Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

ISSN: 0025-2344

Intelligence and Group Differences in Preference for

Breasts over Buttocks
George Francis∗ Emil O. W. Kirkegaard

Abstract

Permanent adipose breasts are unique to humans among primates. We propose that permanent breasts, and a
preference for them, are an adaptation for pair bonding and a slow life history strategy. This theory predicts that races
with a slower life history strategy will prefer breasts to buttocks. To test our hypothesis, we predict preferences for
breasts over buttocks from racial admixture and both national and regional average IQ, which we use as an indicator
of differences in life history. Measures of Breast-Buttock preference are constructed for nations and subnational
regions using the relative frequencies of internet searches on Google, Pornhub and YouPorn for terms related to
breasts versus those related to buttocks. Average IQ correlates with Breast-Buttock preference at the national level
(r = .76, p < .001), within regions of American countries (r = .85, p < .001), US States (r = .54, p < .001) and US
Metro areas (r = .57, p < .001). Controlling for racial ancestry substantially moderated the effect size of intelligence.
We consistently found African ancestry to negatively predict breast preference: across countries (r = −.80, p < .001),
within US states (r = −.91, p < .001), and US Metro areas (r = −.91, p < .001). These results replicated when using
Spanish language search terms within Spanish-speaking countries, suggesting our findings are not peculiar to the
English language. Only for US Metro areas did we find a significant effect size for socioeconomic controls.

Keywords: Sex, Breasts, Intelligence, Life history, Race, Ethnicity

1 Introduction
Human females are the only primates with permanent adipose breasts. In other primates, the breasts only
develop during pregnancy for the purpose of breastfeeding. There have been many theories for the evolution
of perennial breasts in humans including thermoregulation, as a fat source akin to a camel’s hump, and as
an adaptation to bipedalism (Pawłowski & Żelaźniewicz, 2021). Currently, however, there is no agreed upon
explanation for why humans evolved this trait. For a comprehensive review of the literature see Pawłowski
and Żelaźniewicz Pawłowski & Żelaźniewicz (2021). Although the origins of human breasts are uncertain, it
is clear that they currently play the role of a secondary sexual characteristic, developing through puberty
and being arousing to the male members of the species.
A particular challenge for any sexual explanation of the evolution of breasts is that the cues created
by enlarged breasts may have been unattractive to our ancestors. In other primates, enlarged breasts signal
being pregnant, or breast feeding infants and possibly having a male partner. During pregnancy it is near
impossible to have another conception (Roellig et al., 2011) and females also face temporary infertility during
breast feeding, a phenomenon known as “lactational amenorrhea”. A primate attracted to large breasts
not only faces the risk of hostility from another male, but also receives few benefits from any resulting
copulation. That breasts have evolved as a secondary sexual characteristic, despite possibly signalling
sexually unattractive qualities in our ancestors, is a problem we term the “breast paradox”. Pawłowski &
Żelaźniewicz (2021, p. 7) refer to this issue as “the key problem” to sexual explanations of the origins of
permanent breasts.
The breast paradox might be obviated by assuming that perennial breasts evolved for non-sexual
reasons, before being co-opted into their current sexual role. However, it is still possible that the cost of
signalling the undesirable characteristics from having permanent breasts could have been too large for any
non-sexual advantage to be favoured by evolution.
A possible resolution to the breast paradox is to bite the bullet, and state that perennial breasts are
attractive precisely because they signal being pregnant or breastfeeding. A primate male that is attracted
∗
Corresponding Author; Email: [email protected]

279
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

to enlarged breasts may be more likely to stay around whilst the female is in the vulnerable situation of
pregnancy or breast feeding, increasing paternal investment. This would be especially useful if these periods
were more critical in humans compared to other primates – as is evidence by the high body weight of human
newborns. Humans are born at approximately 6% of their mothers’s body weight, compared to only 3% for
chimpanzees (DeSilva, 2011). Furthermore, the brain of a human newborn accounts for 12.3% of its body
weight compared to 10% in chimpanzees (DeSilva, 2011).
Because, in other primates, the size of breasts indicates when the female is fertile, having perennially
enlarged breasts will obscure when adult human females are fertile. Permanent breasts will thus attract men
who are constantly engaged in courting, rather than only temporarily whilst a woman is fertile. Whilst we
are unaware of prior research that makes the same argument, our idea arises naturally out of prior theory on
the origin of concealed ovulation in humans. Humans females have concealed ovulation, which makes it
difficult to tell when a female is fertile. Other primates can be in heat, having detectable estrus – visual
and behavioural cues to indicate they are ovulating. This unusual fact of concealed ovulation has been
speculated to be useful for creating long-term pair bonds and increasing paternal investment (Alexander &
Noonan, 1979). Permanent breasts, by concealing fertility, may play the same role.
This explanation of permanent breasts makes logical sense in the context of humans being pair-bonded
and having a slow life history strategy. Humans are the only primates who live in groups and still manage
to maintain monogamous relations (Benshoof & Thornhill, 1979). Life history strategy refers to the extent
that an organism has evolved to expend resources immediately in reproduction, versus saving them for
later through mechanisms such as growth or paternal investment (Hutchings, 2021). Expending resources
maintaining one mate in a pair bond may be seen as a slow life history strategy (Fletcher et al., 2015),
because males invest resources in the hope of future reproduction rather than immediate reproduction.
Humans generally have a slow life history strategy, compared to many other mammalian species, living for a
long period of time and investing in children who take a long time to mature (Fletcher et al., 2015).
Other theories of the evolution of breasts also point to them having a role in pair bonding. Under the
nubility hypothesis (Marlowe, 1998), adipose breasts evolved as an honest signal of age. Breasts droop with
age, a phenomenon known as breast ptosis, allowing them to signal a woman’s residual reproductive value.
Males interested in finding a partner who can bear children for years to come should prefer women with
large breasts which honestly signal youth. Low levels of breast ptosis would also be attractive for a slow life
history, because it signals who will have further years to nurture children and grandchildren. It has been
speculated that human females, unlike other primates, reach a menopause stage when they focus more of
their attention on nurturing offspring and grandchildren (Williams, 1957).
Another relevant theory comes from Desmond Morris’s book, The Naked Ape (1967). Morris argued
that perennial breasts had evolved to mimic the role of buttocks on the front of the female body, helping to
facilitate the enjoyment of face-to-face coitus, known as ventro-ventral copulation. This position, which
Morris considers the standard way of performing copulation in humans, is extremely rare in other species.
He notes that the purpose of this position may be especially important for humans as a pair-bonding species.
To quote Morris Morris (1967), “The frontal approach means that the in-coming sexual signals and rewards
are kept tightly linked with the identity signals from the partner. Face-to-face sex is ‘personalised sex’.” To
put it otherwise, the face-to-face position is intimate. The pair-bonding role of ventro-ventral copulation is
perhaps implied by its common name as used by laymen, “the missionary position”, a name which connotes
sanctity.
If a preference for breasts facilitates pair bonding, then we may expect races selected for a slow
life history strategy would also have evolved a greater preference for breasts. Races selected for pair
bonding would benefit from a breast preference which facilitated ventro-ventral copulation and encouraged
a preference for youth. Human breasts are still enlarged during pregnancy and breast feeding (Ashley B S et
al., 2020), thus a preference for breasts may still help to ensure that male humans provide resources during
these critical periods.
Racial differences do exist with regards to pair bonding and life history. Much evidence can be noted.
Whites are more likely to marry and remain married than Blacks (Aughinbaugh et al., 2013). Blacks are
more likely to commit adultery than Whites (Wang, 2018). Among men, Asians are lower in sociosexuality

280
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

than Whites, who in turn are lower than Africans (Sprecher et al., 2013). With regards to the need for
parental investment, there are racial differences in speed of maturation. Black newborns have stronger head
control and motor skills than White infants (Freedman & Deboer, 1979). Blacks tends to start and finish
puberty earlier than Whites, who in turn start and finish earlier than East Asians. These puberty differences
have been recorded by events such as growth in genitalia and breast development (Rushton, 1995). Asians
tend to lose their virginity later than Whites, who in turn lose their virginity later than Africans (Rushton &
Bogaert, 1987). This same pattern of East Asians then Whites then Africans appears for a range of life
history traits including intercourse frequency, rates of having twins, gestation time, skeletal development,
and lifespan (Rushton, 1995).
To test whether racial differences in breast preference are a part of life history strategy, we use national
and regional IQ as a proxy for group differences in life history. This relationship was originally suggested by
Rushton Rushton (2010). High intelligence requires costly investment, which may come at the expense of
immediate reproductive effort. Among mammalian species, a larger brain size is associated with a greater
basal metabolic rate (Isler & van Schaik, 2006). As Rushton Rushton (2004) noted1 , “Representing only
2% of body mass, the brain uses about 5% of basal metabolic rate in rats, cats, and dogs, about 10% in
rhesus monkeys and other primates, and about 20% in humans.” It is surely an investment that requires
time to attain a sufficient return on.
Associations between brain size and life history traits among species further suggest that intelligence
is related to organisms’ life history strategies. Among 21 primate species, Smith Smith (1989) found that
brain size correlates .80 to .90 with life span, length of gestation, age of weaning, age of eruption of first
molar, age at complete dentition, age at sexual maturity, interbirth interval, and body weight. Rushton
Rushton (2004) followed up Smith’s research in a study of 234 mammalian species, finding correlations
with longevity (r = .59), gestation time (r = .66), birth weight (r = .16), litter size (r = −.18), age at first
mating (r = .63) and duration of lactation (r = .61), after controlling for weight and body length.
National IQ and group differences in intelligence are likely related to these life history trade-offs.
Brain size causally increases intelligence (Lee et al., 2019) and cranial capacity has been shown to correlate
with mean IQ at .91 among a group of ten races (Rushton, 2010) and at .54 with national IQ (Francis &
Kirkegaard, 2022).
We might also note that intelligence could be related to life history for behavioural reasons. Intelligence
is associated with lower delay discounting, or what can be called patience, in humans (Shamosh & Gray,
2008) and other primates (Stevens, 2014). Apart from intelligence’s general utility in decision making,
the relationship with patience implies it can be useful for long term planning and thus a slow life history
strategy.

2 Data

2.1 Google Trends Breast Preference

The utility, Google Trends (https://trends.google.com/home), allows users to compare the popularity of
search terms. For example, if you compare the terms “red” and “blue” within the United Kingdom for the
year 2022, Google trends states that of all searches containing either ”red” or ”blue”, 54% of searches are
for “red” and 46% of searches include are for “blue”. The maximal number of words that can be compared
at once is five.
To compare the popularity of buttocks and breast per country we tested the relative popularity of
common pornographic words for each body part before choosing the two most popular terms for each. For
buttocks these words were “ass” and “butt”. For breasts, the terms were “boobs” and “tits”. By adding the
relative popularity of “boobs” and “tits” together, we attained the number of searches including “boobs”
or “tits” as a fraction of all searchers including “boobs”, “tits”, “ass” and “butt”. We call this measure
“Google Breast-Buttock preference”.
1 Rushton unfortunately does not give a source for his statistics, but they appear to be correct. Rolfe and Brown (Rolfe &
Brown, 1997) present body oxygen use and mass for different body parts in rats and Humans. The human brain is 2% of
body mass but 20% of oxygen use. In raths that brain is 1.5% of body mass and only uses 3% of oxygen use

281
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

The formula for the Google Breast preference for area i is presented in Equation 1, where each sexual
secondary sexual characteristic refers to the total number of Google searches for the term in area i.

Boobsi + T itsi
Google Breast-Buttock preferencei = (1)
Boobsi + T itsi + Assi + Butti
The time frame for our measurements is from 01/01/2004 to 12/09/2022. Google Trends provides
an option to “include low search volume regions”. We used this option to maximise our sample size. We
collated the results by nations, US states, US metropolitan areas (US Metro areas), and at the level of
regions within the American countries for which we had regional intelligence data (see subsection 2.6 for
details on the regional data).
We also created a Spanish language measure of “Google Tetas-Culos preference”. This used sexual,
Spanish terms for breasts and buttocks. It was calculated with the formula below for regions of Colombia,
Chile, and Mexico, Spanish speaking countries for which we had regional cognitive and socioeconomic data
available.

T etasi
Google Tetas-Culos Preferencei = (2)
T etasi + Culosi

2.2 Pornhub and YouPorn Breast Preference:

In 2014, statistiacians from Pornhub and YouPorn (Pornhub, 2014) released data measuring the popularity
of six categories of body parts by country and US state. These body parts were “Boobs”, “Butts”, “Feet”,
“Legs”, “Pregnant” and “Pussy”. The popularity of each term was given as an index bounded between 0
and 1. Pornhub Pornhub (2014) does explainfor how they measured the popularity of these terms, but we
suppose that it is a measure of frequency of these search terms. We operationalise a measure of breast
preference P ornhubBreastP ref erence for state i by dividing the popularity of Boobsi by the sum of
Boobsi and the popularity of Buttsi .

Boobsi
Pornhub Breast-Buttock preferencei = (3)
Boobsi + Buttsi

2.3 Reliability of Breast Preference Data

Our measures of breast preference are valid if they capture national and regional variation in preference
for breasts over buttocks and reliable to the extent that they are not noisy measures of this underlying
construct.
In Figure 1 we present the correlation between our two measures of breast preference. The extremely
large correlation (r = .91, p < .001) between our two measures evidences strong convergent validity and
implies a strong reliability. We also compare our measures at the level of US states in appendix Figure A3,
finding a correlation of (r = .92, p < .001).

2.4 National IQ
Our measure of national IQ comes from Becker and Rindermann Becker & Rindermann (2016) and is publicly
available in version 1.3.4 of the national IQ dataset2 (Accessed from: https://viewoniq.org/?page
_id=9). National IQ is measured by combining psychometric and student assessment tests administered in
different countries administered by Psychologists and multinational organisations such as the Organisation
for Economically Developed Countries (OECD). National IQ is normed relative to the United Kingdom,
assuming the country had a mean of 100 and a standard deviation of 15 as of the year 1992. Measures
of national cognitive ability show excellent convergent validity (r > .83) (Francis & Kirkegaard, 2022;
Rindermann, 2007), so we do away with replicating our results using alternative measures of national
cognitive ability. For a discussions of the validity and relaibility of national IQ see Warne Warne (2022),
Carl Carl (2022) and Rindermann Rindermann (2018).

2 See column N of the sheet named ’FAV’

282
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Figure 1: Comparsion of Google and Pornhub Breast-Buttock preference Measures across countries.

2.5 National Control Variables

To estimate the racial admixture of each country we used the “World Migration Matrix” created by
Putterman and Weil Putterman & Weil (2010). For each country in the year 2000, the matrix gives the
proportion of people who lived in different countries in the year 1500. For example, the matrix states that
18% of Americans’ ancestors lived in Great Britain in 1500. The authors created the matrix from a mix of
genetic data and textual sources.
We lump the ethnicity of each country in 1500 into one of five racial groups: Europeans, Sub-Saharan
Africans, North-East Asians, South-East Asians and Middle East or North African (MENA). Our assignment
of nations to racial groups is given in table “Ethnic_Assignment.csv” in the supplementary materials. From
this we have the ancestry proportions of countries as of 2000 divided into the five racial groups. For example,
we calculate that the USA is 9.3% Sub-Saharan African in its ancestry. Some ancestry could not be assigned
to our chosen racial groups and is used as the reference group in our regression models.
As an additional control variable, we use nations’ natural logarithm of GDP per capita. The logarithm
of GDP per capita is calculated using real GDP at purchasing power parity3 and population size from the
Penn World Tables version 10.01 (Feenstra et al., 2015). These values were taken from the year 2013
because it is the middle year in the period in which search data from Google Trends was taken. We use
the continental regions for each country as classified by the United Nations United Nations (2023), to
colour-code scatter plots.

2.6 Regional and US State Data

In addition to analysing the relationship between national IQ and breast preference, we also study the
relationship between average cognitive ability and sub-state regions within the Americas. For our explanatory
variables we employ the dataset created by Fuerst and Kirkegaard Fuerst & Kirkegaard (2016). Cognitive
ability for US States and regions of nations in the Americas are estimated from a range of scholastic
test and are standardised against the United Kingdom, employing the same approach used to calculate
national IQs. The proportion of racial admixture, which is European, African and Amerindian was estimated
3 Denoted as cgdpe in the Penn World Tables.

283
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

using both genomic data and socially identified racial ethnicity (SIRE). Cognitive ability and ancestry data
were unavailable for the District of Colombia. Ancestry data was also missing for Hawaii. The authors
estimate the welfare of regions by factor analysing indicators of welfare within nations to extract an S factor
(Kirkegaard, 2014) and then make it comparable by rescaling S factor scores for regions by their nations’
Human Development Index (HDI).

2.7 US Metropolitain Data

Google trends provides us with Breast-Buttock preference at the level of 210 US metropolitan areas. To
create metro level measures of socioeoconomic variation, racial ancestry and average cognitive ability, we
first obtained these values for all 3146 US counties before taking population weighted averages of county
level values for each metro area. We use the data of Schneider Schneider (2020), who assigns US counties
to the metropolitan areas created by Google Trends. We had missing data for 46 or 1.4% of US counties.
Missing values for these counties were imputed with the mean of values for all other counties.
Our county-level measures of socioeconomic variation, racial ancestry and average cognitive ability
were originally collated and created in Pesta et al. Pesta et al. (2021). Average cognitive ability for each
county comes from the Stanford Education Data Archive (SEDA). The data archive pools and norms test
scores from the NAEP and state tests for students in grades 8-13. Data at the county level was available
for 2009-2015. Further details regarding the calculation of these scores can be found in McLaughlin et al.
McLaughlin et al. (2003)
SEDA also provides indicators of socioeconomic status for counties. These indicators are median family
income, proportion of adults with a bachelor’s degree or higher, proportion of adults who are unemployed,
household poverty rate, proportion of households receiving food benefits, and proportion of households with
single mothers. Pesta et al. Pesta et al. (2021) used factor analysis to extract a general factor of these
variables for each county. We refer to this latent variable as the socioeconomic factor.
Finally SEDA also provides estimates of the racial ancestry of each county. This is derived from
self-identified race and ethnicity composition data for students (e.g., “percent Whites in the grade”). These
proportions are based on the annual surveys created by the US Department of Education Common Core of
Data (CCD), from the years 2006-2010.

3 Results

3.1 National Analysis

Scatter plots of national IQ and Breast-Buttock preference are presented in Figures 2 and A1. National
IQ correlates at .78 (p < .001) with Breast-Buttock preference measured with Pornhub data and at
.71 (p < .001) when measured with Google Trends data. Countries within South Asia and the Indian
subcontinent appear as outliers in both datasets, showing a much higher Breast-Buttock preference than
would be predicted by their national IQ. Scatter plots of Sub-Saharan African ancestry and Breast-Buttock
preferences are presented in Figures 3 and A2. Sub-Saharan African ancestry correlates at −0.80 (p < .001)
with Breast-Buttock preference measured with Pornhub data and at −.68 (p < .001) when measured with
Google trends data.
In Table 1 we employ control variables: racial ancestry , and the logarithm of real GDP per capita.
Controlling for racial ancestry substantially reduces the regression beta of national IQ. Across all models,
African ancestry predicts lower preference for breasts, whilst South-East Asian ancestry predicts greater
preference for breasts. North-East Asian Ancestry has no significant relationship with breast preference.
Middle East and North African admixture significantly predicts lower Breast-Buttock preference, only when
the Pornhub data is used and only at p < .05. Raw regression betas for the admixture variables are shown in
Table A1. They indicate that the effects of Sub-Saharan African admixture and South-East Asian ancestry
are of approximately equal magnitude, although opposite in direction.
The logarithm of GDP per capita shows no significant association with Breast-Buttock preference.
This indicates that National IQ’s association with Breast-Buttock preference cannot be explained by GDP
confounding or mediating the relationship.

284
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Figure 2: National IQ and Pornhub Breast-Buttock preference.

Figure 3: Sub-Saharan African Ancestry and Pornhub Breast-Buttock preference.

3.2 Regional Analysis

Figure 4 is a scatterplot of regional intelligence and regional Google Breast-Buttock preference within
nations of the Americas. In Table 2, regional IQ has a standardised beta of .85 (p < .001), which is only
slightly diminished to .80 (p < .001) after employing country fixed effects. The socioeconomic factor is not
significantly associated with Breast-Buttock preference, just as GDP per capita had no association in the
national-level regressions.

285
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Table 1: Regression Models of National Breast-Buttock preference.

Dependent variable:
Google Breast-Buttock preference Pornhub Breast-Buttock preference
(1) (2) (3) (4) (5) (6)
National IQ 0.724∗∗∗ 0.500∗∗∗ 0.491∗∗∗ 0.802∗∗∗ 0.377∗∗∗ 0.352∗∗∗
(0.054) (0.098) (0.126) (0.046) (0.071) (0.090)
Sub-Saharan African Admixture −0.301∗∗ −0.325∗∗ −0.484∗∗∗ −0.500∗∗∗
(0.101) (0.104) (0.074) (0.076)
North-East Asian Admixture −0.019 −0.020 0.054 0.056
(0.064) (0.065) (0.047) (0.048)
South Easst Asian Admixture 0.177∗∗ 0.171∗∗ 0.211∗∗∗ 0.210∗∗∗
(0.055) (0.057) (0.041) (0.042)
Middle East and North African Admixture −0.051 −0.055 −0.097∗ −0.100∗
(0.060) (0.065) (0.044) (0.048)
Log GDPpc −0.006 0.016
(0.098) (0.068)
Constant 0.003 0.011 0.018 −0.004 0.013 0.023
(0.053) (0.054) (0.057) (0.047) (0.040) (0.041)
Observations 184 152 147 193 155 151
R2 0.498 0.591 0.596 0.609 0.770 0.771
Note: ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001
All continuous variables are standardized with a mean of 0 and a standard deviation of 1. Standard errors are in parentheses.

African ancestry has a significant negative association with Breast-Buttock preference (β = −.25, p <
.001), whilst Native American ancestry has no significant association. In the appendix Table A2, regression
betas are shown without standardizing the racial admixture variables. Before controlling for admixture,
Brazil predicted preference for buttocks and Mexico predicted preference for breasts. After employing the
controls, Chile predicts substantially lower preference for breasts, whilst the United States is associated with
higher preference for breasts. To avoid perfect multicollinearity, the constant of the regressions are omitted
so every country has a regression beta.

3.3 Spanish Language Analysis

So far, we have found cognitive ability and racial ancestry to predict internet searches for breasts instead
of buttocks. To test whether our results are unique to English language searches, we repeat our analysis
using the Spanish terms “tetas” (tits) and “culos” (asses). For this test we only employ the Spanish nations
used in our regional analyses, excluding the USA and Brazil. Figure 5 shows a correlation of .46 (p < .001)
between regional IQ and “Tetas-Culos” preference. Figure 6 shows −.56 (p < .001) between Black racial
admixture and Tetas-Culos preference.
Regression models for the Spanish analysis are shown in Table 3. The same models without
standardising racial admixture percentages are represented in Table A3. After controlling for national
fixed effects, regional IQ has a substantial effect size on Tetas-Culos preference (β = .95, p < .001). After
controlling for admixture, regional IQ is no longer statistically significant (β = .23, n.s), unlike in the English
language analysis. African admixture still has a significant effect size when the Spanish terms are used
(β = −.82, p < .001), but now so does Native American admixture (β = −.66, p < .001). The general
socioeconomic factor does not have a significant effect size (β = −.16, n.s).

3.4 US States Analysis

Figure A4, in the appendix, is a scatter plot of US State IQ and Google Breast-Buttock preference showing
a correlation of r = .53 (p < .001). Figure 7 is the same scatter plot using the Pornhub Breast-Buttock
preference data, showing a correlation of r = .54 (p < .001). In regression models only employing African
admixture, we find a substantial effect size of β = −0.87 (p < .001). When admixture is controlled for, state
IQ is barely significant with the Google Breast-Buttock Preference dependent variable (β = 0.22, p < .05) and

286
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Figure 4: Regional IQ and Google Breast-Buttock preference.

Table 2: Regression Models of Regional Breast-Buttock preference in the Americas.

Dependent variable:
Google Breast-Buttock preference
(1) (2) (3)
Regional IQ 0.800∗∗∗ 0.656∗∗∗ 0.597∗∗∗
(0.092) (0.108) (0.132)
African Admixture −0.251∗∗∗ −0.251∗∗∗
(0.059) (0.059)
Native American Admixture 0.117 0.135
(0.119) (0.122)
Socioeconomic Factor 0.041
(0.053)
Brazil −0.401∗∗ −0.208 −0.249
(0.127) (0.145) (0.154)
Chile −0.233 −0.614∗∗∗ −0.642∗∗∗
(0.128) (0.144) (0.149)
Colombia 0.114 −0.016 −0.094
(0.137) (0.134) (0.168)
Mexico 0.274∗∗ −0.075 −0.111
(0.089) (0.146) (0.154)
United States 0.054 0.353∗ 0.444∗
(0.128) (0.145) (0.186)
Observations 145 144 144
R2 0.777 0.812 0.813
Note: ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001.
All continuous variables are standardized with a mean of 0 and a standard
deviation of 1. Standard errors are in parentheses.

not statistically significant when using the Pornhub Breast-Buttock preference measure (β = 0.03, p = n.s).
The socioeconomic factor has no significant effect size regardless of the dependent variable employed.

287
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Figure 5: Regional IQ and Google Tetas-Culos Preference.

Figure 6: African Admixture and Google Tetas-Culos Preference.

3.5 US Metro Area Analysis

Comparing US Metropolitan Statistical Areas (metro areas), average cognitive ability correlates with Google
Breast-Buttock preference (r = .57, p < .001), as shown in Figure 9. However, the black proportion of
metro areas explains a substantial proportion of the variance (r = −.91, p < .001), as shown in Figure 10.
Regression analysis, shown in Table 5, shows that metro area cognitive ability has no significant effect size on

288
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Table 3: Regression Models of Tetas-Culos Preference.

Dependent variable:
Google Tetas-Culos Preference
(1) (2) (3)
Regional IQ 0.573∗∗∗ 0.138 0.267
(0.129) (0.139) (0.168)
African Admixture −0.723∗∗∗ −0.690∗∗∗
(0.145) (0.146)
Native American Admixture −0.325∗∗ −0.364∗∗
(0.114) (0.117)
Socioeconomic Factor −0.163
(0.123)
Chile −0.423 −0.863∗∗∗ −0.913∗∗∗
(0.231) (0.215) (0.217)
Colombia 0.279 0.399 0.475∗
(0.211) (0.201) (0.208)
Mexico −0.002 0.118 0.087
(0.163) (0.156) (0.157)
Observations 70 70 70
R2 0.266 0.482 0.496
Note: ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001.
All continuous variables are standardized with a mean of 0 and a standard
deviation of 1. Standard errors are in parentheses.

Figure 7: US State IQ and Pornhub Breast-Buttock preference.

289
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Figure 8: Black Admixture and Pornhub Breast-Buttock preference in US States.

Table 4: Regression Models of US State Breast-Buttock preference.

Dependent variable:
Google Breast-Buttock preference Pornhub Breast-Buttock preference
(1) (2) (3) (4) (5) (6) (7) (8)
State IQ 0.625∗∗∗ 0.223∗ 0.276 0.515∗∗∗ 0.033 0.124
(0.111) (0.087) (0.137) (0.116) (0.065) (0.101)
African Admixture −0.874∗∗∗ −0.786∗∗∗ −0.778∗∗∗ −0.875∗∗∗ −0.916∗∗∗ −0.903∗∗∗
(0.071) (0.084) (0.086) (0.057) (0.063) (0.063)
Native American Admixture −0.084 −0.072 −0.226∗∗∗ −0.204∗∗
(0.077) (0.082) (0.057) (0.060)
Socioeconomic Factor −0.055 −0.095
(0.110) (0.081)
Observations 50 49 49 49 50 49 49 49
R2 0.397 0.763 0.822 0.823 0.290 0.833 0.892 0.896
Note: ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001.
All continuous variables are standardized with a mean of 0 and a standard deviation of 1. Standard errors are in parentheses.

Breast-Buttock preference after controlling for racial composition. Proportion of Hispanic ethnicity predicts
less breast preference (β = −0.32, p < .001). Proportion of Asian origin predicts less breast preference
(β = 0.20, p < .01), but only after controlling for the socioeconomic factor. The socioeconomic factor
predicts greater breast preference (β = 0.32, p < .001). This is the only regression analysis we perform
where a measure of prosperity significantly predicts Breast-Buttock preference.

4 Discussion
We have found intelligence to be associated with Breast-Buttock preference at the aggregate level. This was
consistent with our theory that a preference for breasts is indicative of a slow life history strategy, assuming
the average intelligence of human groups is an indicator for life history strategy. The finding was robust to

290
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Table 5: Regression Models of US Metro Area Breast-Buttock preference

Dependent variable:
Google Breast-Buttock preference
(1) (2) (3)
Cognitive Ability 0.569∗∗∗ −0.060
(0.057) (0.047)
Black Admixture −0.881∗∗∗ −0.685∗∗∗
(0.035) (0.048)
Hispanic Admixture −0.318∗∗∗ −0.206∗∗∗
(0.036) (0.040)
Asian Admixture 0.014 −0.201∗∗
(0.059) (0.065)
Amerindian Admixture 0.056 0.021
(0.039) (0.037)
Other Admixture −0.043 0.091
(0.062) (0.061)
Socioeconomic Factor 0.332∗∗∗
(0.056)
Observations 208 208 208
R2 0.323 0.780 0.815
Note: ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001.
All continuous variables are standardized with a mean of 0 and a
standard deviation of 1. Standard errors are in parentheses.

Figure 9: US Metro Area Cognitive Ability and Google Breast-Buttock preference.

controls for GDP and socioeconomic success. We found a a significant effect size for the socioeconomic
factor only in the US Metro analysis. Given that this results was only found in this analysis, we suggest
economic prosperity is unlikely to cause breast or buttock preference.
We suggested that cognitive ability would be correlated with breast preference, but would not cause.
Instead, intelligence is likely confounded with life history strategy, which we believe may be causing racial

291
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Figure 10: US Metro Area African Percent and Google Breast-Buttock preference.

differences in breast preference. Controlling for variables measuring racial ancestry reduced the effect size of
intelligence, consistent with our theory. Nevertheless, in many of our analyses cognitive ability still predicted
Breast-Buttock preference after controls. This may have been due to the imprecision of our estimates for
ancestry making them imperfect controls. Our work leaves open the possibility that intelligence has a causal
effect on Breast-Buttock preference.
Whether intelligence causes Breast-Buttock preference may be resolved by studying individuals rather
than groups. If brighter individuals do not prefer breasts to buttocks more than their less intelligent
compatriots from the same ancestry, this would further strengthen our suggestion that intelligence does
not cause these preferences. However, if life history theory is valid for individual differences, we might still
expect intelligence to be confounded with life history differences within groups. It may also be possible to
test whether individual differences in life history causes variation in Breast-Buttock preference. However, it
is debated whether life history theory applies meaningfully within groups (Giudice, 2020; of Menie et al.,
2021; Zietsch & Sidari, 2020).
According to Rushton’s Rushton (1995) theory that East Asians have a slower life history than
Europeans and Black Africans have a faster life history, we predicted that Blacks would be less interested
in breasts and East Asians would have greater interest in breasts. Only the former prediction held true.
In most analyses, East Asian ancestry had no significant association with Breast-Buttock preference. In
fact, in our analysis of US Metro areas, when we controlled for the socioeconomic factor, Asian ancestry
predicted lower breast preference.
African admixture was a strong predictor of lower Breast-Buttock preference, with African ancestry
explaining 83% of the variance among US States in Breast-Buttock preference. This corroborates prior
evidence that Africans are particularly interested in the buttocks relative to the breasts. Francis and
Kirkegaard Francis & Kirkegaard (2023) surveyed sex workers, who reported that Blacks preferred buttocks
to breasts (d = 2.38, p < .001) and the doggy-stlye to missionary position (d = 1.42, p < .001), relative to
Whites.
With regards to cultural stereotypes, African American culture includes the dance of the “twerk”
characterised by making prominent the buttocks through thrusting (Sauphie, 2021). This is not unique to
Africans in America; as Rushton and Bogaert Rushton & Bogaert (1987) have noted ,“in Africa, dances

292
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

have been invented which emphasize undulating rhythms and mock copulation.” Similar preoccupations
with the buttocks can be found in African-American pop music such as the the lyrics ”I like big butts and I
cannot lie” in Sir Mix Alot’s hit song Baby got Back.
Other racial ancestries also showed assocations with greater preference for breasts over buttocks.
In our national level regressions, South-East Asian ancestry had a raw effect similar in magnitude to the
effect of Sub-Saharan admixture, in the opposite direction. India had the highest score on the Google
Breast-Buttock preference measure despite its presumed average IQ of only 76. Under our theory, the result
could be interpreted as South Asians being particularly slow in their life history strategy relative to their
intelligence. We do note that in the Indian princely state of Travancore, it was apparently illegal for women
of the Dalit untouchable caste to cover their breasts in public (Gupta, 2017).
One issue for the interpretation of our results is the meaning of our Breast-Buttock preference
measures, derived from the relative frequency of searches for breasts or buttocks on Google Trend and
pornographic websites. It is theoretically possible that these measures of internet searches are capturing
something other than just Breast-Buttock preference. We might speculate that the representativeness
of masturbators, or the types of search terms they use on these websites, differs systematically with the
intelligence of a country. Our Breast-Buttock measures derived from Google Trends data and from porn
sites correlated highly (r = .92), indicating excellent convergent validity. This would imply that any bias
in our measures must systematically effect both the data derived from pornographic websites as well as
Google Trends.
To deal with concerns that linguistic differences between countries could cause average cognitive
ability to correlate with Breast-Buttock preference, we also looked at regions within countries and used
Spanish search terms within Spanish countries. The consistency of the correlation between intelligence and
breast preference across languages and within countries, suggests our results are not caused by linguistic
differences between groups.
One potential issue is that our measures of Breast-Buttock preference do not isolate interest in the
breasts, but rather analyse it relative to interest in the buttocks. An advantage of our approach, considering
preference as relative, is that it prevents our measure from capturing general interest in secondary sexual
characteristics or even just the propensity to watch pornography online. In practice this approach was
unavoidable because Google Trends does not give raw absolute search counts and the data from Pornhub
and YouPorn also appear to be relative frequencies. Follow-up research at the individual level could attempt
to isolate interest in breasts, relative to sexual interest generally.
Whilst we found intelligence and race differences in breast preference, the cause is uncertain. We
suggested evolved differences in life history strategy were responsible for intelligence and race differences,
however alternative genetic or environmental explanations could be proposed. Although our theory is
speculative, the striking relationship between intelligence, race and breast preference demands explanation,
which we encourage others to explore. We hope that the strong correlates of breast preference may provide
clues to the function of permanent breasts.

Supplementary Materials
All materials used in this study can be found at: https://osf.io/dvqyc/

References
Alexander, R., & Noonan, K. M. (1979). Concealement of ovulation, paternal care, and human social
evolution.

Ashley B S, A., B S Eva, B., & McGuire, K. P. (2020). Anatomy and physiology of the breast during pregnancy.
Disease of the Breast during Pregnancy and Lactation. Retrieved from http://www.springer.com/
series/5584

293
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Aughinbaugh, A., Robles, O., & Sun, H. (2013, 10). Marriage and divorce: patterns by gender, race, and
educational attainment. Monthly Labor Review . doi: 10.21916/mlr.2013.32

Becker, D., & Rindermann, H. (2016, 8). The relationship between cross-national genetic distances and
iq-differences. Personality and Individual Differences, 98 , 300-310. doi: 10.1016/j.paid.2016.03.050

Benshoof, L., & Thornhill, R. (1979). The evolution of monogamy and concealed ovulation in humans
(Vol. 2).

Carl, N. (2022, 7). How useful are national iqs? Retrieved from https://noahcarl.substack.com/p/
how-useful-are-national-iqs

DeSilva, J. M. (2011, 1). A shift toward birthing relatively large infants early in human evolution.
Proceedings of the National Academy of Sciences of the United States of America, 108 , 1022-1027. doi:
10.1073/pnas.1003865108

Feenstra, R. C., Inklaar, R., & Timmer, M. P. (2015, 10). The next generation of the penn world table.
American Economic Review , 105 , 3150-3182. doi: 10.1257/aer.20130954

Fletcher, G. J., Simpson, J. A., Campbell, L., & Overall, N. C. (2015, 1). Pair-bonding, romantic love, and
evolution: The curious case of homo sapiens. Perspectives on Psychological Science, 10 , 20-36. doi:
10.1177/1745691614561683

Francis, G., & Kirkegaard, E. O. W. (2022). National intelligence and economic growth: A bayesian update.
Mankind Quarterly , 63 , 9-78. doi: 10.46469/mq.2022.63.1.2

Francis, G., & Kirkegaard, E. O. W. (2023). Racial differences in sexuality: A sex worker survey. Mankind
Quarterly , 63 , 675-698. Retrieved from https://mankindquarterly.org/papers/63/4/7 doi:
10.46469/mq.2023.63.4.7

Freedman, D. G., & Deboer, M. M. (1979). Biological and cultural differences in early child development.
Retrieved from www.annualreviews.org

Fuerst, J., & Kirkegaard, E. O. W. (2016). Admixture in the americas: Regional and national differences.
Mankind Quarterly . doi: 10.46469/mq.2016.56.3.2

Giudice, M. D. (2020, 11). Rethinking the fast-slow continuum of individual differences. Evolution and
Human Behavior , 41 , 536-549. doi: 10.1016/j.evolhumbehav.2020.05.004

Gupta, V. (2017). Breast tax: Social oppression of dalit women. Contemporary Social Science, 26 .

Hutchings, J. A. (2021). A primer of life histories: Ecology, evolution, and application.

Isler, K., & van Schaik, C. P. (2006, 12). Metabolic costs of brain size evolution. Biology Letters, 2 ,
557-560. doi: 10.1098/rsbl.2006.0538

Kirkegaard, E. O. W. (2014, 9). The international general socioeconomic factor: Factor analyzing
international rankings. Open Differential Psychology . doi: 10.26775/odp.2014.09.08

Lee, J. J., Mcgue, M., Iacono, W. G., Michael, A. M., & Chabris, C. F. (2019). The causal influence of brain
size on human intelligence : Evidence from within-family phenotypic associations and gwas modeling.
Intelligence, 75 , 48-58. Retrieved from https://doi.org/10.1016/j.intell.2019.01.011 doi:
10.1016/j.intell.2019.01.011

Marlowe, F. (1998). The nubility hypothesis the human breast as an honest signal of residual reproductive
value (Vol. 9).

294
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

McLaughlin, D., de Mello, V. B., Blankenship, C., Chaney, K., Esra, P., Hikawa, H., . . . Rahman, T.
(2003). Comparison between naep and state reading assessment results: 2003, volume 1, research and
development report.

Morris, D. (1967). The naked ape. Jonathan Cape Ltd.

of Menie, M. A. W., Luoto, S., Peñaherrera-Aguirre, M., & Sarraf, M. A. (2021, 4). Life history is a major
source of adaptive individual and species differences: a critical commentary on zietsch and sidari (2020).
Evolutionary Psychological Science, 7 , 213-231. doi: 10.1007/s40806-021-00280-2

Pawłowski, B., & Żelaźniewicz, A. (2021, 12). The evolution of perennially enlarged breasts in women: a
critical review and a novel hypothesis. Biological Reviews, 96 , 2794-2809. doi: 10.1111/brv.12778

Pesta, B. J., Fuerst, J. G. R., & Kirkegaard, E. (2021, 2). County-level usa: No robust relationship
between geoclimatic variables and cognitive ability. Journal of Geographical Research, 4 , 48-66. doi:
10.30564/jgr.v4i1.2765

Pornhub. (2014). Boobs, butts & beyond. Retrieved from https://www.pornhub.com/insights/

boobs-butts-legs-feet-searches

Putterman, L., & Weil, D. N. (2010, 11). Post-1500 population flows and the long-run determinants of
economic growth and inequality. The Quarterly Journal of Economics, 125 , 1627-1682. doi: 10.1162/
qjec.2010.125.4.1627

Rindermann, H. (2007, 8). The g-factor of international cognitive ability comparisons: The homogeneity of
results in pisa, timss, pirls and iq-tests across nations. European Journal of Personality , 21 , 667-706. doi:
10.1002/per.634

Rindermann, H. (2018). Cognitive capitalism: Human capital and the wellbeing of nations. Cambridge
University Press.

Roellig, K., Menzies, B. R., Hildebrandt, T. B., & Goeritz, F. (2011, 2). The concept of superfetation: a
critical review on a ‘myth’ in mammalian reproduction. Biological Reviews, 86 , 77-95. doi: 10.1111/
j.1469-185X.2010.00135.x

Rolfe, D. F., & Brown, G. C. (1997). Cellular energy utilization and molecular origin of standard metabolic
rate in mammals. Physiological Reviews, 77 , 731-758. doi: 10.1152/physrev.1997.77.3.731

Rushton, J. P. (1995). Race, evolution, and behavior : a life history perspective. Transaction Publishers.

Rushton, J. P. (2004, 8). Placing intelligence into an evolutionary framework or how fits into the ? matrix of
life-history traits including longevity. Intelligence, 32 , 321-328. doi: 10.1016/j.intell.2004.06.003

Rushton, J. P. (2010). Brain size as an explanation of national differences in iq, longevity, and other life-history
variables. Personality and Individual Differences, 48 , 97-99. doi: 10.1016/j.paid.2009.07.029

Rushton, J. P., & Bogaert, A. F. (1987, 12). Race differences in sexual behavior: Testing an evolutionary
hypothesis. Journal of Research in Personality , 21 , 529-551. doi: 10.1016/0092-6566(87)90038-9

Sauphie, E. (2021). African origins: From new orleans to abidjan, the roots of twerking. Re-
trieved from https://www.theafricareport.com/144303/african-origins-from-new-orleans
-to-abidjan-the-roots-of-twerking/

Schneider, J. (2020). Google trends metro area gis shape file.

Shamosh, N. A., & Gray, J. R. (2008, 7). Delay discounting and intelligence: A meta-analysis. Intelligence, 36 ,
289-305. Retrieved from https://linkinghub.elsevier.com/retrieve/pii/S0160289607001092
doi: 10.1016/j.intell.2007.09.004

295
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Smith, B. H. (1989). Dental development as a measure of life history in primates. Evolution, 43 , 683-688.
doi: 10.1111/j.1558-5646.1989.tb04266.x

Sprecher, S., Treger, S., & Sakaluk, J. K. (2013, 11). Premarital sexual standards and sociosexuality:
Gender, ethnicity, and cohort differences. Archives of Sexual Behavior , 42 , 1395-1405. doi: 10.1007/
s10508-013-0145-6

Stevens, J. R. (2014, 7). Evolutionary pressures on primate intertemporal choice. Proceedings of the Royal
Society B: Biological Sciences, 281 , 20140499. doi: 10.1098/rspb.2014.0499

United Nations. (2023). Standard country or area codes for statistical use (m49). Retrieved from
https://unstats.un.org/unsd/methodology/m49/overview/

Wang, W. (2018). Who cheats more? the demographics of infidelity in america.

https://ifstudies.org/blog/who-cheats-more-the-demographics-of-cheating-in-america.

Warne, R. T. (2022). National mean iq estimates: Validity, data quality, and recommendations. Evolutionary
Psychological Science. doi: 10.1007/s40806-022-00351-y

Williams, G. C. (1957). Pleiotropy, natural selection, and the evolution of senescence.

Zietsch, B. P., & Sidari, M. J. (2020, 11). A critique of life history approaches to human trait covariation.
Evolution and Human Behavior , 41 , 527-535. doi: 10.1016/j.evolhumbehav.2019.05.007

296
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Appendix

Figure A1: National IQ and Google Breast-Buttock preference.

Figure A2: Sub-Saharan African Ancestry and Google Breast-Buttock preference

297
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Table A1: Regression Models of National Breast-Buttock preference. Admixture not standardised.

Dependent variable:
Google Breast-Buttock preference Pornhub Breast-Buttock preference
(1) (2) (3) (4) (5) (6)
National IQ 0.724∗∗∗ 0.500∗∗∗ 0.491∗∗∗ 0.802∗∗∗ 0.377∗∗∗ 0.352∗∗∗
(0.054) (0.098) (0.126) (0.046) (0.071) (0.090)
Sub-Saharan African Admixture −0.701∗∗ −0.756∗∗ −1.130∗∗∗ −1.160∗∗∗
(0.235) (0.242) (0.171) (0.176)
North-East Asian Admixture −0.102 −0.106 0.292 0.301
(0.345) (0.350) (0.254) (0.257)
South-East Asian Admixture 0.872∗∗ 0.844∗∗ 1.040∗∗∗ 1.030∗∗∗
(0.273) (0.280) (0.201) (0.206)
Middle East and North African Admixture −0.168 −0.180 −0.319∗ −0.328∗
(0.199) (0.215) (0.146) (0.157)
Log GDPpc −0.006 0.016
(0.098) (0.068)
Constant 0.003 0.198 0.224∗ −0.004 0.318∗∗∗ 0.339∗∗∗
(0.053) (0.101) (0.104) (0.047) (0.074) (0.076)
Observations 184 152 147 193 155 151
R2 0.498 0.591 0.596 0.609 0.770 0.771
Note: ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001. Admixture variables are not standardized. All other variables are standardized with a mean
of 0 and a standard deviation of 1. Standard errors are in parentheses.

Table A2: Regression Models of Regional Breast-Buttock preference in the Americas. Admixture not Standardised.

Dependent variable:
Google Breast-Buttock preference
(1) (2) (3)
Regional IQ 0.800∗∗∗ 0.656∗∗∗ 0.597∗∗∗
(0.092) (0.108) (0.132)
African Admixture −2.810∗∗∗ −2.810∗∗∗
(0.662) (0.663)
Native American Admixture 0.513 0.594
(0.523) (0.534)
Socioeconomic Factor 0.041
(0.053)
Brazil −0.401∗∗ −0.028 −0.093
(0.127) (0.183) (0.202)
Chile −0.233 −0.434 −0.487
(0.128) (0.265) (0.274)
Colombia 0.114 0.164 0.061
(0.137) (0.261) (0.294)
Mexico 0.274∗∗ 0.105 0.044
(0.089) (0.309) (0.319)
United States 0.054 0.533∗ 0.599∗∗
(0.128) (0.210) (0.227)
Observations 145 144 144
R2 0.777 0.812 0.813
Note: ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001. Admixture variables are not stan-
dardized. All other variables are standardized with a mean of 0 and a
standard deviation of 1. Standard errors are in parentheses.

298
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Table A3: Regression Models of Tetas-Culos Preference. Admixture not standardised.

Dependent variable:
Google Tetas-Culos Preference
(1) (2) (3)
Regional IQ 0.573∗∗∗ 0.138 0.267
(0.129) (0.139) (0.168)
African Admixture −9.230∗∗∗ −8.800∗∗∗
(1.850) (1.860)
Native American Admixture −2.900∗∗ −3.250∗∗
(1.020) (1.050)
Socioeconomic Factor −0.163
(0.123)
Chile −0.423 1.360∗ 1.460∗
(0.231) (0.557) (0.557)
Colombia 0.279 2.630∗∗∗ 2.840∗∗∗
(0.211) (0.577) (0.596)
Mexico −0.002 2.350∗∗∗ 2.460∗∗∗
(0.163) (0.662) (0.663)
Observations 70 70 70
R2 0.266 0.482 0.496
Note: ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001. Admixture variables are not
standardized. All other variables are standardized with a mean of 0 and a
standard deviation of 1. Standard errors are in parentheses.

Figure A3: Comparsion of Breast-Buttock preference Measures in US States

299
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Figure A4: State IQ and Google Breast-Buttock preference in US States

Figure A5: Black Admixture and Google Breast-Buttock preference in US States

300
Mankind Quarterly, 64(2), 279-301 2023 Winter Edition

Table A4: Regression Models of US State Breast-Buttock preference. Admixture not standardised.

Dependent variable:
Google Breast-Buttock preference Pornhub Breast-Buttock preference
(1) (2) (3) (4) (5) (6) (7) (8)
State IQ 0.625∗∗∗ 0.223∗ 0.276 0.515∗∗∗ 0.033 0.124
(0.111) (0.087) (0.137) (0.116) (0.065) (0.101)
African
−11.207∗∗∗ −10.075∗∗∗ −9.976∗∗∗ −11.221∗∗∗ −11.743∗∗∗ −11.569∗∗∗
Admixture
(0.912) (1.077) (1.104) (0.734) (0.802) (1.369)
Native
American −1.923 −1.631 −5.153∗∗∗ −4.648∗∗
Admixture
(1.754) (1.862) (1.301) (1.369)
Socioeconomic
−0.055 −0.095
Factor
(0.110) (0.081)
Observations 50 49 49 49 50 49 49 49
R2 0.397 0.763 0.822 0.823 0.290 0.833 0.892 0.896
Note: ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001. Admixture variables are not standardized. All other variables are standardized with a mean of 0
and a standard deviation of 1. Standard errors are in parentheses.

Table A5: Regression Models of US Metro Area Breast-Buttock preference. Admixture not standardised.

Dependent variable:
Google Breast-Buttock preference
(1) (2) (3)
Cognitive Ability 0.569∗∗∗ −0.060
(0.057) (0.047)
Black Admixture −7.660∗∗∗ −5.960∗∗∗
(0.303) (0.415)
Hispanic Admixture −2.060∗∗∗ −1.330∗∗∗
(0.230) (0.258)
Asian Admixture 0.401 −5.930∗∗
(1.750) (1.930)
Native American Admixture 2.110 0.806
(1.470) (1.400)
Other Admixture −2.040 4.370
(2.980) (2.930)
Socioeconomic Factor 0.332∗∗∗
(0.056)
Constant 0.000 1.030∗∗∗ 0.796∗∗∗
(0.057) (0.070) (0.081)
Observations 208 208 208
R2 0.323 0.780 0.815
Note: ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001. Admixture variables are not
standardized. All other variables are standardized with a mean of 0 and a
standard deviation of 1. Standard errors are in parentheses.

301