You are on page 1of 12

Intelligence 81 (2020) 101465

Contents lists available at ScienceDirect

Intelligence
journal homepage: www.elsevier.com/locate/intell

The costs of being exceptionally intelligent: Compatibility and interpersonal T


skill concerns☆
Gilles E. Gignac , Zoe M.V. Callis

School of Psychology, University of Western Australia, 35 Stirling Highway, Crawley, Western Australia 6009, Australia

ABSTRACT

People tend to rate exceptional levels of IQ (99th percentile) as less attractive than high levels of IQ (90th percentile), and it remains to be determined why.
Furthermore, the desirability of emotional intelligence (EI) in a prospective partner has yet to be investigated. Finally, we sought to determine whether individual
differences in self-assessed and objectively measured IQ/EI correlated with desirability ratings of IQ/EI in a prospective partner. Based on a general community
sample (N = 236) and an undergraduate sample (N = 220), we found that the association between rated desirability and the IQ/EI level of a prospective partner
exhibited a threshold effect at the 90th IQ/EI percentile. Furthermore, a statistically significant decrease in rated desirability between the 90th to the 99th percentiles
was observed for IQ, but not for EI. We found that participants who reduced their ratings of desirability between the 90th and 99th IQ percentiles did so due to
compatibility concerns (≈60%) and social skill concerns (≈40%). We also found that self-assessed IQ and objectively measured IQ correlated positively with
desirability ratings at the 90th IQ percentile, and self-assessed EI (but not objectively measured EI) with desirability ratings at the 90th EI percentile. Finally, we
found that, on average, people ranked/rated EI to be somewhat more desirable than IQ. We interpreted the results as consistent with compatibility theory, active
assortative mating for intelligence, and the possibility that many people subscribe to the stereotype that exceptionally intelligent people suffer from interpersonal
skill difficulties.

1. Introduction levels similar to their own (Śmieja & Stolarski, 2018). Consequently, the
second purpose of this investigation was to determine the ideal level of
Cognitive intelligence (IQ) is one of the most attractive traits in a EI in a prospective romantic partner. Furthermore, on the basis of the
potential romantic partner (Buss et al., 1990; Gignac, Darbyshire, & assortative mating for intelligence literature (Bouchard & McGue,
Ooi, 2018; Goodwin & Tinker, 2002). From a strict biological per- 1981), we hypothesized that individual differences in raters' level of IQ
spective, the desirability of a prospective partner should increase line- and EI would correlate positively with their desirability ratings of a
arly with the prospective partner's level of IQ, as intelligence is asso- prospective partner's level of IQ and EI, respectively. Finally, we sought
ciated with incrementally beneficial biological characteristics to determine which of the intelligences people rate as more attractive in
(Hagenaars et al., 2016). However, exceptional levels of IQ a prospective partner: IQ or EI?
(IQ = 140+; 99th percentile) have been found to be rated, on average,
less attractive than high levels of IQ (i.e., IQ = 120; 90th percentile; 1.1. Partner desirability and cognitive intelligence
Gignac et al., 2018; Gignac & Starbuck, 2019). Gignac et al. (2018)
suggested two possibilities for why people, on average, rate exceptional At a high level of abstraction, cognitive intelligence represents an
intelligence as less attractive: (1) compatibility concerns; and (2) con- entity's capacity to adapt to the environment successfully using cogni-
cerns that highly intelligent people are more likely to have inter- tive processes (McIntosh, Dixon, & Pierson, 2005; Pintner, 1923). In
personal skill deficits. The first purpose of this investigation was to practice, among both academics and lay people alike, the IQ of humans
investigate these two possibilities. is commonly considered to vary across dimensions such as logical
In contrast to IQ, there is relatively little research on the attrac- reasoning, knowledge of worldly facts, memory span and processing
tiveness of emotional intelligence (EI) in a prospective romantic speed, for example (Furnham, 2001; Gignac, 2018; Sternberg, Conway,
partner. Some research suggests that higher levels of EI may be rated Ketron, & Bernstein, 1981). Furthermore, IQ is a well-established po-
progressively attractive (Schutte et al., 2001). However, there is also sitive correlate of socially attractive characteristics such as educational
some evidence to suggest that people may prefer individuals with EI attainment, job status, and earnings (Gottfredson, 2002; Zagorsky,


Correspondence should be addressed to Gilles E. Gignac, School of Psychology, University of Western Australia, 35 Stirling Highway, Crawley, Western Australia,
6009, Australia. E-mail: gilles.gignac@uwa.edu.au; thanks to Travis Dunkley and Brooke Shepherdson for help with data collection.

Corresponding author.
E-mail address: gilles.gignac@uwa.edu.au (G.E. Gignac).

https://doi.org/10.1016/j.intell.2020.101465
Received 14 January 2020; Received in revised form 5 April 2020; Accepted 27 May 2020
0160-2896/ © 2020 Elsevier Inc. All rights reserved.
G.E. Gignac and Z.M.V. Callis Intelligence 81 (2020) 101465

2007), characteristics that tend to be considered attractive in a long- Stumm, 2014).


term romantic partner (Buss & Schmidt, 1993; Buunk, Dijkstra, Secondly, Gignac et al. (2018) suggested that some raters may have
Fetchenhauer, & Kenrick, 2002). Furthermore, the socio-economic had concerns that an exceptionally intelligent person would lack social
benefits of IQ (e.g., earnings) have been reported to manifest them- skills, as there is research to suggest that a non-negligible percentage of
selves progressively, even at the highest levels of IQ (Lubinski & people believe that exceptionally intelligent people have interpersonal
Benbow, 2006). difficulties. For example, Baudson (2016) found that two thirds of
Correspondingly, IQ has been found to be ranked the second or third people in the general community held the view that intellectually gifted
most desirable characteristic in a prospective partner, behind only ‘kind people tend to have greater difficulties with social situations, in com-
and understanding’ and (sometimes) ‘exciting personality’ (Buss et al., parison to people with average IQ (see also Rodríguez-Nieto, Sánchez-
1990; Buss & Barnes, 1986; Goodwin & Tinker, 2002). Beyond the so- González, & Sánchez-Miranda, 2019). Therefore, it is possible that the
cioeconomic benefits, IQ has been argued to be an attractive partner mean reduction in rated desirability for a prospective partner from the
characteristic, as it is associated positively with biological fitness and 90th to the 99th IQ percentiles was observed because a non-negligible
better health (Der, Batty, & Deary, 2009; Lubinski & Benbow, 2006; number of respondents in Gignac et al. (2018) and Gignac and Starbuck
Prokosch, Yeo, & Miller, 2005). Additionally, IQ is known to be sub- (2019) made a perceived trade-off of IQ points for social skills. The two
stantially heritable (Plomin & von Stumm, 2018), and at least some of hypotheses suggested by Gignac et al. (2018), compatibility and social
the identified genes appear to influence the whole spectrum of the IQ skill concerns, could be tested by asking respondents who reduce their
continuum (Zabeneh et al., 2018). Furthermore, large scale genetics ratings of desirability from the 90th to the 99th percentile why they did
research has found that the genes associated with IQ predispose a so.
person to biological benefits in an incremental manner (e.g., Hagenaars
et al., 2016). On the basis of such evidence, Gignac et al. (2018) argued 1.3. Emotional intelligence as an attractive partner trait
that the association between rated attractiveness and the IQ of a pro-
spective partner may be hypothesized to be linear, as the biological and At a high level of abstraction, emotional intelligence may be defined
social benefits of IQ have been found to be largely consistent across the as the “…ability to purposively adapt, shape and select environments
full spectrum of the dimension. Gignac et al. (2018) labelled such an through the use of emotionally relevant processes” (Gignac, 2010a, p.
expectation the ‘biological hypothesis’. 131). Ability-based (maximum performance with tests) and trait-based
However, empirical research on the rated desirability of prospective (combination of self-reported abilities, traits, self-esteem, and well-
partners and their level of IQ has failed to support the biological hy- being) measures are two common approaches to the measurement of EI
pothesis. For example, based on a mixed sample (general community (Petrides, 2011). Both approaches typically measure several positively
and university students; N = 383), Gignac et al. (2018) found that the interrelated dimensions, including perceiving, using, understanding
association between rated desirability of a prospective partner and the and managing emotions (Gignac, 2010b; Petrides, 2011; Salovey &
level of IQ of a hypothetical prospective partner was found to be cubic Mayer, 1990). EI has been shown to predict job performance, in-
(i.e., two bends in the pattern of mean ratings). Specifically, desirability dependently of the effects of IQ (Joseph, Jin, Newman, & O'Boyle, 2015;
ratings increased moderately between the 1st and the 25th IQ percen- O'Boyle Jr, Humphrey, Pollack, Hawver, & Story, 2011), suggesting that
tiles, and then substantially from the 25th to the 75th IQ percentiles. EI may be regarded as a socially attractive characteristic in a person.
Importantly, after a slight increase between the 75th and 90th IQ per- Additionally, EI has been found to correlate positively with several
centiles, a statistically significant decrease in rated desirability was indicators of romantic relationship success (Apostolou, Paphiti, Neza,
observed between the 90th and 99th IQ percentiles. The same pattern of Damianou, & Georgiadou, 2019; Schröder-Abé & Schütz, 2011; Zarch,
results was replicated in Gignac and Starbuck (2019), based on a Marashi, & Raji, 2014). Consequently, given that EI is a dimension
sample of first year university students (N = 214). Thus, the results of associated with socioeconomic and inter-personal benefits, it would be
both Gignac et al. (2018) and Gignac and Starbuck (2019) were not expected that, on average, people would rate EI as a desirable trait in a
consistent with the biological hypothesis. Instead, the results were in- prospective partner.
terpreted to be consistent with the ‘threshold hypothesis’, i.e., the To date, little research has evaluated specifically whether people
benefits of IQ diminish or cease after a certain point (Torrance, 1962). consider EI to be a desirable trait in a romantic partner. For example,
Thus, one or more other factors likely impacted people's views on the the publication of the Buss and Barnes' (1986) Partner Preference Scale
desirability of IQ in a prospective partner. predates the academic publication of the construct of EI (Salovey &
Mayer, 1990) and EI has not been included in any adaptation. In per-
1.2. Why is there a threshold effect? haps the only relevant study, Schutte et al. (2001) found that university
students predicted they would find more satisfying a relationship with
An important limitation of Gignac et al. (2018) and Gignac and someone who was adept at managing their own and others' emotions, in
Starbuck (2019) is that they did not investigate why, on average, people comparison to someone without those abilities.
rated IQ at the 99th percentile as less attractive than the 90th percen- Although the Schutte et al. (2001) study had strengths, it should
tile. However, Gignac et al. (2018) did suggest two possible psychoso- also be acknowledged that, strictly speaking, Schutte et al. (2001) did
cial factors that may have impacted the rated desirability of an ex- not measure the desirability or attractiveness of EI in a prospective
ceptionally intelligent prospective partner: (1) compatibility concerns; partner, per se. Instead, the link between EI and anticipated relation-
and (2) interpersonal skill deficit concerns. ship satisfaction only implies that higher levels of EI are attractive in a
Firstly, Gignac et al. (2018) suggested that exceptional levels of IQ prospective partner. Additionally, Schutte et al. (2001) did not design
may be, on average, rated as less attractive, because raters may be their study to evaluate whether there might be a threshold effect for EI,
concerned that they would be incompatible with someone exceptionally as reported for IQ by Gignac et al. (2018) and Gignac and Starbuck
intelligent. Correspondingly, there is evidence that people consider (2019). Finally, Schutte et al.'s (2001) study did not compare EI with
intellectual compatibility, when evaluating the compatibility of a pro- other known desirable traits, consequently, the degree to which EI may
spective partner (e.g., Jonason et al., 2019; Wilson & Cousins, 2003). be desirable in a partner could not be contextualised with comparisons
Thus, on the basis of compatibility concerns, it is possible that, on to other desirable traits.
average, people rate a person with an IQ at the 99th percentile as less As described above, people may prefer partners who have relatively
attractive than the 90th percentile, because most people self-report higher levels of EI, because of the corresponding interpersonal and
their own IQ to be somewhere between 110 and 120, i.e., less than the socioeconomic benefits (Joseph et al., 2015; Schröder-Abé & Schütz,
99th percentile (Furnham, 2001; Gignac & Zajenkowski, 2019; von 2011). However, as per IQ, it is possible that the association between

2
G.E. Gignac and Z.M.V. Callis Intelligence 81 (2020) 101465

partner attractiveness ratings and partner EI may be consistent with a questionnaire in which the response scale was a typical 5-point Likert
threshold effect, on the basis of compatibility theory (Huston & Houts, scale (e.g., ‘I know a lot of worldly facts,’ and ‘I am poor at solving
1998). That is, as most people rate their EI at substantially less than the logical problems, from ‘strongly disagree’ to ‘strongly agree’). However,
99th percentile (Petrides & Furnham, 2000), exceptional levels of EI composite scores from such a scale may not correspond closely to
may not be perceived to be the most desirable level in a prospective people's perceptions of their own IQ relative to others, which is argu-
partner, on average, in a manner similar to that observed for IQ ably more relevant to the rated desirability of prospective partners
(Gignac, 2019; Gignac et al., 2018). We also note that the degree of distinguished by their specified level of IQ, if assortative mating/com-
friendship among friends has been found to correlate positively with the patibility theory plays a role in the mate selection process. Another
degree of similarity in ability to recognise emotions in faces (Brauer & approach would be to request participants to specify their IQ along a
DePaulo, 1980), a dimension of EI (Schütz & Koydemir, 2018). Such a percentile continuum, an established measurement approach in the
result suggests that people may seek out relationships with people with field (e.g., Furnham, Moutafi, & Chamorro-Premuzic, 2005; Swami,
whom they share similar levels of EI abilities. Furnham, & Zilkha, 2009; von Stumm, 2014). Such a measurement
In summary, although a substantial amount of research has shown approach may help increase the chances of observing a positive asso-
‘intelligent’ to be ranked the second or third most desirable character- ciation between self-rated IQ and ratings of attraction for a hypothetical
istic in a romantic partner (Buss et al., 1990; Gignac et al., 2018; romantic partner with a specified level of IQ. Although the evidence for
Goodwin & Tinker, 2002), no research has compared ‘cognitively in- assortative mating for EI is inconsistent (Schröder-Abé & Schütz, 2011;
telligent’ and ‘emotionally intelligent’ within the context of the Buss/ Śmieja & Stolarski, 2018; Zeidner & Kaluda, 2008), we also explored
Barnes Partner Preference scale. Additionally, although an IQ of 120 the possibility that self-assessed EI measured on a percentile continuum
has been shown to be the most attractive level, on average, in a pro- would relate positively with the ratings of attraction for a hypothetical
spective partner on the basis of the measurement approach employed romantic partner with a specified level of EI.
by Gignac et al. (2018); Gignac (2019), whether EI is associated with a
similar threshold effect remains to be determined. 1.5. Summary and purpose

1.4. Predictors of IQ/EI rated desirability The rated desirability of a prospective partner has been shown to
increase in a nonlinear manner across levels of specified IQ in a pro-
Although Gignac et al. (2018) and Gignac and Starbuck (2019) re- spective partner, such that a prospective partner with exceptional levels
ported mean differences of rated desirability across the IQ percentiles, of IQ (140+) is viewed as, on average, somewhat less attractive than
Gignac et al. (2018) and Gignac and Starbuck (2019) also reported non- high levels of IQ (i.e., 120; Gignac et al., 2018; Gignac & Starbuck,
negligible individual differences in the desirability ratings of the pro- 2019). However, it remains to be determined why the mean reduction
spective partners across the IQ percentiles. Correspondingly, not ev- in rated desirability occurs. On the basis of the suggestions made by
eryone reduced their ratings of desirability from the 90th to the 99th Gignac et al. (2018), we hypothesized that two factors, (1) compat-
percentiles, implying that some people were attracted to exceptional ibility concerns, and/or (2) the stereotype that intellectually gifted
levels of IQ, and even more so than high levels of IQ. What may in- people struggle with interpersonal skills, may play a role.
fluence such individual differences? Additionally, although some research suggests that EI may be re-
Gignac et al. (2018) suggested that a positive association may be garded as an attractive ability/trait in a prospective partner (Schutte
observed between the degree to which people rate the attractiveness of et al., 2001), little to no research has evaluated this possibility directly.
a prospective partner with a specified level of IQ and the objectively Consequently, we sought to determine whether the association between
measured IQ of those raters, as there is evidence for assortative mating desirability ratings of prospective partners and their suggested levels of
for IQ (r ≈ 0.30 to 0.40; Gignac & Zajenkowski, 2019; Watson et al., EI exhibited a nonlinear (threshold) effect similar to that reported for
2004; van Leeuwen, van Den Berg, & Boomsma, 2008), and the corre- IQ.
lation is observed early in the relationship, suggesting that mate se- On the basis of compatibility theory (Huston & Houts, 1998), as well
lection is the cause of the spousal IQ correlation (Mascie-Taylor, 1989). as the evidence of assortative mating for IQ (Mascie-Taylor, 1988), we
However, Gignac (2019) failed to find an association between general also hypothesized that individual differences in self-reported in-
intelligence and individual differences in ratings of desirability of a telligence (IQ and EI) would correlate positively with ratings of desir-
prospective partner with IQs specified at the 75th, 90th, and 99th ability for a prospective partner with a relatively high level of IQ. Fi-
percentiles. By contrast, correlations of approximately −0.20 to −0.25 nally, given the relatively little research on the attractiveness of EI as a
were observed for the corresponding below average IQ percentiles. partner characteristic, it was considered useful to add ‘emotionally in-
Thus, Gignac and Starbuck (2019) suggested that relatively intelligent telligent’ as a 14th prospective partner characteristic to Buss and
people may have a tendency to simply find unattractive relatively un- Barnes' (1986) Partner Preference Scale. It was hypothesized that
intelligent people, rather than find attractive relatively intelligent ‘emotionally intelligent’ would be ranked relatively highly, however,
people. However, Gignac and Zajenkowski (2019) suggested the cor- whether it would be ranked more highly than ‘cognitively intelligent’
relations appeared to be small, considering the magnitude of the as- was a possibility we explored.
sortative mating for IQ effect. Consequently, Gignac (2019) re-
commended replication of the effect, prior to drawing firm conclusions. 2. Method
Gignac and Starbuck (2019) also suggested that a person's self-
perceived IQ, rather than objective IQ, may associate more sub- 2.1. Sample
stantially with their attraction ratings for a potential partner with a
specified level of IQ, as people would be expected to be more aware of For sample 1 (‘general community sample’), we recruited 254 par-
how intelligent they think themselves to be, rather than how intelligent ticipants (Asian 5%; Black/African American = 8%; Multiracial = 5%;
they are objectively. However, Gignac and Zajenkowski (2019) failed to White/Caucasian = 81%) from Amazon's Mechanical Turk (MTurk), a
find an association between self-reported IQ (9-item questionnaire) and platform that has been shown to yield reasonably reliable and valid
rated interest in a prospective partner with a specified level of IQ, data (Buhrmester, Kwang, & Gosling, 2011). The only filter applied in
contrary to expectations, given the evidence for compatibility theory the MTurk recruitment was that the respondents needed to be US re-
and assortative mating for IQ. sidents. Eighteen of the 254 MTurk responders (7.1%) were excluded
One arguable limitation associated with Gignac and Starbuck from the final sample, as they were deemed not to have completed the
(2019) is that they measured self-perceptions of IQ with a multi-item questionnaires entirely seriously (i.e., their pattern of responses to one

3
G.E. Gignac and Z.M.V. Callis Intelligence 81 (2020) 101465

or more of the questionnaires was implausible). Excluding these parti- potential partner (e.g. marriage, children), if you later learned that
cipants did not impact the conclusions made in this investigation. The their emotional intelligence level was such that they were more emo-
final MTurk sample consisted of 236 participants (age M = 39.19; tionally intelligent than [1%, 10%, 25%, 50%, 75%, 90%, 99%] of the
SD = 12.27; range = 20 to 82) with approximately equal numbers of population?” For each of the seven percentile points, participants rated
males (56.4%) and females (43.6%). Finally, the MTurk sample had an their level of interest on a six-point scale: extremely uninterested = 1;
educational profile somewhat more educated than the general US po- very uninterested = 2; uninterested = 3; interested = 4; very inter-
pulation (United States Census Bureau, 2019): some high ested = 5; and extremely interested = 6. This measure was included in
school = 0.4%; high school graduate or graduate equivalency = 12.7%; samples 1 and 2.
associate's technical degree = 5.9%; some college = 15.3%; associate's
academic degree (2 years) = 8.5%; bachelor's degree = 44.9%; master's 2.2.4. Extensions to Percentile Partner Interest Scales: IQ and EI
degree = 8.9%; professional school degree = 3.0%; doctoral de- To investigate why some participants did not rate the 99th per-
gree = 0%; prefer not to say = 0.4%. centile as the most attractive, for sample 1, we followed up the
For sample 2, we recruited 239 English as a first language, first year Percentile Partner Interest Scale (IQ) with the following:
undergraduate psychology students (received partial course credit)
“For the question above, if your level of interest was not highest for
from a predominantly white Anglo-Saxon university. Data from twelve
someone smarter than 99% of the population, why was that so?
of the participants were removed, as they were deemed insufficiently
(Select both reasons below, if both reasons apply).
proficient in English (e.g., looked up English words on their phone
A: I'm worried that someone exceptionally intelligent would be
during testing). Additionally, seven participants did not complete the
socially awkward
entire testing protocol for various reasons (e.g., not feeling well, tech-
B: I'm worried that I would not be compatible with someone who is
nical problem). Thus, the final total sample consisted of 220 partici-
exceptionally intelligent.
pants (69.5% female, age M = 20.70, SD = 6.10; range: 17 to 50). As
C: This question does not apply to me. My highest interest was in
we describe below, the key difference between samples 1 and 2 is that
fact for someone smarter than 99% of the population.”
we had scores from self-reported intelligence and objective intelligence
measures for sample 2. This research was approved by The University of The participants could select options A and/or B, or C, or provide no
Western Australia Human Research Ethics Committee (RA/4/1/6704). answer, if neither A nor B explained the reason for their reduced rat-
The data have been made available via the Open Science Framework: ings.
https://osf.io/sn872/ For sample 2, the extension to the Percentile Partner Interest Scale-
IQ was slightly different. Specifically, those who reduced their rated
2.2. Measures interest from the 90th to the 99thpercentile of IQ were asked why they
did so:
2.2.1. Partner Preference Scale
“You've indicated that you would be less interested in someone who
The Buss/Barnes Partner Preference Scale requires participants to
is smarter than 99% of the population, in comparison to someone
rank 13 traits in a prospective partner, from most important to least
who is smarter than 90% of the population. Why did you reduce
important (Buss & Barnes, 1986). The original traits include: kind and
your rating of attraction?”
understanding, exciting personality, intelligent, physically attractive,
healthy, easygoing, creative and artistic, wants children, college grad- The participants were provided with the same options, A and B, as
uate, good earning capacity, good heredity, good housekeeper, and specified above. Additionally, for sample 2, the participants were given
religious (Buss & Barnes, 1986). For the current study, the Partner the option to provide qualitative comments (free text), in order to
Preference Scale was modified. Specifically, ‘intelligent’ was changed to capture responses beyond the multiple-choice options. Furthermore, we
‘cognitively intelligent’. Additionally, ‘emotionally intelligent’ was also made available the option of providing free-text responses to those
added as a 14th trait. This measure was included in samples 1 and 2. who reduced their partner interest ratings between the 90th and the
For sample 1, the presentation order of the traits was randomized across 99th EI percentiles, in order to gain some insights into why some par-
participants. ticipants may have reduced their rating of attractiveness.

2.2.2. Percentile Partner Interest Scale–Cognitive Intelligence (IQ) 2.2.5. Self-Reported IQ and EI
The Percentile Partner Interest Scale measures the degree to which a Self-reported cognitive intelligence (SRIQ) was measured in sample 2
respondent finds a prospective partner attractive on the basis of a with the Self-Report Intelligence Questionnaire (SRIQ; Gignac, Stough,
specified level of a trait, in this case, IQ. As per Gignac et al. (2018), & Loukomitis, 2004). The SRIQ is a 9-item questionnaire whereby each
participants were asked, “Suppose you were single (if you're not) and item is responded to on 5-point Likert scale (strongly disagree to
you met someone single that, at first sight, potentially interested you. strongly agree). Three of the items are keyed negatively. Each item is
How interested would you be in the person as a potential partner (e.g. roughly representative of one of the subtests within the Wechsler Adult
marriage, children), if you later learned that their intelligence level was Intelligence Scales (e.g., ‘I know a lot of worldly facts’; ‘I am poor at
such that they were smarter than [1%, 10%, 25%, 50%, 75%, 90%, solving logical problems’). The ninth item is a general indicator (‘I am
99%] of the population?” For each of the seven percentile points, intelligent’). Internal consistency reliability was estimated at α = 0.70.
participants rated their level of interest on a six-point scale: extremely Self-reported emotional intelligence (SREI) was measured in sample 2 with
uninterested = 1; very uninterested = 2; uninterested = 3; inter- the slight adaptation of the 70-item Genos Emotional Intelligence In-
ested = 4; very interested = 5; and extremely interested = 6. This ventory (Genos EI; Palmer, Stough, Hamer, & Gignac, 2009). Although
measure was included in samples 1 and 2. self-report in nature (7-point Likert scale; 1 = Never, 2 = Almost
Never, 3 = Seldom, 4 = Sometimes, 5 = Usually, 6 = Almost Always,
2.2.3. Percentile Partner Interest Scale–Emotional Intelligence (EI) 7 = Always), Genos EI is arguably better characterized as a self-report
An adaptation of the Percentile Partner Interest Scale was ad- measure of EI, rather than a mixed-model (combination of abilities,
ministered to investigate the degree to which a respondent found a traits, self-esteem, well-being, for example) measure of EI (see Petrides,
prospective partner attractive on the basis of a specified level of EI. 2011), as the items are all closely aligned with the capacity to use EI
Specifically, participants were asked, “Suppose you were single (if skills (Gignac, 2010b). Example items include, ‘I am effective in helping
you're not) and you met someone single that, at first sight, potentially others feel positive,’ and ‘I fail to handle stressful situations effectively’.
interested you. How interested would you be in the person as a Six of the items are keyed negatively. Internal consistency reliability

4
G.E. Gignac and Z.M.V. Callis Intelligence 81 (2020) 101465

was estimated at α = 0.84 for the total scale scores. 2008). The brief STEU consists of 19 vignettes designed to represent a
situation that can be expected to elicit an emotion from the principal
2.2.6. Self-Assessed Intelligence: IQ and EI character in the vignette. The testee must select from five response
For sample 2, self-assessed cognitive intelligence (SAIQ) was measured alternatives the most appropriate emotion for the vignette. Based on the
with a single item designed to measure a person's IQ on a percentile results of several investigations (Austin, 2010; Libbrecht & Lievens,
scale.1 The instructions include a definition of IQ and a description of a 2012; Schlegel, Fontaine, & Scherer, 2017), the STEU may be con-
percentile: sidered a valid indicator of ability-based EI, on the basis of its positive
correlation with the Mayer-Salovey-Caruso Emotional Intelligence Test
“Cognitive intelligence represents the ability to acquire knowledge, (MSCEIT), a social theory of mind ability test, and an emotional re-
reason with the knowledge, and use it to solve logical problems cognition test, for example. However, based on sample 2, we found the
effectively. People differ with respect to their cognitive intelligence internal consistency reliability to be very low α = 0.26 (discussed fur-
and can have a low, an average or a high level. Using the following ther in the results below).
scale, please indicate where you would place yourself compared to
other people. Specifically, slide the scale to the appropriate point 2.3. Procedure
corresponding to your level of cognitive intelligence relative to
others.” For sample 1 (‘general community sample’), all testing was com-
The participant responds to the item with a slider scale that ranges pleted on a computer with the questions hosted on Qualtrics. A
from 1 to 100 (See Fig. S1A). Self-assessed emotional intelligence (SAEI) Qualtrics link was provided through the MTurk platform. After com-
was measured with a single item designed to measure a person's EI on a pleting the informed consent form, the participants completed the Buss/
percentile scale. The instructions include a definition of EI and a de- Barnes Partner Preference Scale. Next, the participants completed the
scription of a percentile: Percentile Partner Interest Scales (IQ and EI; order randomized for
sample 1). Finally, the participants completed the self-assessed IQ and
“Emotional intelligence is the ability to perceive emotions, use the self-assessed EI single-item scales (order randomized for sample 1).
emotions to help with thinking, understand emotional meanings, Total testing time was approximately 7 min.
and manage emotions. People differ with respect to their emotional For sample 2 (‘undergraduate sample’), participants were tested in-
intelligence and can have a low, an average or a high level. Using person by one of three research assistants. After completing the consent
the following scale, please indicate where you would place yourself form, the participants completed the demographic questions, a self-re-
compared to other people. Specifically, slide the scale to the ap- port narcissism scale (different study), the self-reported IQ ques-
propriate point corresponding to your level of emotional in- tionnaire, the Advanced Vocabulary test, the Advance Progressive
telligence relative to others.” Matrices, the Buss/Barnes ranking questionnaire, the Situational Test of
The participant responds to the item with a slider scale that ranges Emotional Understanding, Genos EI, a novel test of emotional re-
from 1 to 100 (See Fig. S1B). cognition (different study), Letter-Number Sequencing, the Connections
test, the Percentile Partner Interests Scales (IQ and EI), and the single-
item self-assessed intelligence scale (IQ and EI). All of the tests were
2.2.7. Task-Based Cognitive Intelligence (TBIQ)
administered on a computer, with the exception of the Letter-Number
For sample 2, task-based IQ was measured across four dimensions of
Sequencing and Connections tests (administered face-to-face). The total
cognitive ability: processing speed, crystallized intelligence, fluid in-
testing time was approximately 55 min.
telligence, and memory span. Processing speed was measured with the
two subtests (Numbers and Numbers/Letters) within the Connections
2.4. Data analysis
battery (Salthouse et al., 2000). These tests are very similar to the well-
known Trails A/Trails B tests (Reitan, 1958). On the basis of the cor-
All of the sample 1 and sample 2 quantitative analyses were per-
relations between the two processing speed tests, the internal con-
formed with SPSS and Amos version 25.0. First, we analysed the
sistency reliability of the overall processing speed composite scores was
Partner Percentile Interest Scales' (IQ and EI) data with a factorial re-
estimated at 0.56. Crystallized intelligence was measured with the Ad-
peated measures ANOVA approach (for samples 1 and 2). Specifically,
vanced Vocabulary Test (AVT; Gignac, Shankaralingam, Walker, &
we tested a 2 × 7 repeated measures factorial ANOVA with intelligence
Kilpatrick, 2016), a 21-item multiple choice test comprised of relatively
type as a main effect (two levels; cognitive and emotional) and an in-
difficult words to define, by selecting the most appropriate of the five
telligence level main effect (seven levels/percentiles). If the data were
alternatives (α = 0.67 this sample). Fluid intelligence was measured with
consistent with a threshold effect, we anticipated a statistically sig-
the odd items from the Advanced Progressive Matrices (APM; Raven,
nificant nonlinear trend in the data (i.e., cubic effects). Furthermore, we
1966). Participants were given 10 min to complete the APM short form
hypothesized a statistically significant reduction in the rated mean
(α = 0.71 this sample). Finally, memory span was measured with a
desirability of IQ from the 90th to the 99th percentiles. We explored
slight adaptation of the Letter–Number Sequencing Test (LNS;
such a possibility with the EI data.
Wechsler, 2008; α = 0.80 this sample). On the basis of a factor analysis
Next, for samples 1 and 2, we estimated the percentages of people
(maximum likelihood) and the extraction of single factor, we obtained
who did not rate the 99th IQ and EI percentiles as the most desirable.
the following standardized g-loadings: APM = 0.72; Connec-
Furthermore, for sample 2, the open-ended responses were analysed
tions = 0.68; LNS = 0.65; and AVT = 0.63.
with a qualitative strategy. In light of the lack of prior empirical re-
search on the reasons underlying the threshold effect of IQ in the
2.2.8. Task-Based Emotional Intelligence (TBEI) context of partner attraction, as well as the lack of research examining
For sample 2, task-based EI was measured with the brief version of EI as a partner trait, grounded theory methodology was drawn upon to
the Situational Test of Emotional Understanding (STEU; Allen, analyse the open-ended responses. The grounded theory approach in-
Weissman, Hellwig, MacCann, & Roberts, 2014; MacCann & Roberts, volves systematic data coding and analysis procedures to derive in-
ductively a theory about a phenomenon (Strauss & Corbin, 1990). See
1
For the purposes of this investigation, we distinguished between self-re- the Supplementary Section for further details and Tables S1 and S2 for
ported intelligence, which was measured with a typical multi-item ques- the coding structures.
tionnaire that includes a Likert response scale, versus self-assessed intelligence, To estimate the associations between the individual difference
which was measured with a single-item and a percentile type scale. variables and rated interest in a prospective partner with a specified

5
G.E. Gignac and Z.M.V. Callis Intelligence 81 (2020) 101465

level of IQ, we performed latent Pearson correlations between the interaction effect was significant for both samples 1 and 2 (see Table 1).
percentile-based self-assessed measures (IQ and EI) and the responses to As can be seen in Fig. 1 (panels A and B), the greatest numerical di-
the Percentile Partner Interest Scales (IQ and EI). On the basis of a vergence between the IQ and EI rated desirability means occurred from
factor analysis with the extraction of two factors (maximum likelihood; the 75th to the 99th intelligence percentiles. A pair of follow-up 2 × 2
promax rotation), we found that the 10th IQ percentile loaded on the factorial within-subjects ANOVAs isolated at the 90th and 99th in-
first IQ factor at 0.99 and 0.97 for samples 1 and 2, respectively; and telligence percentiles identified a statistically significant interaction
the 90th IQ percentile loaded on the second IQ factor at 0.99 and 0.90 across both samples 1 and 2 (see Table 1). Furthermore, for IQ, statis-
and for samples 1 and 2, respectively. Correspondingly, we found that tically significant reductions in rated desirability were observed be-
the 10th EI percentile loaded on the first EI factor at 0.99 and 0.97 for tween the 90th and 99th percentiles across both samples: sample 1, t
samples 1 and 2, respectively; and the 90th EI percentile on the second (235) = 4.51, p < .001, g = 0.18; sample 2, t(219) = 4.31, p < .001,
EI factor at 0.99 and 0.99, respectively. Consequently, for the latent g = 0.21. By contrast, for EI, the corresponding difference was not
correlational analyses, we used the 10th percentile and the 90th per- found to be significant statistically across both samples: sample 1, t
centile observed scores to represent the lower and upper intelligence (235) = −1.85, p = .066, g = −0.07; sample 2, t(219) = −0.98,
percentiles, respectively. The latent variable correlations were esti- p = .328, g = −0.04.
mated via maximum-likelihood estimation.
Finally, for samples 1 and 2, we analysed the Buss/Barnes Partner 3.2. Percentage reducers and reasons: IQ
Preference Scale rank data, in order to evaluate the attractiveness of the
‘cognitively intelligent’ and ‘emotionally intelligent’ mate character- To appreciate further the nature of the effects, we created scatter
istics. As conventional non-parametric statistical techniques often yield (jitter) plots depicting the association between the desirability ratings
less insightful information than parametric analyses (Cooper & Weekes, associated with 90th and 99th IQ/EI percentiles (sample 1). As can be
1983), we employed a least squares regression-based paired-samples t- seen in Fig. 2, a substantial number of participants did not change their
test established by Hedberg and Ayers (2015), in order to test the dif- ratings from the 90th to the 99th percentiles (observations along the
ference between mean mate characteristic ranks associated with the line of best fit). Furthermore, some participants increased their desir-
Buss/Barnes Partner Preference Scale. The procedure involves calcu- ability ratings (observations above the line of best fit). However, by
lating difference scores between the two paired variables, mean cen- comparison, a larger number of observations were below the lines of
tering one of the paired variables, and then regressing the differences best fit, suggesting an overall reduction in desirability rating from the
scores onto the mean centered variable. In this analysis, the constant 90th to the 99th percentiles.
associated with the regression equation represents the mean difference In specific numerical terms, for sample 1, 28.0% reduced, 60.2% did
and can be tested for statistical significance (Hedberg & Ayers, 2015). not change, and 11.9% increased their rating of desirability between
As the paired variables of interest in this portion of the investigation the 90th and 99th IQ percentiles; and for sample 2, 9.5% reduced,
were measured with ranks, bias-corrected accelerated confidence in- 75.5% did not change, and 11.4% increased their rating of desirability
tervals were estimated via bootstrapping (2000 replications), i.e., an between the 90th and 99th IQ percentiles. Next, for sample 1, we ex-
estimation method that does not assume interval/ratio measurement amined the reasons why the people who did not respond ‘extremely
(Efron & Tibshirani, 1993). interested’ for the 99th IQ percentile (n = 165) did so. We found that
59.4% (n = 98) of the participants specified that they were concerned
3. Results that there would be compatibility issues with someone exceptionally
intelligent. By comparison, we found that 29.7% (n = 49) of those
Excluding the rank data from the Buss/Barnes Partner Preference participants (n = 165) specified that they were concerned that someone
Scale, all of the variable distributions were considered sufficiently with exceptional intelligence would be socially awkward. Finally,
normal for parametric analyses (i.e., skew < |2.0| kurtosis < |9.0|; 10.3% (n = 17) selected both reasons and 10.9% (n = 18) did not select
Gignac, 2019; see Table S3 and Table S4 for complete descriptive sta- either reason.
tistics). Furthermore, no outliers were identified based on the Hoaglin With respect to sample 2, a total of 75 participants (34.1%) reduced
and Iglewicz (1987) inter-quartile range outlier labelling rule with a 3.0 their attractiveness rating for a prospective partner with elevated levels
multiplier. of IQ (i.e., beyond the 75th IQ percentile). More than half (n = 43;
57.3%) of the participants who reduced their desirability rating se-
lected, “I'm worried that I would not be compatible with someone who
3.1. Percentile partner interest scale
is exceptionally intelligent”. A somewhat smaller percentage (44.0%;
n = 33) selected, “I'm worried that someone exceptionally intelligent
To determine whether the ratings of desirability for a prospective
would be socially awkward.” Some participants (n = 12; 16.0%) se-
partner were consistent with a nonlinear (threshold) effect, a 2 × 7
lected both options. A total of 19 participants from sample 2 who re-
factorial repeated measures ANOVA was conducted with intelligence
duced their rating of desirability beyond the 75th IQ percentile also
type (cognitive and emotional) as a within-subjects factor and the in-
provided open-ended comments (k = 24) to help explain their con-
telligence percentiles (1st to 99th) as another within-subjects factor,
cerns. As can be seen in Table 2, on the basis of a qualitative analysis,
across samples 1 and 2. We note that the assumption of sphericity was
we identified five categories of comments: (1) arrogance (k = 7), (2)
violated, consequently, the Huynh-Feldt adjusted results are reported.
incompatible (k = 6), (3) low EI (k = 5), (4) socially awkward (k = 3),
The intelligence type main effect (emotional versus cognitive) was
(5) too competitive (k = 3). Thus, the most commonly observed free-
not significant statistically for samples 1 or 2 (see Table 1).2 By con-
text response (9.3% of reducers) pertained to concerns surrounding
trast, the intelligence percentile main effect was significant statistically
arrogance in a person with a very high IQ. For further narrative sur-
for samples 1 and 2 (see Table 1). As can be seen in Fig. 1 (panels A and
rounding the IQ qualitative results, see the Supplementary Section.
B), both IQ and EI were associated with a pattern of means consistent
with a significant cubic effect (see Table 1 for statistics), as hypothe-
3.3. Percentage reducers and reasons: EI
sized. Additionally, the intelligence type by intelligence percentile
With respect to EI, for sample 1, 11.0% reduced, 67.4% did not
2
Sample 1: cognitive intelligence M = 3.72, SD = 0.77; emotional in- change, and 21.6% increased their desirability rating from the 90th to
telligence M = 3.69, SD = 0.78; Sample 2: cognitive intelligence M = 3.64, the 99th EI percentiles. For sample 2, 2.3% reduced, 84.5% did not
SD = 0.64; emotional intelligence M = 3.70, SD = 0.64. change, and 22.7% increased their desirability rating from the 90th to

6
G.E. Gignac and Z.M.V. Callis Intelligence 81 (2020) 101465

Table 1
ANOVA results across Sample 1 and Sample 2.
F df p partial η2

Sample 1 (N = 236)
Main Effect: Intelligence Type 0.66 1, 235 0.417 0.003
Main Effect: Intelligence Percentile 282.28 1.65, 387.03 < 0.001 0.546
Cubic Effect: Intelligence Percentile 169.48 1, 235 < 0.001 0.419
Interaction: Intelligence Type by Percentile 40.60 2.52, 591.31 < 0.001 0.147
Follow-Up: 90th to 99th percentiles 30.03 1, 235 < 0.001 0.113
Follow-Up: 75th to 90th percentiles 18.23 1, 235 < 0.001 0.072

Sample 2 (N = 213)
Main Effect: Intelligence Type 1.97 1, 219 0.162 0.009
Main Effect: Intelligence Percentile 316.71 1.85, 404.35 < 0.001 0.591
Cubic Effect: Intelligence Percentile 164.59 1, 219 < 0.001 0.429
Interaction: Intelligence Type by Percentile 29.35 2.17, 475.73 < 0.001 0.118
Follow-Up: 90th to 99th percentiles 19.10 1, 219 < 0.001 0.080
Follow-Up: 75th to 90th percentiles 10.31 1, 219 < 0.001 0.045

Panel A Panel B
6.0
6.0
5.5
IQ 5.5
5.0 EI IQ
5.0
Desirability Rating

4.5 Desirability Rating EI


4.5
4.0
4.0
3.5
3.5
3.0
3.0
2.5
2.5
2.0
2.0
1.5
1.5
1.0
1.0
1 10 25 50 75 90 99
1 10 25 50 75 90 99
Intelligence Percentile
Intelligence Percentile

Fig. 1. Plot of means depicting the nonlinear (cubic) trend between intelligence (emotional and cognitive) percentile and desirability rating (N = 236);
IQ = cognitive intelligence; EI = emotional intelligence; error bars represent 95% confidence intervals; panel A = general community sample (N = 236); panel
B = undergraduate university sample (N = 220).

the 99th EI percentiles. skew = −0.86) correlated negatively with the 10th IQ percentile, r
As noted above, although we failed to identify a significant mean (234) = −0.34, p < .001, and positively with the 90th IQ percentile, r
difference between the 90th and 99th EI percentile mean desirability (234) = 0.39, p < .001, as hypothesized. Furthermore, EI self-assessed
ratings, several participants from sample 2 (n = 53; 24.1%), none- with the single-item percentile indicator (M = 64.73; SD = 21.63;
theless, reduced their partner interest rating between the 75th and 90th skew = −0.83) correlated negatively with the 10th EI percentile, r
EI percentiles, or between the 90th and 99th EI percentiles. (234) = −0.24, p < .001, and positively with the 90th EI percentile, r
Furthermore, all 53 participants provided open-ended responses to help (234) = 0.37, p < .001, as hypothesized. Thus, people who self-as-
explain their reduction. As can be seen in Table 3, all of the free-text sessed their IQ relatively higher tended to rate less desirable a pro-
responses (k = 67) were coded to one or more of eight categories: (1) spective partner with relatively low IQ, as well as rate more desirable a
perceived incompatibility (k = 26), (2) constantly analyse me (k = 12), prospective partner with relatively high IQ. The same pattern emerged
(3) too emotions focussed (k = 8), (4) judgemental (k = 6), (5) assume for self-assessed EI, again, as hypothesized.
how I'm feeling (k = 5), (6) too intense (k = 5), (7) intimidating Next, with respect to sample 2, recall that, in addition to self-as-
(k = 3), and (8) boring relationship (k = 2). Thus, the two most com- sessed IQ (single item) and the Percentile Partner Interest Scale, we also
monly observed comments corresponded to the ‘perceived incompat- had scores from self-reported IQ (multi-item questionnaire), and task-
ibility’ category (49.1% of reducers) and the ‘constantly analyse me’ based measures of IQ. As can be seen in Table 4, self-assessed IQ, self-
category (22.6% of reducers). The six remaining categories were rela- reported IQ, and task-based IQ correlated negatively with the 10th IQ
tively less commonly observed (i.e., ≤ 15%). For further narrative percentile and positively with the 90th IQ percentile, as hypothesized.
surrounding the EI qualitative results, see the Supplementary Section. We also note that self-assessed IQ and self-reported IQ correlated po-
sitively with the task-based composite measure of IQ to the same degree
3.4. Predicting desirability ratings: IQ and EI (both r = 0.49). Thus, there was no evidence to suggest that the mea-
surement of self-perceptions of IQ with a percentile approach was any
First, with respect to sample 1, we found IQ self-assessed with the better than a conventional questionnaire approach. A multiple regres-
single-item percentile indicator (M = 66.14; SD = 20.75; sion structural equation model failed to identify any statistically

7
G.E. Gignac and Z.M.V. Callis Intelligence 81 (2020) 101465

Panel A Panel B

6 6

PPS EI 99th Percentile


PPS IQ 99th Percentile
5 5

4 4

3 3

2 2

1 1

1 2 3 4 5 6 1 2 3 4 5 6

PPS IQ 90th Percentile PPS EI 90th Percentile

Fig. 2. N = 236 (sample 1); scatter plots depicting the association between the Percentile Partner Interest Scale (PPS) desirability ratings for the 90th and 99th
percentiles (Panel A = IQ; Panel B = EI); the larger number of observations below the line of best fit especially apparent for the IQ data suggests that a greater
number of people reduced their ratings of desirability from the 90th to the 99th IQ percentiles.

Table 2 internal consistency reliability of the task-based measure of EI was so


Categories of reasons why people reduced their ratings of interest in a pro- low (i.e., 0.26), we did not interpret those results.
spective partner with an elevated IQ.
Categories k Sample response
3.5. Ranked mate characteristics
Arrogance 7 “Belittle a partner who is not as smart as them.”
Incompatible 6 “I'm worried that they would be smarter than I am.”
With respect to sample 1, kindness (M = 3.03; SD = 2.23) was nu-
Low EI 5 “I'm worried that because there is such a high level of
intelligence, his emotional intelligence … would be low.” merically the most highly ranked mate characteristic, based on the
Socially Awkward 3 “It has something to do with social awkwardness…” Buss/Barnes Partner Preference Scale (see supplementary section; Fig.
Too Competitive 3 “Im [sic] worried that intellectual debates with the S3, panel A). The ‘emotionally intelligent’ (M = 4.34; SD = 2.58) and
partner will end up in stalemate or become too heated.” ‘cognitively intelligent’ mate (M = 4.72; SD = 3.07) characteristics
were the second and third most highly ranked characteristics, nu-
Note. A total of 19 participants (sample 2) provided open-ended (free text)
responses to help explain why they reduced their interest rating; the frequencies
merically (see Table S5 for all descriptive statistics). A robust regres-
(k) correspond to the number of participants who made such a free-text com- sion-based paired t-test of the difference between the ‘emotionally in-
ment; some people generated comments that included more than one category, telligent’ and ‘cognitively intelligent’ mean ranks was significant
consequently, the total number of comments (k = 24) is greater than the total statistically, ΔM = −0.38, 95%CI: −0.74/−0.02, suggesting that EI
number of participants (n = 19) who provided free text responses. was ranked more desirable than IQ, on average (Cohen's d = −0.13).
We note that the ‘emotionally intelligent’ and ‘cognitively intelligent’
significant unique effects (see Supplementary Section; Fig. S2). ranks were also correlated positively (Spearman r = 0.27, p < .001),
Additionally, with respect to sample 2, recall that we had self-assed suggesting some tendency for higher EI ranks to be associated with
EI (single-item), self-reported EI (multi-item questionnaire), a task- higher IQ ranks, although the correlation was not so large as to suggest
based measure of EI, as well as scores from the Percentile Partner the two traits were viewed very similarly. Similar results were reported
Interest Scale – EI. As can be seen in Table 5, only the self-assessed EI for sample 2 (see supplementary section; Table S6 and Fig. S3, panel B).
indicator correlated significantly negatively and significantly positively
with the 10th and 90th EI percentiles, respectively. Finally, because the

Table 3
Categories of reasons why people reduced their ratings of interest in a prospective partner with an elevated EI.
Categories k Sample response

Perceived incompatibility 26 “I don't think i [sic] would be very compatible with someone too emotionally intelligent as I am not extremely emotionally intelligent”
Constantly analyse me 12 “I would not want them to constantly analyse my emotions and read into them.”
Too emotions focussed 8 “…I feel that the relationship would be too challenging and that a lot of the relationship would be focused on how we are feeling emotionally.”
Judgemental 6 “I'm worried they'd be … very critical of me”
Assume how I'm feeling 5 “I would not want to be with someone who never needed to ask how I was feeling and just assumed they knew.”
Too intense 5 “Someone who is super emotionally intelligent could get intense”
Intimidating 3 “intimidated by someones [sic] emotional intelligence”
Boring relationship 2 “Relationship would most likely end up boring or uninteresting.”

Note. A total of 53 participants (sample 2) provided open-ended (free text) responses to help explain why they reduced their interest rating; the frequencies (k)
correspond to the number of participants who made such a free-text comment; some people generated comments that included more than one category, consequently,
the total number of comments (k = 67) is greater than the total number of participants (n = 53) who provided free text responses.

8
G.E. Gignac and Z.M.V. Callis Intelligence 81 (2020) 101465

Table 4
Bivariate correlations between latent and observed variables included in the structural equation model: Cognitive intelligence.
SAIQ SRIQ TBIQ PPSIQ-10th PPSIQ-90th M SD skew

SAIQ 1.0 66.24 13.56 −0.18


SRIQ 0.58 1.0 3.59 0.49 0.07
TBIQ 0.49 0.49 1.0 0.00 0.77 −0.03
PPSIQ-10th −0.17 −0.15 −0.24 1.0 2.74 1.24 0.28
PPSIQ-90th 0.10 0.13 0.21 −0.34 1.0 4.53 1.09 −0.84

Note. N = 220; all correlations in bold were significant statistically (p < .05); SAIQ = self-assessed cognitive intelligence; SRIQ = self-reported cognitive in-
telligence; TBIQ = task-based cognitive intelligence; PPSIQ = Partner Preference Scale – Cognitive Intelligence; SAIQ, SRIQ and TBIQ were specified as latent
variables.

Table 5 level of IQ. Such comments are consistent with the assortative mating
Bivariate correlations between latent and observed variables. theory for IQ, that is, people tend to select partners with compatible
1. 2. 3. 4. 5. M SD skew levels of IQ (Gignac & Zajenkowski, 2019; Jonason et al., 2019). We
also note that our qualitative results are not supportive of the notion
1. SAEI 1.0 67.64 16.36 −0.73 that the typical person is satisfied with a moderate level of IQ in a
2. SREI 0.60 1.0 4.77 0.56 −0.12
partner, on the basis of the suggestion that such a level of intelligence
3. TBEI 0.17 0.31 1.0 13.48 1.96 −0.50
4. PPSEI-10th −0.14 −0.11 0.02 1.0 2.45 1.21 0.47
may be regarded as sufficient to solve daily life problems (Li, Bailey,
5. PPSEI-90th 0.30 0.36 0.05 −0.25 1.0 4.92 0.93 −1.17 Kenrick, & Linsenmeier, 2002).
Additionally, across both samples, a non-negligible percentage of
Note. N = 220; all correlations in bold were significant statistically (p < .05); the responders (≈ 40%) expressed concerns that an exceptionally in-
SAEI = self-assessed emotional intelligence; SREI = self-reported emotional telligent person would be more likely to struggle with interpersonal
intelligence; TBEI = task-based emotional intelligence; PPSEI = Partner skills. Such a result is consistent with the stereotype that exceptionally
Preference Scale – Emotional Intelligence; SAEI, SREI and TBEI were specified
intelligent/gifted people have interpersonal difficulties (França-Freitas
as latent variables.
et al., 2014; Zeidner & Shani-Zinovich, 2011). It is interesting that our
sample 2 was composed of university students and the stereotype was,
4. Discussion
nonetheless, clearly observed, as 34% (n = 75) of the sample expressed
concerns about interpersonal skills (e.g., socially awkward; low EI) with
Across both samples, we replicated the threshold effect for IQ, such
respect to the exceptionally intelligent. Such a result is essentially
that a mean reduction in rated desirability was observed from the 90th
consistent with Rodríguez-Nieto et al. (2019) who found that 25% of
to the 99th IQ percentiles. By contrast, although a threshold effect was
university students expressed the view that gifted people are more
observed for EI, as well, a mean reduction was not observed across the
likely to have socioemotional problems.
90th to the 99th EI percentiles. On the basis of a qualitative analysis, we
Consequently, we believe several respondents may have made a
found that concerns for IQ compatibility were the most frequently
trade-off, such that they prioritized the perceived social costs of ex-
specified, followed by interpersonal skill concerns. Furthermore, we
ceptional IQ over the biological benefits of IQ, even though the biolo-
found that self-assessed and objective IQ scores correlated positively
gical and socio-economic benefits of IQ extend across the full spectrum
with desirability ratings for a prospective partner with an IQ at the 90th
of IQ (Hagenaars et al., 2016; Lubinski & Benbow, 2006). Such a trade-
percentile, whereas only self-assessed EI scores correlated positively
off may be unnecessary, if the notion that exceptional levels of IQ are
and significantly with the 90th EI percentile desirability ratings.
associated with inter-personal difficulties is only a stereotype (Baudson,
Finally, across both samples, ‘emotionally intelligent’ was ranked sta-
2016). Thus, it would be important to determine the association be-
tistically significantly higher (more desirable) than ‘cognitively in-
tween objective IQ and objective interpersonal skills in future research.
telligent’. We discuss each of these key results below.

4.2. EI threshold effect and reasons


4.1. IQ threshold effect and reasons
The current study is the first to specifically examine what level of EI
Although IQ has been found to be consistently one of the most is rated the most attractive in a prospective partner. Across both sam-
highly ranked characteristics in a potential partner (Buss et al., 1990; ples, we observed a threshold effect, similar to IQ, with attraction
Buss & Barnes, 1986; Goodwin & Tinker, 2002), the association be- peaking at the 90th EI percentile. Thus, the results of our investigation
tween degree of desirability and level of IQ was found to correspond to provide more detailed insight into the degree to which EI is attractive,
a threshold effect in two studies (Gignac et al., 2018; Gignac & in comparison to Schutte et al. (2001) who found that participants
Starbuck, 2019). Consistent with these two previous studies, a small expected to have greater relationship satisfaction with someone who
(Cohen, 1992) but statistically significant reduction in mean rated de- could recognise and manage emotions, in comparison to a partner who
sirability of a prospective partner was observed from the 90th to the could not.
99th IQ percentiles, across both samples in this investigation. Thus, we Unlike IQ, however, our participants were, on average, essentially
confirmed further the threshold effect of IQ, with respect to rated equally interested in a potential partner who was described as being
partner attractiveness, i.e., an IQ of approximately 120 appears to be more emotionally intelligent than 90% and 99% of the population;
the most attractive in a prospective partner, on average. thus, rated desirability essentially peaked at the 90th percentile, sug-
A unique contribution associated with this investigation is that we gesting a threshold effect. This suggests that a high level of EI, rather
uncovered why some participants did not rate the 99th IQ percentile as than an exceptional level, is sufficient to elicit the maximum possible
the most attractive. Specifically, we found that more than half of the mean level of rated attraction from others.
respondents (≈ 60%) expressed concerns relevant to intellectual Gignac and Starbuck (2019) found that physical attractiveness and
compatibility. Some participants stated they were concerned that they kindness yielded similar threshold effects to that observed for EI, i.e., no
would have different interests to an exceptionally intelligent person, as significant mean change from the 90th to the 99th percentiles. Thus, on
well as have difficulty communicating with a partner with a very high the basis of the results of this investigation, as well as those of Gignac

9
G.E. Gignac and Z.M.V. Callis Intelligence 81 (2020) 101465

et al. (2018); Gignac and Zajenkowski (2019), it would appear that 4.4. Which is more desirable, IQ or EI?
exceptional levels of several attributes yield, on average, essentially no
meaningful incremental benefits with respect to the perception of Although Schutte et al. (2001) found that people expected to enjoy a
prospective partner desirability, even though several of those char- romantic relationship with someone who could recognise and manage
acteristics produce greater socio-economic benefits across the whole emotions more than someone who could not, our results offer a clearer
spectrum of the characteristic (Lubinski & Benbow, 2006; Patel & indication of the attractiveness of EI. Specifically, we found that EI was
Wolfe, 2019). the second most desirable trait in a potential partner, behind only ‘kind
Although there was no significant difference between the mean and understanding’, on the basis of a modified Buss/Barnes Partner
partner interest ratings between potential partners with EI levels at the Preference Scale (Buss & Barnes, 1986). In previous research, the
90th and 99th percentiles, some participants did reduce the their de- second most highly ranked characteristic in a partner has been reported
sirability ratings from the 90th to the 99th EI percentiles. In a manner consistently to be ‘exciting personality’ or ‘intelligent’ (Buss et al., 1990;
similar to IQ, perceived incompatibility was the most common reason Gignac et al., 2018). Thus, by distinguishing EI from IQ, our in-
for reducing ratings of interest for the exceptionally emotionally in- vestigation uncovered more detailed insights into how people perceive
telligent. More specifically, some participants were worried the dis- the value of intelligence, emotional and cognitive, with respect to a
parity in EI between themselves and the partner would result in an potential partner: both are valued relatively highly, but EI more so.
inability to relate to them, that they would be intimidated by someone Our results may be considered consistent with the empirical findings
with high EI, or that a partner high in EI would be judgemental of them. that suggest that EI is associated positively with partner closeness,
These results are consistent with the qualitative results we obtained for commitment, relationship success and relationship satisfaction
IQ, underscoring the compatibility theory of mate selection (Huston & (Apostolou et al., 2019; Schröder-Abé & Schütz, 2011). Furthermore, EI
Houts, 1998). has been found to predict job performance uniquely, i.e., independently
Also consistent with the compatibility theory, in this context, is that of the effects of personality and IQ (Joseph et al., 2015; O'Boyle Jr et al.,
people tend to overestimate their IQ and EI, when measured on a self- 2011), suggesting potential socio-economic benefits. Thus, on average,
assessed percentile continuum. Specifically, on average, people tend to our results suggest that people are likely aware of these potential
self-report their IQ and EI at between somewhere equal to 112 to 120 benefits, both relationship-related and socio-economic, when con-
on the conventional IQ scale (Petrides & Furnham, 2000), which cor- sidering a relationship with someone relatively high in EI. By com-
responds to somewhere between the 80th and 90th percentiles. Corre- parison, IQ has not been found to correlate positively with relationship
spondingly, for both IQ and EI, the mean rated desirability level for a satisfaction (Gignac & Zajenkowski, 2019). Consequently, on average,
prospective partner peaked at approximately the 90th percentile for people may have ranked IQ somewhat lower than EI, as there is some
both IQ and EI, in this investigation. awareness among adults that IQ offers fewer relationship-related ben-
efits, in comparison to EI.
4.3. Predictors of IQ and EI desirability We note that, although both EI and IQ were ranked relatively
highly, people did not view EI and IQ as essentially the same char-
We found that both self-assessed IQ and objectively measured IQ acteristic. Based on supplementary analyses, the EI and IQ ranks cor-
correlated negatively with desirability ratings for a relatively less in- related (Spearman) at only r(234) = 0.27 (p < .001) and r
telligent prospective partner (10th IQ percentile) and positively for a (211) = 0.11 (p = .111), in samples 1 and 2, respectively.
relatively more intelligent prospective partner (90th IQ percentile). The Correspondingly, based on individual differences research, EI and IQ
true score correlations were in the relatively small to typical range for have been reported to correlate positively only relatively weakly
individual differences research (Gignac & Szodorai, 2016). (r < 0.30; O'Boyle Jr et al., 2011; van Rooy & Viswesvaran, 2004).
Although a substantial amount of research has established the as- Thus, from both an individual differences in ability perspective, as well
sortative mating for IQ effect (Bouchard & McGue, 1981), how precisely as an individual differences in perceived partner characteristic desir-
the effect arises remains to be determined (Plomin & Deary, 2015). ability perspective, IQ and EI are substantially different constructs.
Some research suggests that assortative mating for IQ may be caused by
a third variable: educational propinquity (Phillips, Fulker, Carey, & 4.5. Limitations
Nagoshi, 1988). However, on the basis of a close examination, Stevens
(1991) found that assortative mating for education (and, by inference, Although this investigation was associated with several strengths,
IQ) likely does not arise through educational propinquity. Corre- there were also several limitations. First, we are unsure of the number
spondingly, our results suggest that people have some awareness of of Hispanics who completed the survey in sample 1, as such information
their own level of objective IQ (r ≈ 0.50), and that people who rate was not collected. Additionally, MTurk workers may be especially test-
their IQ relatively higher tend to find more desirable a prospective wise participants. Consequently, the degree to which our sample 1 re-
partner with a higher IQ. Consequently, our results suggest that assor- sults truly generalise to the US population may be questioned.
tative mating for IQ may arise, at least in part, via active assortment, Furthermore, the second sample consisted entirely of first year under-
i.e., direct effects that arise through individual differences in mating graduate students, a population known to be associated with range
preferences (Watson et al., 2004). restriction in IQ (Gignac et al., 2004; Osmon & Jackson, 2002). Thus,
With respect to EI, our measure of objective EI yielded internal the correlations between objective intelligence and the desirability
consistency reliability so low (α = 0.26) as to be unusable. However, ratings reported in this investigation are likely underestimates. Conse-
we found that our measure of self-reported EI correlated significantly quently, a future study with an ethnically diverse, general community
and positively with ratings of desirability for a prospective partner with sample, and objective measures of IQ, would be useful.
a high level of EI (90th percentile). The true score correlation was re- We also acknowledge that the manner in which the Partner
latively large (Gignac & Szodorai, 2016). Such a result suggests the Percentile Interest Scales measured the participants' responses may
possibility for active assortment for EI. However, to date, there is only a have impacted the results. Specifically, although the participants did
small amount of research on the assortative mating for EI and the re- not have to know what a percentile was in order to respond to the items
sults are inconsistent (Schröder-Abé & Schütz, 2011; Śmieja & Stolarski, on the scale, it is possible that focussing on frequencies (e.g., 1 out of
2018; Zeidner & Kaluda, 2008). More high quality research in this area 100), rather than percentages (“Smarter than 99% of the population.”),
is encouraged, particularly considering the interpersonal and in- may yield more valid participant responses. Additionally, the percen-
trapersonal benefits associated with EI and relationship satisfaction tiles associated with the scale increase in a non-linear manner (see X-
(Amitay & Mongrain, 2007; Schröder-Abé & Schütz, 2011). axes Fig. 1), which may have impacted the results. Ultimately, it would

10
G.E. Gignac and Z.M.V. Callis Intelligence 81 (2020) 101465

be useful to try to replicate the results with alternative approaches to References


measuring people's desirability ratings.
Additionally, we did not measure relationship status in this in- Allen, V. D., Weissman, A., Hellwig, S., MacCann, C., & Roberts, R. D. (2014).
vestigation. Instead, we asked participants to imagine they were single, Development of the situational test of emotional understanding–brief (STEU-B) using
item response theory. Personality and Individual Differences, 65, 3–7.
if they were not, when rating the desirability of the prospective partner. Amitay, O. A., & Mongrain, M. (2007). From emotional intelligence to intelligent choice
There is research to suggest that single people, on average, rate pro- of partner. The Journal of Social Psychology, 147(4), 325–343. https://doi.org/10.
spective partners more positively (Simpson, Gangestad, & Lerma, 3200/SOCP.147.4.325-344.
Apostolou, M., Paphiti, C., Neza, E., Damianou, M., & Georgiadou, P. (2019). Mating
1990). Additionally, some work suggests that people in a relationship performance: Exploring emotional intelligence, the dark triad, Jealousy and
are less motivated to provide valid ratings of desirability for a pro- Attachment Effects. Journal of Relationships Research, 10.
spective partner (Lydon, 2010). Thus, the results reported in this in- Austin, E. J. (2010). Measurement of ability emotional intelligence: Results for two new
tests. British Journal of Psychology, 101(3), 563–578.
vestigation may be moderated by relationship status. Future research to Baudson, T. G. (2016). The mad genius stereotype: Still alive and well. Frontiers in
evaluate this possibility is indicated. Correspondingly, we also ac- Psychology, 7, 368. https://doi.org/10.3389/fpsyg.2016.00368.
knowledge that most people would not be expected to have a large Bouchard, T. J., & McGue, M. (1981). Familial studies of intelligence: A review. Science,
212(4498), 1055–1059.
amount of experience with people who are exceptionally intelligent;
Brauer, D. V., & DePaulo, B. M. (1980). Similarities between friends in their under-
consequently, it may have been difficult for many of our participants to standing of nonverbal cues. Journal of Nonverbal Behavior, 5(1), 64–68. https://doi.
provide well considered ratings for a person with an IQ at the 99th org/10.1007/BF00987057.
percentile, for example. Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon's Mechanical Turk: A new
source of inexpensive, yet high-quality, data? Perspectives on Psychological Science,
Furthermore, in study 2, the open-ended responses were captured 6(1), 3–5.
after the participants responded to the multiple-choice item to help Buss, D. M., Abbott, M., Angleitner, A., Asherian, A., Biaggio, A., Blanco-Villasenor, A., ...
explain why they reduced their ratings of desirability. It is possible that Ekehammar, B. (1990). International preferences in selecting mates: A study of 37
cultures. Journal of Cross-Cultural Psychology, 21(1), 5–47. https://doi.org/10.1177/
other reasons were not captured, as some participants may have felt 0022022190211001.
that selecting one of the options was a sufficient explanation. Future Buss, D. M., & Barnes, M. (1986). Preferences in human mate selection. Journal of
research should first present participants with a fully open-ended Personality and Social Psychology, 50(3), 559–570. https://doi.org/10.1037/0022-
3514.50.3.559.
question and then follow-up with the multiple-choice item on the next Buss, D. M., & Schmidt, D. P. (1993). Sexual strategies theory: An evolutionary per-
screen. spective on human mating. Psychological Review, 100, 204–232.
Additionally, the study design and analyses undertaken in this in- Buunk, B. P., Dijkstra, P., Fetchenhauer, D., & Kenrick, D. T. (2002). Age and gender
differences in mate selection criteria for various involvement levels. Personal
vestigation isolated each characteristic. Although such an approach Relationships, 9(3), 271–278.
may have advantages that facilitate clear interpretations of results, it is Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159.
a limitation, as people, in practice, likely consider a multitude of fac- Cooper, R. A., & Weekes, A. J. (1983). Data, models and statistical analyses. Totowa, NJ:
Barnes & Noble.
tors, simultaneously, when considering the attractiveness of a pro-
Der, G., Batty, G. D., & Deary, I. J. (2009). The association between IQ in adolescence and
spective partner (Luo & Klohnen, 2005), factors that may also interact a range of health outcomes at 40 in the 1979 US National Longitudinal Study of
with each other. For example, on the basis of the qualitative comments, Youth. Intelligence, 37(6), 573–580. https://doi.org/10.1016/j.intell.2008.12.002.
some participants expressed concerns that a prospective partner very Efron, B., & Tibshirani, R. (1993). An introduction to the bootstrap. Boca Raton, FL:
Chapman and Hall/CRC.
high in IQ would be relatively lower in EI. Thus, we naturally ac- Zabaneh, M. L. P. D., Del Prette, A., & Del Prette, Z. A. P. (2014). Social skills of gifted and
knowledge that the manner in which the process of mate attraction and talented children. Estudos de Psicologia. Estudos de Psicologia (Natal), 19(4),
selection occurs in nature is much more complex than that suggested by 288–295.
Furnham, A. (2001). Self-estimates of intelligence: Culture and gender difference in self
the research design we employed in this investigation. Correspondingly, and other estimates of both general (g) and multiple intelligences. Personality and
we acknowledge that we measured ratings of interest, not actual be- Individual Differences, 31(8), 1381–1405.
haviour. Consequently, it remains to be determined the degree to which Furnham, A., Moutafi, J., & Chamorro-Premuzic, T. (2005). Personality and intelligence:
Gender, the big five, self-estimated and psychometric intelligence. International
people who rate intelligence highly actually make dating/mating de- Journal of Selection and Assessment, 13(1), 11–24.
cisions on the basis of those reported preferences. Gignac, G. E. (2010a). On a nomenclature for emotional intelligence research. Industrial
Finally, we did not report sex differences in this investigation. On and Organizational Psychology, 3(2), 131–135. https://doi.org/10.1111/j.1754-9434.
2010.01212.x.
the basis of the male and female mean ranks reported for the Buss/
Gignac, G. (2010b). Seven-factor model of emotional intelligence as measured by Genos
Barnes Partner Preference Scale in a large-scale international in- EI: A confirmatory factor analytic investigation based on self-and rater-report data.
vestigation (Buss et al., 1990), we estimated a correlation of 0.93 be- European Journal of Psychological Assessment, 26(4), 309–316.
Gignac, G. E. (2018). In V. Zeigler-Hill, & T. Shackelford (Eds.). The SAGE handbook of
tween the male and female mean ranks. Thus, although some differ-
personality and individual differences. (pp. 439–464). Thousand Oakes, CA: SAGE.
ences between males and females may exist, they are probably not large Gignac, G. E. (2019). How2statsbook (Online Edition 1.0). Perth, Australia: Author. www.
(see Supplementary Section). Nonetheless, we encourage sex differ- how2statsbook.com.
ences (and sexual orientation) research in this area, in order to de- Gignac, G. E., Darbyshire, J., & Ooi, M. (2018). Some people are attracted sexually to
intelligence: A psychometric evaluation of sapiosexuality. Intelligence, 66, 98–111.
termine whether the effects reported, here, were moderated by sex. https://doi.org/10.1016/j.intell.2017.11.009.
Gignac, G. E., Shankaralingam, M., Walker, K., & Kilpatrick, P. (2016). Short-term
5. Conclusion memory for faces relates to general intelligence moderately. Intelligence, 57, 96–104
10.1016/j.intell.2016.05.001.
Gignac, G. E., & Starbuck, C. L. (2019). Exceptional intelligence and easygoingness may
Elevated levels of IQ and EI are highly valued characteristics in a hurt your prospects: Threshold effects for rated mate characteristics. British Journal of
prospective partner, although EI somewhat more than IQ. Reasons for Psychology, 110(1), 151–172. https://doi.org/10.1111/bjop.12342.
Gignac, G. E., Stough, C., & Loukomitis, S. (2004). Openness, intelligence, and self-report
the mean reduction in desirability of a prospective partner with ex- intelligence. Intelligence, 32(2), 133–143.
ceptional levels of IQ revolve primarily around issues of compatibility, Gignac, G. E., & Szodorai, E. T. (2016). Effect size guidelines for individual differences
and, secondarily, concerns for perceived social skill difficulties. It is researchers. Personality and Individual Differences, 102, 74–78.
Gignac, G. E., & Zajenkowski, M. (2019). People tend to overestimate their romantic
arguably important to determine conclusively whether there is a non-
partner's intelligence even more than their own. Intelligence, 73, 41–51. https://doi.
linear association between IQ and social skills/EI, as some people may org/10.1016/j.intell.2019.01.004.
be making an unnecessary trade-off, in the context of mate selection. Goodwin, R., & Tinker, M. (2002). Value priorities and preferences for a relationship
partner. Personality and Individual Differences, 32(8), 1339–1349. https://doi.org/10.
1016/S0191-8869(01)00122-2.
Appendix A. Supplementary data Gottfredson, L. S. (2002). g: Highly general and highly practical. In R. J. Sternberg, & E. L.
Grigorenko (Eds.). The general factor of intelligence: How general is it? (pp. 331–380).
Supplementary data to this article can be found online at https:// Mahwah, NJ: Erlbaum.
Hagenaars, S. P., Harris, S. E., Davies, G., Hill, W. D., Liewald, D. C., Ritchie, S. J., ...
doi.org/10.1016/j.intell.2020.101465.

11
G.E. Gignac and Z.M.V. Callis Intelligence 81 (2020) 101465

Group, C. C. P (2016). Shared genetic aetiology between cognitive functions and factor mediated by developmental stability. Intelligence, 33(2), 203–213. https://doi.
physical and mental health in UK Biobank (N= 112 151) and 24 GWAS consortia. org/10.1016/j.intell.2004.07.007.
Molecular Psychiatry, 21(11), 1624. https://doi.org/10.1038/mp.2015.225. Raven, J. C. (1966). Advanced progressive matrices. New York: Psychological Corporation.
Hedberg, E. C., & Ayers, S. (2015). The power of a paired t-test with a covariate. Social Reitan, R. M. (1958). Validity of the Trail Making Test as an indicator of organic brain
Science Research, 50, 277–291. damage. Perceptual and Motor Skills, 8(3), 271–276.
Hoaglin, D. C., & Iglewicz, B. (1987). Fine-tuning some resistant rules for outlier labelling. Rodríguez-Nieto, M. C., Sánchez-González, A. S., & Sánchez-Miranda, M. P. (2019). How
Journal of the American Statistical Association, 82, 1147–1149. https://doi.org/10. are the gifted? Point of view of university students. Educational Process: International
1080/01621459.1987.10478551. Journal, 8(2), 123.
Huston, T., & Houts, R. (1998). The psychological infrastructure of courtship and mar- van Rooy, D. L., & Viswesvaran, C. (2004). Emotional intelligence: A meta-analytic in-
riage: The role of personality and compatibility in romantic relationships. In T. vestigation of predictive validity and nomological net. Journal of Vocational Behavior,
Bradbury (Ed.). The developmental course of marital dysfunction (pp. 114–151). New 65(1), 71–95.
York: Cambridge University Press. Salovey, P., & Mayer, J. D. (1990). Emotional intelligence. Imagination, Cognition and
Jonason, P. K., Marsh, K., Dib, O., Plush, D., Doszpot, M., Fung, E., ... Di Pietro, K. (2019). Personality, 9(3), 185–211. https://doi.org/10.2190/DUGG-P24E-52WK-6CDG.
Is smart sexy? Examining the role of relative intelligence in mate preferences. Salthouse, T. A., Toth, J., Daniels, K., Parks, C., Pak, R., Wolbrette, M., & Hocking, K. J.
Personality and Individual Differences, 139, 53–59. https://doi.org/10.1016/j.paid. (2000). Effects of aging on efficiency of task switching in a variant of the trail making
2018.11.009. test. Neuropsychology, 14(1), 102–111.
Joseph, D. L., Jin, J., Newman, D. A., & O'Boyle, E. H. (2015). Why does self-reported Schlegel, K., Fontaine, J. R., & Scherer, K. R. (2017). The Nomological network of emotion
emotional intelligence predict job performance? A meta-analytic investigation of recognition ability. European Journal of Psychological Assessment, 1–12.
mixed EI. Journal of Applied Psychology, 100(2), 298–342. Schröder-Abé, M., & Schütz, A. (2011). Walking in each other's shoes: Perspective taking
van Leeuwen, M., Van Den Berg, S. M., & Boomsma, D. I. (2008). A twin-family study of mediates effects of emotional intelligence on relationship quality. European Journal of
general IQ. Learning and Individual Differences, 18(1), 76–88. https://doi.org/10. Personality, 25(2), 155–169. https://doi.org/10.1002/per.818.
1016/j.lindif.2007.04.006. Schutte, N. S., Malouff, J. M., Bobik, C., Coston, T. D., Greeson, C., Jedlicka, C., ...
Li, N. P., Bailey, J. M., Kenrick, D. T., & Linsenmeier, J. A. (2002). The necessities and Wendorf, G. (2001). Emotional intelligence and interpersonal relations. The Journal
luxuries of mate preferences: Testing the tradeoffs. Journal of Personality and Social of Social Psychology, 141(4), 523–536. https://doi.org/10.1080/
Psychology, 82(6), 947. https://doi.org/10.1037/0022-3514.82.6.947. 00224540109600569.
Libbrecht, N., & Lievens, F. (2012). Validity evidence for the situational judgment test Schütz, A., & Koydemir, S. (2018). Emotional intelligence: What it is, how it can be
paradigm in emotional intelligence measurement. International Journal of Psychology, measured and increased and whether it makes us successful and happy. In T. Todd
47(6), 438–447. Shackelford, & V. Zeigler-Hill (Eds.). The SAGE handbook of personality and individual
Lubinski, D., & Benbow, C. P. (2006). Study of mathematically precocious youth after 35 differences (pp. 397–423). Thousand Oaks, CA: Sage.
years: Uncovering antecedents for the development of math-science expertise. Simpson, J. A., Gangestad, S. W., & Lerma, M. (1990). Perception of physical attrac-
Perspectives on Psychological Science, 1(4), 316–345. https://doi.org/10.1111/j.1745- tiveness: Mechanisms involved in the maintenance of romantic relationships. Journal
6916.2006.00019.x. of Personality and Social Psychology, 59, 1192–1201.
Luo, S., & Klohnen, E. C. (2005). Assortative mating and marital quality in newlyweds: A Śmieja, M., & Stolarski, M. (2018). Assortative mating for emotional intelligence. Current
couple-centered approach. Journal of Personality and Social Psychology, 88(2), Psychology, 37(1), 180–187. https://doi.org/10.1007/s12144-016-9501-8.
304–326. Sternberg, R. J., Conway, B. E., Ketron, J. L., & Bernstein, M. (1981). People's conceptions
Lydon, J. E. (2010). How to forego forbidden fruit: The regulation of attractive alter- of intelligence. Journal of Personality and Social Psychology, 41(1), 37–55.
natives as a commitment mechanism. Social and Personality Psychology Compass, 4, Stevens, G. (1991). Propinquity and educational homogamy. Sociological Forum, 6(4),
635–644. 715–726.
MacCann, C., & Roberts, R. D. (2008). New paradigms for assessing emotional in- Strauss, A., & Corbin, J. (1990). Basics of qualitative research. Sage publications.
telligence: Theory and data. Emotions, 8(4), 540–551. von Stumm, S. (2014). Intelligence, gender, and assessment method affect the accuracy of
Mascie-Taylor, C. G. N. (1988). Assortative mating from psychometric characteristics. In self-estimated intelligence. British Journal of Psychology, 105(2), 243–253.
C. G. N. Mascie-Taylor, & A. J. Boyce (Eds.). Human mating patterns (pp. 62–82). Swami, V., Furnham, A., & Zilkha, S. (2009). Estimates of self, parental, and partner
Cambridge, MA: Cambridge University Press. multiple intelligence and their relationship with personality, values, and demo-
Mascie-Taylor, C. N. (1989). Spouse similarity for IQ and personality and convergence. graphic variables: A study in Britain and France. The Spanish Journal of Psychology,
Behavior Genetics, 19(2), 223–227. https://doi.org/10.1007/BF01065906. 12(2), 528–539.
McIntosh, D. E., Dixon, F. A., & Pierson, E. E. (2005). Use of intelligence tests in the Torrance, E. P. (1962). Guiding creative talent. Englewood Cliffs, NJ: Prentice-Hall.
identification of giftedness. In D. P. Flanagan, & P. L. Harrison (Eds.). Contemporary United States Census Bureau (2019). Educational attainment in the United States: 2018.
intellectual assessment (pp. 504–520). New York, NY: Guilford Press. Retrieved from https://www.census.gov/data/tables/2018/demo/education-
O'Boyle, E. H., Jr., Humphrey, R. H., Pollack, J. M., Hawver, T. H., & Story, P. A. (2011). attainment/cps-detailed-tables.html.
The relation between emotional intelligence and job performance: A meta-analysis. Watson, D., Klohnen, E. C., Casillas, A., Nus Simms, E., Haig, J., & Berry, D. S. (2004).
Journal of Organizational Behavior, 32(5), 788–818. Match makers and deal breakers: Analyses of assortative mating in newlywed cou-
Osmon, D. C., & Jackson, R. (2002). Inspection time and IQ: Fluid or perceptual aspects of ples. Journal of Personality, 72(5), 1029–1068.
intelligence? Intelligence, 30(2), 119–127. Wilson, G., & Cousins, J. (2003). Partner similarity and relationship satisfaction:
Palmer, B. R., Stough, C., Hamer, R., & Gignac, G. E. (2009). Genos emotional intelligence Development of a compatibility quotient. Sexual and Relationship Therapy, 18(2),
inventory. In C. Stough, D. Saklofske, & J. D. Parker (Eds.). Advances in the mea- 161–170.
surement of emotional intelligence (pp. 103–118). New York: Springer. Zabaneh, D., Krapohl, E., Gaspar, H. A., Curtis, C., Lee, S. H., & Lubinski, D. (2018). A
Patel, P. C., & Wolfe, M. T. (2019). In the eye of the beholder? The returns to beauty and genome-wide association study for extremely high intelligence. Molecular Psychiatry,
IQ for the self-employed. Strategic Entrepreneurship Journal. https://doi.org/10.1002/ 23(5), 1226–1232.
sej.1323 Advance online publication. Zagorsky, J. L. (2007). Do you have to be smart to be rich? The impact of IQ on wealth,
Petrides, K. V. (2011). Ability and trait emotional intelligence. In T. Chamorro-Premuzic, income and financial distress. Intelligence, 35(5), 489–501.
A. Furnham, & S. von Stumm (Eds.). The Blackwell-Wiley handbook of individual dif- Zarch, Z. N., Marashi, S. M., & Raji, H. (2014). The relationship between emotional in-
ferences (pp. 656–678). New York: Wiley. telligence and marital satisfaction: 10-year outcome of partners from three different
Petrides, K. V., & Furnham, A. (2000). Gender differences in measured and self-estimated economic levels. Iranian Journal of Psychiatry, 9(4), 188. Retrieved from https://
trait emotional intelligence. Sex Roles, 42(5–6), 449–461. search-proquest-com.ezproxy.library.uwa.edu.au/docview/1640731647?
Phillips, K., Fulker, D. W., Carey, G., & Nagoshi, C. T. (1988). Direct marital assortment accountid=14681.
for cognitive and personality variables. Behavior Genetics, 18(3), 347–356. Zeidner, M., & Kaluda, I. (2008). Romantic love: What's emotional intelligence (EI) got to
Pintner, R. (1923). Intelligence testing. New York: Holt, Rinehart, & Winston. do with it? Personality and Individual Differences, 44(8), 1684–1695. https://doi.org/
Plomin, R., & Deary, I. J. (2015). Genetics and intelligence differences: Five special 10.1016/j.paid.2008.01.018.
findings. Molecular Psychiatry, 20(1), 98. Zeidner, M., & Shani-Zinovich, I. (2011). Do academically gifted and nongifted students
Plomin, R., & von Stumm, S. (2018). The new genetics of intelligence. Nature Reviews differ on the Big-Five and adaptive status? Some recent data and conclusions.
Genetics, 19(3), 148–159. Personality and Individual Differences, 51(5), 566–570. https://doi.org/10.1016/j.
Prokosch, M. D., Yeo, R. A., & Miller, G. F. (2005). Intelligence tests with higher g- paid.2011.05.007.
loadings show higher correlations with body symmetry: Evidence for a general fitness

12

You might also like