all the
Table 4 Multiple regression analysis predicting sex ratio at birth by IQ–LVCD after accounting for fertility, logGDP, maternal age, polygyny
prevalence, son preference, latitude, low birth weight and neonatal mortality rate
OLS regression model
Unstandardized Coefficients Std. Error Beta t Sig. VIF
(Constant) 0.968 0.028 34.842 0.000
IQ–LVCD 0.001 0.000 0.484 3.022 0.004 3.028
Fertility 0.000 0.002 0.032 0.149 0.882 5.350
Latitude 0.000 0.000 −0.065 −0.591 0.558 1.428
logGDP 0.005 0.004 0.191 1.289 0.205 2.586
Maternal age 0.000 0.001 0.080 0.742 0.462 1.379
Polygyny prevalence 0.002 0.005 0.051 0.345 0.732 2.563
Son preference 0.018 0.008 0.277 2.222 0.032 1.827
Low birth weight 0.000 0.000 −0.021 −0.166 0.869 1.954
Neonatal mortality rate 0.000 0.000 −0.240 −1.035 0.307 6.325
Continent 0.000 0.001 −0.020 −0.137 0.892 2.424
Ridge regression model (λ= 0.1)
Unstandardized Coefficients Std.Err. Beta t Sig.
(Constant) 0.976 0.025 38.939 0.000
IQ–LVCD 0.001 0.000 0.410 2.878 0.006
Fertility 0.000 0.002 −0.035 −0.205 0.839
Latitude 0.000 0.000 −0.043 −0.404 0.688
logGDP 0.005 0.004 0.181 1.342 0.187
Maternal age 0.000 0.001 0.086 0.828 0.413
Polygyny prevalence 0.002 0.004 0.059 0.447 0.658
Son preference 0.016 0.008 0.246 2.137 0.039
Low birth weight 0.000 0.000 −0.009 −0.071 0.943
Neonatal mortality rate 0.000 0.000 −0.206 −1.141 0.261
Continent 0.000 0.001 0.016 0.121 0.904
OLS model accounted for 66.1% of the variance in sex ratio at birth, F10, 40 = 7.79, p<0.001. Ridge regression model accounted for 62.45% of the
variance in sex ratio at birth, F10, 40= 6.65, p <0.001. VIF = variance inflation factor
Table 5 Partial correlation of independent variables and sex ratio at birth, after the effect of each variable was removed
Control variable↓IQ (LVE) Fertility Latitude logGDP Maternal age Polygyny prevalence Son preference LBW NNMR
IQ (LVE) −0.04 0.19 −0.01 −0.11 −0.02 0.30
**
0.10 0.13
Fertility 0.45
***
0.30
**
−0.01 −0.09 −0.08 0.37
***
0.04 0.12
Latitude 0.50
***
−0.33
***
0.24
*
−0.10 −0.27
**
0.28
**
−0.09 −0.21
*
logGDP 0.55
***
−0.36
***
0.41
***
−0.14 −0.26
**
0.33
***
−0.09 −0.13
Maternal age 0.62
***
−0.48
***
0.46
***
0.35
***
−0.36
***
0.38
***
−0.20
*
−0.32
***
Polygyny prevalence 0.54
***
−0.34
***
0.40
***
0.21
*
−0.04 0.42
***
−0.07 −0.14
Son preference 0.60
***
−0.50
***
0.42
***
0.32
***
−0.21
*
−0.44
***
−0.31
***
−0.37
***
Low birth weight 0.60
***
−0.44
***
0.43
***
0.28
**
−0.09 −0.32
***
0.41
***
−0.26
**
Neonatal mortality rate 0.57
***
−0.39
***
0.40
***
0.16 −0.09 −0.23
*
0.39
***
−0.03
*
p<0.05,
**
p<0.01,
***
p<0.001, two-tailed
Naturwissenschaften (2013) 100:559–569 565
Author's personal copy
variables, as these factors correlate with each other and with
SRB (Table 1). These sources of endogeneity must be consid-
ered during interpretation of the regression models.
Numerous studies have tested TWH in humans, but the
results have been largely equivocal (Fujita, Roth et al.2012;
Kolk and Schnettler 2012; Pollet and Nettle 2010;Song
2012; Stein et al. 2004a,b; Hopcroft 2005; Almond and
Edlund 2007; Pollet et al. 2009). However, as TWH pre-
dictions hold true only at individual level, present results
could not be considered as support for TWH.
That SRB is better predicted by the average national IQ
than by fertility rate, polygyny prevalence, and son preference
suggests that cross-cultural variation in SRB is not simply a
consequence of variation in women's mating tactics. This is
important, as mating tactics are believed to have strong influ-
ence on offspring sex ratio (Whiting 1993; Barber 2004;
Hesketh and Xing 2006). Indeed, in my sample, average
wealth, which select for sons also showed a weaker correlation
with SRB as compared to average national IQ. This could be
due to intelligence being a reflection of genetic quality, with
very high heritability (McClearn, Johansson et al. 1997),
whereas wealth is an acquired resource that is neither
directly related to genetic quality nor shows high heritabil-
ity. Mace et al. 2003) noted that low birth sex ratio within
African populations is correlated with fertility. They
suggested that this may be genetically determined, as black
Americans living in the USA also have SRB values similar
to that of African populations (Martin et al. 2002). How-
ever, in my analysis, IQ correlated stronger than fertility
with SRB, across the world. Human populations with
higher wealth and adequate healthcare facilities have lower
fertility rate (r=−0.737, p<0.001) and higher IQ scores (r
=0.649, p<0.001), which suggest that with improvement
in socioeconomic condition brought-about by growth of
wealth, variables like fertility and IQ could show marked
improvements. Hence, it is unlikely that the relation be-
tween SRB and fertility as suggested by Mace et al. (2003)
or SRB and IQ from my analysis is solely mediated by
genetic factors. Also, it is unlikely that the cross-cultural
variation in birth sex ratio is the result of parental efforts to
enhance reproductive success because humans breed mostly
within cultures.
Hence, the logical basis for a positive correlation of SRB
and average national IQ could be derived from the sex
differences in the costs of offspring production. Male fetus
grow faster (Marsal et al. 1996) and require higher parental
investment during gestation (Tamimi et al. 2003), which
means that women should be in optimal condition to meet
the cost of male offspring. Indeed, male fetus is more
spontaneously aborted than female fetus due to nutritional
deficiencies and exposure to environmental toxins (Byrne
1987; Boklage 2005). It is therefore possible that various
factors that influence maternal investment ability (indirectly
measured by cognitive ability in this study) are more likely
Table 6 Robust clustered mul-
tiple regression analysis
predicting sex ratio at birth by
IQ (LVE and WEAM) after
accounting for fertility, logGDP,
maternal age, polygyny preva-
lence, son preference, latitude,
low birth weight, and neonatal
mortality rate (data are clustered
by continent)
Std. err. adjusted for six clusters
in continent; R-squared first
model= 0.491, second
model= 0.477; N= 109
Coef. Robust std. err. tP>t[95 % Conf. interval]
Model I
IQ-LVE 0.001 0.000 5.190 0.004 0.001 0.002
Fertility −0.002 0.002 −1.210 0.279 −0.008 0.003
Latitude 0.000 0.000 1.360 0.233 0.000 0.001
logGDP 0.001 0.005 0.220 0.834 −0.013 0.015
Maternal age −0.001 0.001 −0.950 0.384 −0.005 0.002
Polygyny prevalence −0.005 0.004 −1.390 0.224 −0.014 0.004
Son preference 0.025 0.004 6.550 0.001 0.015 0.035
Low birth weight 0.000 0.000 −0.090 0.932 −0.001 0.001
Neonatal mortality rate 0.000 0.000 1.830 0.127 0.000 0.001
Constant 0.991 0.035 28.190 0.000 0.901 1.081
Model II
IQ–WEAM 0.001 0.000 2.660 0.045 0.000 0.002
Fertility −0.003 0.002 −1.590 0.172 −0.008 0.002
Latitude 0.000 0.000 1.730 0.145 0.000 0.001
logGDP 0.003 0.006 0.500 0.640 −0.012 0.018
Maternal age −0.001 0.001 −1.060 0.339 −0.005 0.002
Polygyny prevalence −0.006 0.004 −1.420 0.215 −0.016 0.005
Son preference 0.027 0.004 6.310 0.001 0.016 0.037
Low birth weight 0.000 0.000 0.050 0.965 −0.001 0.001
Neonatal mortality rate 0.000 0.000 1.680 0.153 0.000 0.001
Constant 0.998 0.027 36.500 0.000 0.928 1.068
566 Naturwissenschaften (2013) 100:559–569
Author's personal copy
than genetic factors to form the basis for striking cross-
cultural variation in SRB, without any adaptive significance.
However, my analysis still fails to rule out the role of
genetic factors, but provide a strong support for the relation
between condition and sex ratio. Alternatively, many stud-
ies have showed a lack of association between parental
condition (measured by physical indices) and offspring
sex ratio (Cramer and Lumey 2010; Pollet and Nettle
2010;Steinetal.2004a,b), which means that cognitive
ability may have some direct correlation with sex ratio, if
parental condition is not influencing SRB variation across
cultures. Interestingly, a direct correlation between IQ and
SRB could be constructed by considering the results of
two studies: (1) men with higher IQ scores produce high
quality semen (Arden et al. 2009b;Pierce2009)and(2)
red deer with high quality semen father proportionately
larger number of male offspring (Gomendio 2006), which
could produce a positive correlation of cognitive ability
andoffspringsexratio.
Similarly, the role of other variables not considered in
this study cannot be ruled out. For example, ambient tem-
perature (Helle et al. 2008), geography (Chambliss 1949;
Navara 2009), environmental stressors (Fukuda et al. 1998;
Zorn et al. 2002), and their interactions (Helle et al. 2009)
associated with SRB variation in large human
populations, are likely to play a marked role in the
cross-cultural variation of SRB. It is therefore, necessary
to establish the extent to which such factors contribute
to SRB variation by conducting further studies using
world-wide data.
Lack of data for all the countries lead to unintentional
exclusion of many populations in my regression models.
However, the averages for all the study variables were
comparable with global values. This limitation should be
noted before generalizing the correlation of IQ and SRB to
excluded populations. Cross-cultural studies employing
small sample size are criticized for exacerbation of the
problem of non-independence (i.e., individual country,
which is unit of analysis, may not be necessarily truly
independent statistical data point). To overcome this prob-
lem, I have coded the countries by continents and included
this variable in the regression models, to make each country
an independent data point (Navara 2009). Further, clustered
regression models, where data were clustered by continents,
also produced results (Table 6) similar to linear and ridge
regression models.
In summary, I report a positive relationship between
cross-cultural variation in the birth sex ratio and average
national IQ (a proxy for condition). Across cultures, as
average national IQ increases, humans produce more sons.
While previous research has shown systematic variation of
SRB within a single culture (Gibson and Mace 2003), my
results demonstrate systematic cross-cultural variation of
sex ratio at birth. The striking differences in SRB across
cultures, however, most likely do not have adaptive signif-
icance, as reproductive constraints imposed by parental
condition (which varies across countries) explain the SRB
variation better than the parental efforts to enhance repro-
ductive success. Further research is needed to determine the
causal nature of strong correlation of SRB and cognitive
ability. This could be achieved by retrospective or prospective
analysis of offspring sex differences in a large cohort of
subjects with distinct cognitive abilities.
Acknowledgments Opinions expressed here are solely those of the
author and not necessarily of any institute, where the author was
employed since the work was started in 2005. I would also disclose
that this body of work was conducted at home as a hobby starting from
the year 2005. I dedicate this research work to my loving wife Savitri
and sweet daughter Adhya (Gubbi) for tolerating my indulgence in this
work.
Funding No funding to disclose.
Conflicts of interest None declared.
References
Almond D, Edlund L (2007) Trivers–Willard at birth and one year:
evidence from US natality data 1983–2001. Proc Biol Sci
274(1624):2491–2496
Anderson NH (1961) Scales and statistics: parametric and nonparamet-
ric. Psychol Bull 58:305–316
Arden R, Gottfredson LS et al (2009a) Does a fitness factor contribute
to the association between intelligence and health outcomes?
Evidence from medical abnormality counts among 3654 US
Veterans. Intelligence 37(6):581–591
Arden R, Gottfredson LS et al (2009b) Intelligence and semen quality
are positively correlated. Intelligence 37(3):277–282
Barber N (2004) Sex ratio at birth, polygyny, and fertility: a cross-
national study. Biodemography Soc Biol 51(1):71–77
Boklage CE (2005) The epigenetic environment: secondary sex ratio
depends on differential survival in embryogenesis. Hum Reprod
20(3):583–587
Bruckner T, Catalano R (2007) The sex ratio and age-specific male
mortality: evidence for culling in utero. Am J Hum Biol
19(6):763–773
Byrne J, Warburton D et al (1987) Male excess among anatomically
normal fetuses in spontaneous abortions. Am J Med Genet
26(3):605–611
Cagnacci A, Renzi A et al (2003) The male disadvantage and the
seasonal rhythm of sex ratio at the time of conception. Hum
Reprod 18(4):885–887
Cameron EZ (2004) Facultative adjustment of mammalian sex ratios in
support of the Trivers–Willard hypothesis: evidence for a mech-
anism. Proc Biol Sci 271(1549):1723–1728
Cameron EZ, Dalerum F (2009) A Trivers–Willard effect in contem-
porary humans: male-biased sex ratios among billionaires. PLoS
One 4(1):e4195
Catalano RA (2003) Sex ratios in the two Germanies: a test of the
economic stress hypothesis. Hum Reprod 18(9):1972–1975
Naturwissenschaften (2013) 100:559–569 567
Author's personal copy
Catalano R, Bruckner T et al (2005) Fetal death sex ratios: a test of the
economic stress hypothesis. Int J Epidemiol 34(4):944–948
Catalano R, Bruckner T et al (2006) Exogenous shocks to the human
sex ratio: the case of September 11, 2001 in New York City. Hum
Reprod 21(12):3127–3131
Catalano R, Ahern J et al (2009) Gender-specific selection in utero
among contemporary human birth cohorts. Paediatr Perinat
Epidemiol 23(3):273–278
Chambliss R (1949) The geographic factor in the human sex ratio at
birth. Soc Forces 28(2):190–195
Cramer JS, Lumey LH (2010) Maternal preconception diet and the sex
ratio. Hum Biol 82(1):103–107
Cronk L (2007) Boy or girl: gender preferences from a Darwinian point
of view. Reprod Biomed Online 15:23–32
Davis DL, Gottlieb MB et al (1998) Reduced ratio of male to female
births in several industrial countries: a sentinel health indicator?
JAMA 279(13):1018–1023
Eppig C, Fincher CL et al (2010) Parasite prevalence and the
worldwide distribution of cognitive ability. Proc Biol Sci
277(1701):3801–3808
Falkner FT, Tanner JM (1986) Human growth: a comprehensive trea-
tise. Plenum Press, New York
Fox J (1991) Regression diagnostics. Sage Publications, Newbury
Park, Calif
Fujita M, Roth EA et al (2012) Low serum vitamin A mothers
breastfeed daughters more often than sons in drought-ridden
northern Kenya: a test of the Trivers and Willard hypothesis. Evol
Hum Behav 33(4):357–364
Fukuda M, Fukuda K et al (1998) Decline in sex ratio at birth after
Kobe earthquake. Hum Reprod 13(8):2321–2322
Gibson MA, Mace R (2003) Strong mothers bear more sons in rural
Ethiopia. Proc Biol Sci 270(1):S108–109
Gomendio M, Malo AF et al (2006) Male fertility and sex ratio at birth
in red deer. Science 314(5804):1445–1447
Grant VJ (2007) Could maternal testosterone levels govern mammalian
sex ratio deviations? J Theor Biol 246(4):708–719
Grant VJ, Irwin RJ (2005) Follicular fluid steroid levels and subse-
quent sex of bovine embryos. J Exp Zool A Comp Exp Biol
303(12):1120–1125
Grant VJ, Irwin RJ et al (2008) Sex of bovine embryos may be related
to mothers' preovulatory follicular testosterone. Biol Reprod
78(5):812–815
Hardy I (2002) Sex ratios: concepts and research methods. Cambridge
University Press, Cambridge
Helle S, Helama S et al (2008) Temperature-related birth sex ratio bias
in historical Sami: warm years bring more sons. Biol Lett
4(1):60–62
Helle S, Helama S et al (2009) Evolutionary ecology of human birth
sex ratio under the compound influence of climate change, famine,
economic crises and wars. J Anim Ecol 78(6):1226–1233
HeskethT,XingZW(2006)Abnormalsexratiosinhuman
populations: causes and consequences. Proc Natl Acad Sci U S
A 103(36):13271–13275
Hopcroft RL (2005) Parental status and differential investment in sons
and daughters: Trivers–Willard revisited. Social Forces Social
Forces 83(3):1111–1136
James WH (1971) Cycle day of insemination, coital rate, and sex ratio.
The Lancet 297(7690):112–114
James WH (1980) Gonadotrophin and the human secondary sex ratio.
Br Med J 281(6242):711–712
James WH (1986) Hormonal control of sex ratio. J Theor Biol
118(4):427–441
James WH (1987) The human sex ratio. Part 1: a review of the
literature. Hum Biol 59(5):721–752
Kolk M and Schnettler S (2012). Parental status and gender prefer-
ences for children: is differential fertility stopping consistent with
the Trivers–Willard hypothesis? J Biosoc Sci. doi:10.1017/
S0021932012000557
Lowe CR, McKeown T (1950) The sex ratio of human births related to
maternal age. Br J Soc Med 4(2):75–85. doi:10.1136/jech.4.2.75
Lynn R (2006) Race differences in intelligence: an evolutionary analysis.
Washington Summit Publishers, Augusta, GA
Lynn R, Mikk J (2009) National differences in intelligence and educa-
tional attainment. Intelligence 35(2):115–121
Lynn R, Vanhanen T (2006) IQ and global inequality. Washington
Summit Publishers, Augusta, Ga
Mace R, Jordan F et al (2003) Testing evolutionary hypotheses about
human biological adaptation using cross-cultural comparison.
Comp Biochem Physiol A Mol Integr Physiol 136(1):85–94
Macmahon B, Pugh TF (1953) Influence of birth order and maternal
age on the human sex ratio at birth. Br J Prev Soc Med 7(2):83–86
Marsal K, Persson PH et al (1996) Intrauterine growth curves based on
ultrasonically estimated foetal weights. Acta Paediatr 85(7):843–848
Martin JA, Hamilton BE et al (2002) Births: final data for 2000.
National vital statistics reports: from the Centers for Disease
Control and Prevention, National Center for Health Statistics.
National Vital Statistics System 50(5):1–101
Mathews F, Johnson PJ et al (2008) You are what your mother eats:
evidence for maternal preconception diet influencing foetal sex in
humans. Proc Biol Sci 275(1643):1661–1668
McClearn GE, Johansson B et al (1997) Substantial genetic influence
on cognitive abilities in twins 80 or more years old. Science
276(5318):1560–1563
Moller H (1996) Change in male:female ratio among newborn infants
in Denmark. Lancet 348(9030):828–829
Navara KJ (2009) Humans at tropical latitudes produce more females.
Biol Lett 5(4):524–527
O'Brien R (2007) A caution regarding rules of thumb for variance
inflation factors. Qual Quant 41(5):673–690
Parazzini F, La Vecchia C et al (1998) Trends in male:female ratio
among newborn infants in 29 countries from five continents. Hum
Reprod 13(5):1394–1396
Pierce A, Miller G et al (2009) Why is intelligence correlated with
semen quality?: Biochemical pathways common to sperm and
neuron function and their vulnerability to pleiotropic mutations.
Commun Integr Biol 2(5):385–387
Pollet T, Nettle D (2010) No evidence for the generalized Trivers–
Willard hypothesis from British and rural Guatemalan data. J Evol
Psychol 8(1):57–74
Pollet TV, Fawcett TW et al (2009) Sex-ratio biasing towards daughters
among lower-ranking co-wives in Rwanda. Biol Lett 5(6):765–768
Price B (1977) Ridge regression: application to nonexperimental data.
Psychol Bull 84(4):759–766
Rindermann H (2007) The g-factor of international cognitive ability
comparisons: the homogeneity of results in PISA, TIMSS, PIRLS
and IQ-tests across nations. Eur J Personal 21(5):667–706
Rojansky N, Brzezinski A et al (1992) Seasonality in human repro-
duction: an update. Hum Reprod 7(6):735–745
Rushton JP (1995) Race, evolution, and behavior: a life history per-
spective. Transaction Publishers, New Brunswick, N.J., USA
Sameroff AJ, Seifer R et al (1987) Intelligence quotient scores of 4-
year-old children: social-environmental risk factors. Pediatrics
79(3):343–350
Shieh Y-Y, Fouladi RT (2003) The effect of multicollinearity on
multilevel modeling parameter estimates and standard errors. Educ
Psychol Meas 63(6):951–985. doi:10.1177/0013164403258402
Song S (2012) Does famine influence sex ratio at birth? Evidence from
the 1959–1961 Great Leap Forward Famine in China. Proc Biol
Sci 279(1739):2883–2890
Stein AD, Barnett PG et al (2004a) Maternal undernutrition and the sex
ratio at birth in Ethiopia: evidence from a national sample. Proc
Biol Sci 271(3):S37–39
568 Naturwissenschaften (2013) 100:559–569
Author's personal copy
Stein AD, Zybert PA et al (2004b) Acute undernutrition is not associ-
ated with excess of females at birth in humans: the Dutch hunger
winter. Proc Biol Sci 271(4):S138–141
Tamimi RM, Lagiou P et al (2003) Average energy intake among
pregnant women carrying a boy compared with a girl. BMJ
326(7401):1245–1246. doi:10.1136/bmj.326.7401.1245
Teitelbaum MS (1970) Factors affecting the sex ratio in large
populations. J Biosoc Sci 2:61–71
Templer DI, Arikawa H (2006) Temperature, skin color, per capita income,
and IQ: an international perspective. Intelligence 34(2):121–139
United Nations Development Programme (2009) Human development
report 2009: Overcoming barriers: human mobility and develop-
ment. Available online at http://hdr.undp.org/en/media/HDR_
2009_EN_Complete.pdf
United States Central Intelligence Agency (2011) The world factbook.
Central Intelligence Agency, Washington, DC
Whiting JWM (1993) The effect of polygyny on sex ratio at birth. Am
Anthropol 95(2):435–442
Wicherts JM, Dolan CV et al (2010) A systematic literature review of
the average IQ of sub-Saharan Africans. Intelligence 38(1):1–20
World Bank (2011) World development indicators 2010. World Bank,
Washington
World Health Organization (2011) Global health risks mortality and
burden of disease attributable to selected major risks. World
Health Organization, Geneva
Zorn B, Sucur V et al (2002) Decline in sex ratio at birth after 10-day
war in Slovenia: brief communication. Hum Reprod 17(12):
3173–3177
Naturwissenschaften (2013) 100:559–569 569
Author's personal copy
