Testing a multidimensional model of putative evolved human motives

Abstract

The Assessment of Individual Motives-Questionnaire (AIM-Q) was designed to assess 15 latent dimensions based on a relatively new evolutionary theory of human motivation. The present studies are the first to examine these dimensions using confirmatory factor analysis in an attempt to test the model proposed by the theory. Study 1 (N = 1,411) explored ways of producing a short version tapping the 15 dimensions. Fit indices suggested that a model consisting of 60 items tapping 15 weak to moderately correlated dimensions best described the data. Four alternative models that tested a completely orthogonal structure and a unidimensional structure were found to poorly represent the data. Study 2 (N = 490) successfully cross-validated the 60-item 15 dimension version using a different sample. These results support the multidimensional motivational theory on which the AIM-Q is based as well as the shorter version used to assess its dimensions.

Full text

ORIGINAL PAPER
Testing a multidimensional model of putative evolved human
motives
Larry C. Bernard •Andrew Lac
Published online: 1 May 2013
Springer Science+Business Media New York 2013
Abstract The Assessment of Individual Motives-Ques-
tionnaire (AIM-Q) was designed to assess 15 latent
dimensions based on a relatively new evolutionary theory
of human motivation. The present studies are the first to
examine these dimensions using confirmatory factor anal-
ysis in an attempt to test the model proposed by the theory.
Study 1 (N=1,411) explored ways of producing a short
version tapping the 15 dimensions. Fit indices suggested
that a model consisting of 60 items tapping 15 weak to
moderately correlated dimensions best described the data.
Four alternative models that tested a completely orthogonal
structure and a unidimensional structure were found to
poorly represent the data. Study 2 (N=490) successfully
cross-validated the 60-item 15 dimension version using a
different sample. These results support the multidimen-
sional motivational theory on which the AIM-Q is based as
well as the shorter version used to assess its dimensions.
Keywords Motivation Motives Evolution Individual
differences Confirmatory factor analysis
Introduction
The results of two studies that support a new multidi-
mensional model of human motivation based on evolu-
tionary theory are presented. We briefly trace the
development of the theory, discuss the multiple hypothe-
sized dimensions of individual differences derived from it,
and place it in the context of past and present motivational
theories. The two confirmatory factor analyses (CFA)
reported here are the culmination of a programmatic series
of earlier work and are the crucial test of the 15 factor
model upon which the theory rests. These analyses are also
crucial to the development of a parsimonious and psy-
chometrically sound instrument that could serve as an
impetus for future research on the putative (i.e., currently
hypothetical) evolutionary basis of the model, its cross-
cultural relevance, and its predictive validity.
An evolutionary theory of human motives
Motivation science has experienced a reinvigoration of
interest in recent years (e.g., Ryan 2012; Sorrentino and
Higgins 1986; Shah and Gardner 2008). Consistent with
this renewed interest, Bernard et al. (2005a) developed a
theory of motivation that combined the perspectives of
evolutionary and individual differences with motivational
psychology. They proposed that ‘‘ultimate’’ factors—cir-
cumstances and conditions that shaped a species’ devel-
opment in its environments of evolutionary adaptedness
(EEAs; see Hagen 2005, for a discussion of this term)—are
partly responsible for purposeful human behavior today.
They posited that domain-specific motives evolved in the
EEA to prepare the brain for purposeful action in the face
of the most common challenges encountered by humans in
the past: (1) how to obtain and maintain resources; (2) how
to compete for and attract a mate; (3) how to form coali-
tions of kin and non-kin for mutual support; and (4) how to
rationalize existence when the only certainty is death.
Bernard et al. (2005a) conceptualized motives as
behavior-activating neural circuits in the brain. These cir-
cuits arose independently through evolutionary selection
pressures that resulted in brains that are prepared to
L. C. Bernard (&)A. Lac
Psychology Department, Loyola Marymount University,
Los Angeles, CA 90045, USA
e-mail: lbernard@lmu.edu
URL: http://myweb.lmu.edu/lbernard
123
Motiv Emot (2014) 38:47–64
DOI 10.1007/s11031-013-9360-7

facilitate specific behavioral syndromes at different
strength levels. A behavioral syndrome is a correlated suite
of behaviors (Roff 2001; Sih et al. 2004a,b). Individual
differences in strength levels of behavioral syndromes are
due to different phenotypes within the same behavioral
dimension. Each phenotype is based on a genotype that
survived selection pressures because it had different
advantages and disadvantages at different times in the
EEAs. Genes set a reaction range, an upper and lower
limit, for the potential expression of any given motive
phenotype while environmental conditions facultatively
calibrate the strength level of an individual’s phenotype
within this range (Buss 2009; Nettle 2006; Lukaszewski
and Roney 2011; Platt and Sanislow 1988).
In the renewal of motivation science, the term ‘‘motive’’
has been preferred to ‘‘trait,’’ which is used widely in
personality psychology. ‘‘Motive’’ is defined as ‘‘…apre-
disposition to behave in a directed fashion…act(ing) as the
motor for action…(and) energizing purposive behavior that
serves a function for the individual’’ (Fiske 2008, p. 4).
Bernard et al. (2005a) identified the hypothetical chal-
lenges faced by humans in the EEAs and linked them to 15
motive dimensions that should address them. They pro-
posed that the motives should be evident in individual
differences in contemporary human behavior. Listed in
order of the size of the social domain in which they
developed, the motives are: (1) Individual Domain—
Environmental Inquisitiveness, Illness Avoidance, Threat
Avoidance; (2) Dyadic Domain—Aggression, Interper-
sonal Inquisitiveness, Sex, and the four ‘‘competitive/sta-
tus’’ motives Appearance, Mental, Physical, Wealth; (3)
Small Group Domain—Commitment; (4) Large Group
Domain—two of three ‘‘cooperative’’ motives Altruism
and Social Exchange; and (5)Very Large Group Domain—
the third cooperative motive Legacy and Meaning. See
Table 1for the recurring fitness challenge and conceptual
definition of each motive.
Evolutionary psychology holds that the mind is ‘‘multi-
modular.’’ As Tooby and Cosmides (1992) recognized:
‘‘Our ability to perform most of the environmentally
engaged, richly contingent activities that we do depends on
the guiding presence of a large number of highly special-
ized psychological mechanisms’’ (italics added; p. 39).
Tooby and Cosmides spoke of an evolved human mental
structure made up of multiple domain-specific modules that
‘‘ …does not constrain; it creates or enables’’ (p. 39). (Note
the decidedly motivational connotation of the use of
‘‘creates’’ and ‘‘enables.’’) Evolutionary psychologists
hypothesize that these various domain-specific modules
evolved to address the problems and challenges faced by
humans in the past (see, Buss 2012, for specific examples).
They then test these hypothesized modules using standard
social science methods (Confer et al. 2010).
Bernard et al. (2005a) believed that the multi-
dimensional/multifactorial frameworks of individual dif-
ferences psychology matched well with the evolutionary
psychology concept of mental modularity.
1
Nevertheless,
research on these motive dimensions and advances in
evolutionary theory necessitated a revision in the theory
(Bernard 2012). This revision incorporated new thinking
about how heritable individual differences can develop
through the mechanisms of balancing/fluctuating selection
(Penke et al. 2007) and trade-offs between relative
advantages and disadvantages in different environments
(Nettle 2006).
As an example, Fig. 1depicts how these mechanisms
might give rise to the ‘‘Commitment’’ motive. It outlines a
cycle that begins with constant and random genetic muta-
tions that give rise to changes in brain structure and
function. These changes result in individual differences in
the strength of a reaction range in a domain-specific
dimension that produces different levels of correlated
behaviors (syndromes), which are expressed by individuals
in their environments. Continuing the cycle, Fig. 1sug-
gests a few of the many potential trade-offs that may result
from the Commitment behavioral syndrome. The individ-
ual who transfers resources to a mate and offspring has a
resources loss but may gain by increasing the viability of
offspring raised with a committed partner. Depending on
specific environmental conditions, the resource transfer
may be harmful or beneficial to the survival of the indi-
vidual’s genes in the long term and those pressures,
through balancing selection, operate in the selection of the
initial genetic variations.
However, Fig. 1is also simplified and does not include
two important influences that operate in this cycle. One is
learning (through association and reinforcement), which
can modify specific behaviors in the syndrome to better
match them to local environmental conditions. This may be
reflected in more conscious goal strategies that optimize
behavior suited to the environment, but such optimization
can also be the result of implicit learning (e.g., Amso and
Davidow 2012; Paynter et al. 2010). The other influence
not depicted in Fig. 1is environmental interaction directly
on gene expression and function (e.g., Dick 2011). This
epigenetic role for the social environment does not depend
on the long-term selection pressures of evolution and is
currently an active research area in behavioral genetics
(Szyf et al. 2010; Toyokawa et al. 2012).
Bernard et al.’s (2005a) theory allowed for the possi-
bility that a few of the motives might not be empirically
(i.e., statistically) independent. The reason, which was not
1
At the time the theory was proposed, evolutionary psychology was
somewhat hostile to individual differences; there has since been a re-
thinking of this attitude (e.g., Buss and Hawley 2011).
48 Motiv Emot (2014) 38:47–64
123

stated at the time, is that once developed in humans, the
motives could co-evolve. This co-evolution of motive
dimensions could result in small to moderate correlations
between two or more motives. This approach of positing
ultimate (evolutionary) explanations for individual differ-
ences in proximal (present day) dimensions of motive
strength may provide better understanding of the factors
contributing to contemporary purposeful human behavior.
The theory in context
Evolution and individual differences
Until recently, the psychology of individual differences has
been much more a part of personality than evolutionary
psychology (Buss 1991,1999,2009). This is because
individual differences research is the study of within
Table 1 Recurring fitness challenges, human motives, and conceptual definitions
Recurring fitness challenge Motive Conceptual definition
How to maintain organismic integrity
How to identify environmental resources and hazards
Environmental
Inquisitiveness
Use resources to explore the physical environment; evaluate resource
availability and hazards in new and different things, places, and
situations
Illness
Avoidance
Use resources to maintain bodily integrity and health
Threat
Avoidance
Use resources to maintain the safety of one’s person; avoid hazards
and challenges to one’s person and resources
How to compete for mates
How to develop social status and mating desirability;
attract a mate
Aggression Use resources to acquire and control additional resources; challenge
and intimidate others for control of resources; approach challenges
to one’s person and resources with combative and intimidating
displays and actions
Interpersonal
Inquisitiveness
Use resources to explore the social environment; test limits,
traditions, and how others act, react, and interact; compare oneself
to others
Appearance
a
Use resources to compete for status on the basis of physical
appearance
Mental
a
Use resources to compete for status on the basis of intellectual
capacities and knowledge, as well as displays of skills, abilities,
and talents
Physical
a
Use resources to compete for status on the basis of physical strength,
endurance, size, shape, and stature
Wealth
a
Use resources to compete for status on the basis of acquiring material
resources
Sex Seek sexual activity
How to establish cooperative relationships for mating
and support of offspring
Commitment
b
Transfer of resources to mates and offspring; development of tender,
intimate, supportive attachments with mate and close kin
How to develop individual-to-individual reciprocation
among kin and non-kin
Altruism
b
Transfer of resources to kin without expectation of immediate self-
benefit (but at a cost to oneself that is generally lower than the
benefits to others multiplied by their degree of genetic relatedness)
Social
Exchange
b
Enter into reciprocal, mutually beneficial exchanges of resources
with non-kin; share resources fairly and without cheating; do what
is legally and socially prescribed and avoid what is proscribed
How to develop individual-to-group-to-individual
reciprocation among non-kin (‘‘institutionalized
reciprocity’’)
Legacy
b
Transfer of resources to institutions that benefit non-kin as much as,
or more than, kin without the expectation of direct reciprocity to
oneself
Meaning Use resources to identify with, construct, and maintain a philosophy,
purpose, or rationalization for existence (and non-existence);
attempt to arrive at an understanding and peace with the presumed
purpose of life
Resources include time, energy, effort, money, and reputation. Since the theory was introduced by Bernard et al. (2005a), the motives have
been renamed to better reflect their underlying dimensions and redefined and recategorized to conform more closely to evolutionary theory; this
table reflects the theory in its current form
a
‘‘S’’ or ‘‘status motive’’ the goal of which is to increase one’s social standing through competitive behavior intended to increase personal
resources within a behavioral domain
b
‘‘C’’ or ‘‘cooperative motive’’ the goal of which is to increase one’s reputation for reliability as a partner in relationships and coalitions through
direct and indirect reciprocal transfers of resources
Motiv Emot (2014) 38:47–64 49
123

species variation along a dimension, while evolutionary
biology and psychology study invariance in a dimension to
the extent that it ultimately becomes a distinction in
adaptation between species. Theoretically, adaptations are
forged in the historical processes of selection, which
reduces variations until an adaptation becomes universal
within a species (Tooby and Cosmides 1990,1992).
Some evolutionary psychologists have proposed that a
rapprochement between individual differences and evolu-
tionary psychology might benefit both (e.g., Buss 2009)
and researchers have begun to study individual differences
in evolved traits (e.g., Figueredo et al. 2005,2007; Mac-
Donald 2005; Nettle 2006; Penke et al. 2007). Bernard
et al.’s (2005a) theory is among them.
Because most psychologists are familiar with individual
differences as applied to personality and some of the
motive labels are similar to those used in personality psy-
chology, the motives have sometimes been mistaken for
personality traits. However, the motives are conceptualized
as ‘‘traits of action.’’ Each latent motive construct repre-
sents an independently evolved, measurable, and heritable
reaction range in the strength of activity in a specific
domain (Bernard 2012). Therefore, the motives cannot be
hierarchical. The phenotypic reaction range, however, can
be calibrated (moved slightly higher or lower) depending
on environmental conditions (e.g., the availability of
resources, population size), learning, experience, and cul-
ture. Bernard et al. (2005a) proposed that a consistent
strength level in a motive dimension could be considered a
trait and such a trait should be assessed in the same manner
as personality traits.
Motivation
For much of the early history of psychology, motivation
was a vibrant area of theory and research. The first
approaches to the study of motivation may be broadly
characterized as biological. In his Principles of Psychol-
ogy, James (1890) introduced the concept of ‘‘instinct.’’ An
instinct is ‘‘…a complex, unlearned response that is now,
or was phylogenetically, adaptive and…present in all
members of a species’’ (Harriman 1977). Others used dif-
ferent terms such as ‘‘needs,’’ which activate behavior to
satisfy a physiological deficit, and ‘‘drives,’’ an internal
physiological or psychological impetus to action (Hull
1943) and ‘‘purposive strivings,’’ which emphasized the
ends toward which behavior is directed and the purpose an
instinct serves (McDougall, as cited in Boring 1950).
Eventually, the biological approach gave way to
behaviorism, which emphasized external factors in
Fig. 1 The development of motive dimensions maintained by fluctuating/balancing selection: example with the Commitment motive
50 Motiv Emot (2014) 38:47–64
123

motivation, such as the role that environmental conse-
quences play in shaping motivation. Thorndike’s (1911)
‘‘law of effect’’ focused on the consequences of behavior
(reward and punishment). Thorndike also allowed for
inferences about mind, but other behaviorists, like Watson
(1913) and Skinner (1938,1971), de-emphasized mental
processes altogether. These early biological and behavioral
approaches were criticized for being too mechanistic and
therefore interest in psychology and motivation shifted to
cognition (Nielsen and Day 1999; Tingergen 1951). Tol-
man’s (1932) cognitive-behaviorism and Bandura’s (1977)
social learning theory helped lead the way in this shift.
Some of the most active recent research areas in moti-
vation have resulted from this shift to the cognitive-beha-
vioral/social learning perspective: achievement motivation
(McClelland et al. 1976), goals (e.g., Locke and Latham
1990; Miller et al. 1960), intrinsic-extrinsic motivation
(e.g., Deci 1975; Deci and Ryan 1987), self-awareness
(Silvia and Duval 2004), self-perception (Bem 1967), and
self-regulation (e.g., Carver and Scheier 1981,1998; Oet-
tingen et al. 2004). These theories each have in common a
conception of people as conscious, rational beings whose
motivations are less determined by environmental conse-
quences and less reactive than instincts, needs, or drives.
The cognitive theories allow for more conscious motiva-
tional processes than the biological and behavioral
approaches. The cognitive perspective has been highly
heuristic for psychological research and may, at least in
part, be responsible for the reinvigoration of interest that
motivation science has experienced in recent years (e.g.,
Ryan 2012; Sorrentino and Higgins 1986; Shah and
Gardner 2008).
Holt (1931) had criticized behavioral theories of moti-
vation as being descriptive without being explanatory.
Bernard et al. (2005a) suggested the same criticism could
be made of cognitive theories, even though they have
substantial predictive power. They observed that, while the
three dominant historical paradigms—biological, behav-
ioral, and cognition—have each contributed to the under-
standing of human motivation, the field could benefit from
comprehensive, multidimensional theories that incorporate
biological, behavioral, and environmental influences, as
well as conscious and non-conscious processes. Although
research has focused on the theory’s evolutionary founda-
tion and assessment of the motives, Bernard et al. antici-
pated that future research would eventually incorporate
these additional influences.
The multidimensionality of Bernard et al. (2005a) the-
ory contrasts with most recent uni- or bi-dimensional
approaches that are primarily cognitive (e.g., self-regula-
tion, achievement motivation, and intrinsic-extrinsic
motivation). However, it is consistent with other multidi-
mensional approaches for understanding motivation.
Bernard et al. (2005a) acknowledged Cattell and his col-
leagues’ (Cattell 1957; Cattell et al. 1963) elaborate
motivational theory that resulted in development of the
Motivation Analysis Test over 50 years ago (MAT; Cattell
et al. 1964). The MAT’s 10 scales were constructed
empirically through factor analysis and the instrument was
innovative, but hampered by problems with ipsative scor-
ing of motive scales (Bernard et al. 2005b). Bernard et al.
also acknowledged a more recent multidimensional
approach, based on sensitivity theory (Reiss 2000; Reiss
and Havercamp 1996,1998). In the cognitive tradition, it
posits that universal end goals serve as fundamental
motives accounting for psychologically significant behav-
ior. Like the MAT, its motives were not specified a priori
by the researcher and thus were developed empirically
through exploratory factor analysis, resulting in the isola-
tion of 15 or 16 motives.
These multidimensional approaches employed strictly
exploratory methods without an a priori framework to
identify motive dimensions. However, without guidance by a
formal theory, factor analytic models have no logical basis
for determining how many motive dimensions there should
be, what items should comprise each dimension, and whether
the dimensions should be independent or related. By using
the problems and challenges predicted by mainstream evo-
lutionary psychology to identify the motives, Bernard et al.
(2005a) began with a logically imposed theoretical limit on
the number and types of dimensions to be considered as well
as explanations for why these particular conceptualizations
of dimensions should exist and not others.
One multidimensional measure of individual differences
that was not developed solely through an empirical factor
analytic approach is the Personality Research Form (PRF;
Jackson 1997). The PRF is relevant to motivation because
it was based on Murray’s (1938) concept of needs, which
provided a theoretical basis for its scales. The PRF Form
A/B is comprised of scales that assess: Achievement,
Affiliation, Aggression, Autonomy, Dominance, Endur-
ance, Exhibition, Harmavoidance, Impulsivity, Nurturance,
Order, Play, Social Recognition, and Understanding. Like
the MAT, the PRF assesses dimensions that can be inter-
preted as domain-specific, because they are of relatively
narrow bandwidth.
However, none of these multidimensional approaches
incorporate evolutionary theory. Cattell did suggest there
would be a genetic basis to personality and motivation
(e.g., Cattell and Dreger 1977; Cattell et al. 1963), but did
not develop it, and neither the MAT nor PRF scales were
explicitly based on evolutionary theory. In addition, in a
review of 75 years of motivation research, only two
instruments were categorized as biologically-based, and
none were identified as evolutionary-based (Mayer et al.
2007).
Motiv Emot (2014) 38:47–64 51
123

Going back to the beginning of psychology, a variety of
approaches—biological, behavioral, and cognitive—have
been used to study human motivation. Each of these has
contributed to our understanding of motivation, but there
have been few attempts to unite them in a comprehensive
manner. There have been other multidimensional approa-
ches as well, but Bernard et al. (2005a) have argued that
any such an approach would benefit by being grounded in
evolutionary theory, though not to the exclusion of other
factors such as cognition, learning, and culture. Evolu-
tionary theory provides a solid ground for motivational
theory, because it addresses the why of contemporary
behavior and places logical constraints on theory. Without
an answer to why we are left with only the description of
behavior, and description without explanation is ultimately
unsatisfying. For this reason, Bernard et al. (2005a) new
theory is important and potentially heuristic. If the renewed
interest and activity in motivation research is to have any
relevance to current directions in evolutionary psychology
and biology, it is necessary to engage in this type of
research.
Assessing individual motives
To test the theory, a reliable and valid method of assessing
the 15 motive dimensions had to be constructed. This was
done in a programmatic series of studies that began with an
iterative process of several rounds of item writing, testing,
re-writing, and re-testing that defined and refined the
dimensions psychometrically. This process resulted in what
is called the Assessment of Individual Motives-Question-
naire (AIM-Q; Bernard et al. 2008). The AIM-Q is an
umbrella term for a variety of refined versions whose
subscales all assess the same 15 motive dimensions. Each
version is based on a different item-type: AIM1 is a
200-item resource allocation version; AIM2 is a 200-item
self-description version; and AIM3 is a 15 single-item self-
endorsement version of the motive dimensions. These three
versions permitted a multitrait-multimethod (MTMM;
Campbell and Fiske 1959) analysis of convergent and
discriminant validity of the dimensions via correlations
(Bernard et al. 2008). The results were supportive of the
construct validity of the 15 motives regardless of method
used.
The AIM2 had the best overall psychometric properties
and was used, for a time, in subsequent validation studies.
However, the AIM1 resource allocation items—i.e., how
one ‘‘spends’’ one’s money, time, effort, reputation—better
fits the theoretical framework of evolutionary psychology.
After the original item analysis, only about 50 % of the
items in the AIM1 directly involved resource allocation.
Therefore, a new iterative item development process was
begun to revise the AIM1 (Bernard 2011). This process
resulted in the 199-item AIM1R, consisting of items that
adhere more to the allocation of resources. However, the
AIM1R was still rather lengthy, so further item analyses
and exploratory factor analyses led to development of a
briefer version, the 120-item B-AIM1R (Bernard 2011).
B-AIM1R scales retained their good internal consistency
reliabilities (a=.77–.93) and scale scores were strongly
correlated with scores on the AIM1R (r=.80–.98).
Research has supported the validity of the motives in
several of the AIM-Q versions. For example, ratings of
oneself and ratings of oneself by others (friends and rela-
tives) are convergent, suggesting that the motives are not
psychological artifacts, but heterogeneous dimensions that
others are able to observationally discern of participants
(Bernard 2009). Second, there should be some overlap
between individual differences in personality and motiva-
tion. Indeed, there are logical relationships between the
motives and the five factor model of personality (FFM;
Digman 1996; Costa and McCrae 1992; Goldberg 1993),
but the magnitude of these relationships is not great, sug-
gesting the motives and FFM personality traits do not
overlap substantially. Third, the motives are able to predict
self-reported behavior that was confirmed by reports of
others as well (Bernard 2009). For example, linear com-
binations of AIM2 motives accounted for reasonably large
proportions of the variance in several important behaviors
including smoking (R
Adj.
2
=.15), drinking alcohol
(R
Adj.
2
=.21), use of illegal substances (R
Adj.
2
=.17),
holding positions of leadership (R
Adj.
2
=.29), and exercis-
ing (R
Adj.
2
=.42). When this study was replicated with the
AIM1R, the motives accounted for slightly higher pro-
portions of variance in these same behaviors (Bernard
2011). Fourth, there is convergent validity between the
motives and other measures of aggression, cognition,
playfulness, and sexuality (Bernard 2007a). Finally,
although all AIM-Q items were written to be gender neu-
tral, a few observed sex differences on some motives have
been observed, which is entirely consistent with evolu-
tionary theory (Bernard 2007b).
Present studies
Prior studies of the AIM-Q have used a variety of empirical
methods, from exploratory factor analysis (EFA) to
observer ratings, to define and refine understanding of the
15 motive constructs and its corresponding measured
subscales. This body of research provided considerable
empirical evidence in support of the theory’s original
assumptions about the assessment of individual differences
in motivation, but it has not provided direct support of the
assessment of the entire measurement framework upon
which the theory rests. Prior studies of the AIM-Q have
used analytic approaches that did not offer fit indices to
52 Motiv Emot (2014) 38:47–64
123

assess all the dimensions together in an overall model. So,
up to now, the 15 independent latent motives, as postulated
by the theory, have not been evaluated together in a
comprehensive model.
Confirmatory factor analysis (CFA) is a logical and
necessary next step to demonstrate the fit of the instrument
based on the theory to data. One advantage of CFA is that it
allows for the explicit assessment, and therefore statistical
removal, of item measurement error via the estimation of
latent factors (Borsboom et al. 2003; Brown 2006).
Another benefit of CFA is that the factor structure is
hypothesized and specified in advance, as each item is
permitted to load only on its proposed factor, in contrast to
the less conservative method of EFA (Brown 2006). Prior
to this CFA, the 15 motive model lacked the critical
empirical demonstration of the connection to the Bernard
et al. (2005a) original theory. The present studies are
essential to testing whether the 15 dimensional model
adequately fits data and to determine whether the under-
lying factor structure is plausible, which should serve as the
necessary bridge to further studies to test the validity of the
theory in more complex ways.
A limitation of CFA is that it can become unwieldy
when applied to large models. One reason for this is that
the more variables (in this case, scale items) that are
included, the greater the potential for results involving
factor cross-loadings and poor factor loadings that may
cause a model to fail. Therefore, the use of CFA with the
AIM-Q had to wait for the construction and refinement of a
short and reliable scale such as the B-AIM1R. Even so, a
model with 15 dimensions and 120 variables is near the
upper limits of the procedure. In fact, we did not find a
single instance of such a large multidimensional model
tested with CFA in the literature to date.
We report the results of two studies. The first explored
the factor structure of the 120-item B-AIM1R, with the
intention of finding out if the full 15-factor model ade-
quately fits. If not, the first study would examine whether a
shorter 15-factor version would provide better fit. Study 2
then subjected the results of Study 1 to cross-validation.
Study 1
Objective
The objective of Study 1 was to determine if the 120-item
B-AIM1R, and the 15 latent dimensions it is based on,
provided the best fit to the data. If the 120-item model did
not yield adequate fit indices, items with the lowest factor
loadings would be deleted, and the fit of a simpler model
re-estimated. Thus, the purpose was to retain the most
exemplary items in each of the 15 factors, as a shorter
version of B-AIM1R items was anticipated to produce a
more parsimonious scale by omitting poorly loading items.
Offering a shorter version would encourage wider usage of
the scale because it would take less time for respondents to
complete.
Method
Sample
The sample consisted of N=1,411 participants ranging in
age from 17 to 78 years (M=29.30, SD =15.19). One-
third of the sample were university students and two-thirds
were adults living in diverse geographic regions, mostly
within California. Sixty-four percent of the sample were
female. Participants reported diverse ethnicity: 58 %
White/Euro-American (non-Hispanic), 12 % Latino, 11 %
Asian American, 9 % Mixed (two or more ethnicities), 4 %
African American, and 6 % other.
Instrument
The previously described 120-item B-AIM1R (Bernard
2011) was used in Study 1. Participants endorsed items that
concerned the allocation of personal resources (e.g.,money,
time, effort, reputation) consistent with evolutionary theory
as being personally true or untrue using this 6-point Likert
scale: 1 (Completely untrue), 2 (Mostly untrue), 3 (Slightly
untrue), 4 (Slightly true), 5(Mostly true), and 6 (Completely
true). Examples of items are listed in Table 2.
Procedure
University students participated in return for partial course
credit. Participants met in small groups with a research
assistant who explained the procedure, requested informed
consent, and ensured that they created an appropriate per-
sonal identification number (PIN). The PIN was used to
track participation for credit, while keeping responses
anonymous. Students were provided with a URL and given
instructions to complete the B-AIM1R online anonymously
at their own convenience within 24–48 h. Each student
participant was also instructed to recruit two non-students
who were ‘‘…ideally, one relative (parent, grandparent,
aunt, sibling, etc.) and one acquaintance (friend, neighbor,
teacher, coach, etc.) who is 25–80 years of age.’’ The age
limits were provided to suggest an extensive range. This
type of sampling procedure, often used in sociological
research, provides an efficient and reliable way to obtain
samples of participants who are otherwise missed in lab-
oratory studies (Heckathorn 1997). This procedure has
been used throughout the development of the AIM-Q
Motiv Emot (2014) 38:47–64 53
123

Table 2 The 60 item scale
Factor Item Statement
F1 V1 I make the effort to explore something I have not seen before
V2 I am curious about new and different places because they may have something to offer
V3 I explore (or have explored) unknown places to understand them better
V4 I am interested in going to different places in order to learn something that could help me
F2 V5 I wash my hands before I eat anything, even a quick snack
V6 I am very careful not to touch any surfaces after I wash my hands in a public place
V7 I try to avoid any contact with people when there is an illness going around
V8 I avoid traveling when infectious diseases are circulating
F3 V9 I spend a lot of effort to avoid physical threats of any kind
V10 I stay clear of people who are engaged in a conflict situation for fear of harm
V11 I leave a situation when people become hostile toward each other
V12 I leave when a fight breaks out between people in order to stay safe
F4 V13 I try to intimidate others to keep them from challenging me
V14 I use physical threats so that others know not to get in my way
V15 I use verbal threats so that others know not to get in my way
V16 I take physical risks so others will fear me
F5 V17 I play with, jest, or mock people to discover their reactions
V18 I tease friends and acquaintances because it helps me understand them better
V19 I like to play around with people to see how they react, even if they get angry at times
V20 I try to understand others by playfully teasing them to see how they react
F6 V21 I spend money and time on stylish hair cuts to improve my appearance
V22 I spend money or time to improve my appearance such as getting a tan, spa treatments, or other similar things
V23 I buy expensive shoes or other accessories to look attractive
V24 I spend money on grooming (hair cuts, coloring, tanning) to look my best
F7 V25 I show off my understanding of abstract or complex ideas so people will respect me
V26 I engage in discussions or debates to develop a reputation as a smart and influential person
V27 I put a lot of effort into developing my own special skills and abilities to impress others
V28 I put effort into developing my talents and abilities in order to impress others
F8 V29 I train hard to have as strong a body as I can
V30 I make the effort to stay in peak physical condition compared to others
V31 I work on my body to be as physically competitive as I can
V32 I train hard to have the reputation of being in peak physical condition
F9 V33 I buy (or would buy) expensive or rare luxury goods so that others will know I am successful
V34 I like to have valuable things like cars, watches, or jewelry that other people notice
V35 I put a lot of time and energy into having expensive possessions
V36 I spend a lot of money to buy things that make others envious
F10 V37 I spend time thinking about how to have more sex
V38 I spend (or have spent) a lot of time and energy to have sex
V39 I invest (or have invested) time seeking a lot of different sexual partners
V40 I spend (or have spent) a lot of time pursuing sex
F11 V41 I take time to offer emotional support and affection to the person I am involved with romantically
V42 I make the effort to do the things that show I care for my spouse or partner
V43 I give of myself and my time in a romantic relationship
V44 I give what is necessary of my time and resources in romantic relationships
F12 V45 I sacrifice (or would sacrifice) to help my distant relatives
V46 I would let (or have let) distant relatives live with me temporarily if they needed shelter
V47 I would take slight personal risks to my safety to help distant relatives
V48 I would invest time and effort to help a distant relative even if there were no immediate advantage to me
54 Motiv Emot (2014) 38:47–64
123

versions. Students provided their recruits with a PIN and
the URL to access the survey online at their convenience.
A potential concern for this sampling procedure was that
the student participants, rather than recruit someone, might
complete the survey three times themselves. However, the
length of the survey seemed to discourage this practice.
Post-experimental debriefings were conducted on partici-
pants who reported little difficulty in recruiting acquain-
tances who participate out of social obligation. None
reported faking in anonymous follow-ups. Nevertheless, to
reduce any possibility of faking, participants were asked to
email the researcher if they could not obtain two recruits,
but still wished to receive credit. Two did report difficulty
and were granted full credit.
The fact that participants can complete the B-AIM1R
online at their own convenience probably improves com-
pliance. It also allows students to obtain recruits without
any geographic restrictions. Technical difficulties such as
dropped sessions or failure to complete a session have
occurred rarely, affecting \3 % in prior studies of the
Assessment of Individual Motives-Questionnaire (Bernard
et al. 2008). No such problems were encountered in this
sample.
Analysis
The EQS 6.2 software (Bentler 2005) was used to specify
and estimate the confirmatory factor analyses. As the study
sought to derive a factor structure that optimally repre-
sented the underlying data, maximum-likelihood served as
the method of estimation. For the purpose of model iden-
tification, the scale of each latent factor was set to 1 and
item measurement errors were estimated (Ullman 2007). In
the initial 120-item model, the 15 motive factors were
specified to be inter-correlated, with each factor tapped by
its 8 corresponding items from the B-AIM1R. If a
reasonable fitting model did not result, only items with the
highest factor loadings would be retained, and the simpler
15-factor model would be re-estimated. The fit of four
alternative models would also be estimated to rule out the
plausibility of these competing frameworks.
Pre-analysis of the data indicated the 120 individual
items did not depart drastically from a normal distribution,
with reasonable levels of skewness (-1.91 to 1.98). Even
with departures from normality, the sizable sample of the
study should produce appropriate standard error estimates
based on the central limit theorem (Tabachnick and Fidell
2007). Goodness of fit of the model to the sample data was
evaluated using several indices recommended by Byrne
(2006). The model v
2
statistic should be non-significant,
but this test is prone to erroneously rejecting a correct
model if the sample size is large (Bollen 1989). Also
inspected were the Comparative Fit Index (CFI) and
Incremental Fit Index (IFI), which compare the hypothe-
sized model against the independence model while taking
into account sample size (Byrne 2006). The values of these
two fit indices could range from 0 to 1.00, with higher
values representing greater fit to the data (Ullman and
Bentler 2003). Values of .90 or higher indicate good fit.
The Root Mean-Square Error of Approximation (RMSEA)
considers the extent of misfit between the hypothesized
model and the observed data. The RMSEA is adequately
sensitive in detecting model misspecification, yields
appropriate information about model quality, and has the
additional advantage that confidence intervals can be built
around RMSEA values (MacCallum and Austin 2000).
Values below .05 indicate close fit, between .05 and .08
indicate fair fit, between .08 and .10 indicate mediocre fit,
and above .10 indicate poor fit (MacCallum et al. 1996).
Hu and Bentler (1998) found that the Standardized Root
Mean-Square Residual (SRMR) is useful in detecting
model misspecification and suggested that values below .08
Table 2 continued
Factor Item Statement
F13 V49 I take care of all my responsibilities to others
V50 I do what is required to keep a contract or bargain I have made
V51 I strive to be honest and fair
V52 I make an effort to treat others the way I want to be treated
F14 V53 I contribute to charitable organizations that do good work in the world
V54 I donate my time, effort, or money to causes that promote health, reduce famine, or improve education in other countries
V55 I donate my time, effort, or money to non-profit community organizations that can make a better future for others
V56 I contribute time, effort, and/or money to make the world better for future generations
F15 V57 I put a lot of effort into investigating different philosophies and approaches to life
V58 I take time to investigate different religions and philosophies
V59 I learn about various religions or philosophies to understand the purpose of life better
V60 I study various philosophies of life and its meaning in order to find out what is common in all
Motiv Emot (2014) 38:47–64 55
123

Table 3 Study 1: item loadings, means, and standard deviations, of the 60 item scale
Factor Item Factor loading SD
Standardized
coefficient
Ztest Mean
F1 Environmental Inquisitiveness V1 .75 30.82 4.36 1.13
V2 .77 31.65 4.55 1.23
V3 .77 31.86 4.17 1.31
V4 .72 28.90 4.30 1.33
F2 Illness Avoidance V5 .57 19.94 3.54 1.52
V6 .64 22.64 3.14 1.54
V7 .67 23.91 3.54 1.36
V8 .61 21.32 3.48 1.49
F3 Threat Avoidance V9 .52 18.23 4.15 1.36
V10 .61 21.80 3.75 1.35
V11 .63 22.65 4.08 1.26
V12 .70 25.53 3.90 1.40
F4 Aggression V13 .71 29.10 2.51 1.41
V14 .81 34.94 1.94 1.30
V15 .81 35.25 2.15 1.39
V16 .79 33.79 2.00 1.33
F5 Interpersonal Inquisitiveness V17 .80 34.63 3.12 1.50
V18 .81 35.22 3.13 1.52
V19 .84 37.27 2.99 1.49
V20 .84 37.52 3.15 1.51
F6 Appearance V21 .75 30.95 3.15 1.61
V22 .73 29.72 2.94 1.64
V23 .73 29.94 3.12 1.56
V24 .76 31.76 3.52 1.54
F7 Mental V25 .67 26.72 2.99 1.40
V26 .69 28.09 3.08 1.40
V27 .80 34.32 3.49 1.43
V28 .80 34.30 3.29 1.49
F8 Physical V29 .82 36.53 3.43 1.55
V30 .85 38.91 3.44 1.49
V31 .87 40.16 3.29 1.58
V32 .86 39.70 3.01 1.60
F9 Wealth V33 .78 34.17 2.49 1.48
V34 .78 33.86 3.10 1.59
V35 .83 37.51 2.75 1.47
V36 .84 38.06 2.39 1.41
F10 Sex V37 .73 30.49 2.63 1.59
V38 .76 32.33 2.60 1.63
V39 .77 33.20 1.93 1.44
V40 .90 41.70 2.23 1.56
F11 Commitment V41 .73 30.46 5.05 1.17
V42 .77 32.29 4.87 1.23
V43 .84 36.92 4.70 1.34
V44 .80 34.48 4.82 1.24
F12 Altruism V45 .65 24.76 4.30 1.25
V46 .64 24.03 4.68 1.30
56 Motiv Emot (2014) 38:47–64
123

are diagnostic of good fit. Given the large sample size and
the number of statistical analyses, a conservative p\.001
was used in all tests.
Results and discussion
The 120-item scale, consisting of 15 correlated factors,
each represented with 8 items, yielded a model with mixed
fit indices, v
2
=28,363.71, df =6,916, p\.001,
CFI =.79, IFI =.79, although RMSEA =.047 [90 % CI
.046–.047] and SRMR =.068. Factor loadings ranged
from .39 to .84, with 22.5 % (27 of 120) of these values
loading below .60, indicating that not all items adequately
tapped the hypothesized dimensions. Inter-factor correla-
tions ranged from -.56 to .69. Scrutiny of these results
using the Lagrange multiplier test (Chou and Bentler 1990)
to determine potential modifications that might improve the
model, the relatively poor fit stemmed from a combination
of unacceptable item loadings and items that could load
across several factors. Items with poor loadings on
hypothesized dimensions were most likely to cross-load
across factors. Consistent with our initial concerns, the
model contained too many items.
Consequently, the measurement model was refined to
produce internally reliable and purer subscales uncontami-
nated by less representative items. This was accomplished
systematically by retaining only the four items with the
highest standardized loadings in each of the hypothesized
factors. Because internal consistency via Cronbach’s alpha is
a function of both strength of inter-item correlations and the
number of items representing the factor, this ensured that
factors were balanced with an equal number of items (Crano
and Brewer 2002). This procedure to produce 15 correlated
factors with the most representative content yielded a final
60-item scale (Table 2) that showed good fit indices overall,
v
2
=5,385.68, df =1,605, p\.001, CFI =.91,
IFI =.92, RMSEA =.041 [90 % CI .040–.042],
SRMR =.046. As presented in Table 3, all items were
found to be significantly indicative of their factors, with
standardized loadings ranging from .52 to .90, all p\.001.
Only 3.3 % (2 of 60) of the items loaded below .60.
Next, discriminant validity was assessed to demonstrate
that none of the 15 factors were isomorphic, although they
were all expected to be somewhat correlated as the factors
were designed to capture an evolutionary model of human
motivation. An inter-factor correlation below an absolute
value of .80 has been recommended as a cut-off to dem-
onstrate that two factors are conceptually distinct (Brown
2006; Mahoney et al. 1995). Based on this criterion, dis-
criminant validity was shown across all possible pairs of
factors, with inter-factor correlations ranging from -.59 to
.69 (Table 4). Furthermore, constraints were imposed on
the CFA by forcing every combination of each pair of
factors, in separate analyses, to be perfectly correlated
(r=1.0). These tests revealed that none of the constraints
were tenable, suggesting that all 15 factors were statisti-
cally non-identical, all ps \.001. For each subscale, the
computed average of items, standard deviation, and inter-
nal consistency reliability are presented in Table 5. Reli-
abilities indicated that the items within each subscale were
Table 3 continued
Factor Item Factor loading SD
Standardized
coefficient
Ztest Mean
V47 .67 25.63 4.01 1.33
V48 .78 30.51 4.33 1.22
F13 Social Exchange V49 .67 26.69 4.86 1.00
V50 .65 25.59 4.97 0.98
V51 .75 30.94 5.20 1.01
V52 .73 29.55 5.16 0.98
F14 Legacy V53 .81 34.65 4.19 1.44
V54 .71 28.76 3.61 1.51
V55 .83 36.01 4.02 1.52
V56 .72 29.77 4.14 1.31
F15 Meaning V57 .78 33.59 3.60 1.43
V58 .82 36.57 3.49 1.45
V59 .87 39.47 3.43 1.46
V60 .83 36.91 3.35 1.50
All factor loadings are significant at p\.001
Motiv Emot (2014) 38:47–64 57
123

cohesively related (alphas of .71–.91), despite the relatively
short 4-item subscale lengths.
Comparison of the 60-item and 120-item versions is
shown in Table 5. Subscales using the 60-item version
exhibited only a negligible or minor decrement in reliability
in comparison to subscales computed using the 120-item
version. Each subscale across formats showed strong cor-
respondence with correlations of .87 or higher, all ps\.001.
Taken together, these patterns of parallel findings across
versions suggest that the briefer instrument yielded similar
properties to the longer instrument.
The credibility of four alternative models was evaluated
using the final 60-item scale. First tested was a completely
independent CFA model, specified by not allowing any of the
15 latent factors to be correlated. Results show that this com-
pletely orthogonal 15-factor structure was unsupported by the
Table 4 Study 1: inter-factor correlations of the 60-item scale
Factor F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15
F1 Environmental Inquisitiveness
F2 Illness Avoidance .04
F3 Threat Avoidance .12 .43
F4 Aggression -.07 .12 -.28
F5 Interpersonal Inquisitiveness .15 -.07 -.15 .51
F6 Appearance .05 .19 .26 .13 .02
F7 Mental .23 .05 .01 .50 .52 .31
F8 Physical .12 .14 -.05 .37 .23 .25 .43
F9 Wealth -.01 .12 -.01 .56 .36 .67 .69 .36
F10 Sex .01 -.02 -.26 .52 .39 .15 .33 .27 .41
F11 Commitment .28 -.01 .16 -.27 -.08 .11 .03 .08 -.07 -.06
F12 Altruism .31 -.06 .02 -.12 .05 .01 .07 .07 -.06 -.10 .36
F13 Social Exchange .37 .01 .34 -.59 -.23 -.06 -.17 -.08 -.33 -.41 .62 .45
F14 Legacy .39 .12 .20 -.14 -.14 .14 -.02 .06 -.12 -.21 .26 .34 .42
F15 Meaning .53 .05 .04 .11 .15 .03 .27 .14 .01 .14 .18 .21 .14 .38
Correlations of at least |.11| are significant, p\.001
Correlations are taken from the confirmatory factor analysis
Table 5 Study 1: subscale means, standard deviations, and reliablities of the 60-item and 120-item scales
Factor 60-item scale 120-item scale Subscale correlations
of 60-item and
120-item versions
Mean SD aMean SD a
F1 Environmental Inquisitiveness 4.35 1.02 .84 4.25 0.93 .89 .95
F2 Illness Avoidance 3.43 1.09 .72 3.63 0.88 .77 .88
F3 Threat Avoidance 3.97 0.98 .71 4.13 0.79 .77 .92
F4 Aggression 2.15 1.13 .86 2.63 0.95 .84 .88
F5 Interpersonal Inquisitiveness 3.10 1.31 .89 3.00 1.14 .90 .94
F6 Appearance 3.18 1.29 .83 3.25 1.13 .88 .94
F7 Mental 3.21 1.16 .83 3.25 1.08 .89 .95
F8 Physical 3.29 1.39 .91 3.63 1.29 .94 .97
F9 Wealth 2.68 1.28 .88 2.75 1.24 .93 .96
F10 Sex 2.35 1.32 .87 1.75 1.11 .91 .95
F11 Commitment 4.86 1.05 .86 5.13 0.87 .89 .94
F12 Altruism 4.33 0.99 .78 4.38 0.88 .84 .91
F13 Social Exchange 5.05 0.78 .80 5.00 0.73 .83 .87
F14 Legacy 3.99 1.20 .85 4.00 1.05 .88 .95
F15 Meaning 3.47 1.27 .89 3.75 1.18 .93 .95
a=Cronbach’s alpha. All correlations significant at p\.001
58 Motiv Emot (2014) 38:47–64
123

data, v
2
=11,131.51, df =1,710, p\.001, CFI =.79,
IFI =.79, RMSEA =.063 [90 % CI .061–.064],
SRMR =.158. The correlated five-factor model, combining
the items of the dimensions within each of the five recurring
fitness challenges (as described in Table 1) as factors instead,
produced unacceptable fit indices, v
2
=2,398.06, df =1,700,
p\.001, CFI =.52, IFI =.52, RMSEA =.094 [90 % CI
.093–.096], SRMR =.094. The uncorrelated five-factor
model of domains was also implausible, v
2
=24,347.02,
df =1,710, p\.001, CFI =.49, IFI =.49, RMSEA =.097
[90 % CI .096–.098], SRMR =.131. Next, a one-factor CFA
model was estimated to rule out the possibility of an over-
arching dimension that fully captured and explicated the the-
oretical framework. This particular model, stipulating the
absence of subscale discriminant validity, was specified by
forcing all items to load on a unitary latent factor. Results show
that the one-factor structure was untenable and poorly
approximated the data, v
2
=34,338.40, df =1,710,
p\.001, CFI =.26, IFI =.26, RMSEA =.116 [90 % CI
.115–.117], SRMR =.142. v
2
difference tests found that the
correlated 15-factor model yielded a significantly better fit over
all four alternative models, all ps\.001.
Study 2
The results of Study 1 provided preliminary support for the
15 factor model consisting of 60 items. However, given the
Table 6 Study 2: item loadings, subscale means, standard deviations,
and reliabilities of the 60-item scale
Factor Item Factor loading SD
Standardized
coefficient
Ztest Mean
F1 Environmental
Inquisitiveness
V1 .70 16.05 4.42 1.10
V2 .69 15.71 4.62 1.09
V3 .73 17.17 4.10 1.29
V4 .70 16.06 4.46 1.24
F2 Illness
Avoidance
V5 .62 12.60 3.34 1.55
V6 .66 13.46 3.31 1.55
V7 .57 11.60 3.58 1.39
V8 .57 11.47 3.55 1.57
F3 Threat
Avoidance
V9 .39 7.81 4.03 1.45
V10 .64 13.60 3.90 1.39
V11 .61 12.90 4.07 1.31
V12 .78 17.16 3.90 1.46
F4 Aggression V13 .74 18.40 2.62 1.43
V14 .82 21.20 2.05 1.40
V15 .78 19.66 2.25 1.47
V16 .81 20.82 2.14 1.33
F5 Interpersonal
Inquisitiveness
V17 .77 19.36 3.17 1.57
V18 .79 19.90 3.30 1.49
V19 .84 21.76 3.19 1.53
V20 .79 19.83 3.42 1.49
F6 Appearance V21 .78 19.65 3.36 1.67
V22 .75 18.35 2.86 1.68
V23 .70 16.77 3.32 1.59
V24 .83 21.13 3.57 1.57
F7 Mental V25 .58 13.23 3.23 1.35
V26 .69 16.29 3.36 1.38
V27 .81 20.17 3.59 1.45
V28 .80 20.11 3.43 1.44
F8 Physical V29 .80 20.95 3.61 1.51
V30 .84 22.50 3.63 1.51
V31 .88 23.98 3.44 1.66
V32 .85 22.76 3.23 1.65
F9 Wealth V33 .74 18.45 2.86 1.50
V34 .76 19.29 3.48 1.58
V35 .87 23.38 2.86 1.51
V36 .81 21.20 2.54 1.41
F10 Sex V37 .70 17.01 2.58 1.55
V38 .76 18.87 2.60 1.64
V39 .80 20.34 1.93 1.44
V40 .90 24.45 2.20 1.54
F11 Commitment V41 .75 18.60 5.06 1.21
V42 .80 20.48 4.99 1.13
Table 6 continued
Factor Item Factor loading SD
Standardized
coefficient
Ztest Mean
V43 .81 20.88 4.93 1.12
V44 .87 23.36 4.85 1.24
F12 Altruism V45 .72 17.02 4.38 1.26
V46 .62 13.95 4.80 1.23
V47 .75 17.95 4.21 1.26
V48 .83 20.52 4.39 1.24
F13 Social Exchange V49 .62 13.93 5.00 0.88
V50 .66 15.06 5.04 0.93
V51 .73 17.28 5.28 0.93
V52 .68 15.73 5.27 0.93
F14 Legacy V53 .79 19.33 4.16 1.33
V54 .72 17.17 3.66 1.44
V55 .74 17.76 3.75 1.50
V56 .77 18.96 3.93 1.32
F15 Meaning V57 .77 19.31 3.66 1.42
V58 .81 21.10 3.51 1.47
V59 .86 23.12 3.50 1.46
V60 .83 21.61 3.30 1.44
Motiv Emot (2014) 38:47–64 59
123

systematic truncation of items from the scale based on
empirical tests, further corroboration of the revised factor
structure suggested by the theory is necessary. Therefore, the
final 60-item model resulting from Study 1 was subjected to a
cross-validation by applying the CFA to another sample. If
the structure of the model was cross-validated in Study 2,
additional support would be offered for both the underlying
structure of the revised theory of evolved human motivation
and the adequacy of the 60 B-AIM1R items.
Method
Sample
The sample consisted of N=490 participants ranging in age
from18to78 years(M=27.96, SD =13.50). One-third of the
sample were university students and two-thirds were adults liv-
ing in diverse geographic regions, mostly within California.
Fifty-six percent of the sample were female. Participants
reported diverse ethnicity: 52 % White/Euro-American (non-
Hispanic), 14 % Asian American, 12 % Latino, 11 % mixed
(two or more ethnicities), 7 % African American, and 4 % other.
Instrument
The final set of 60 items isolated in Study 1 were admin-
istered in Study 2. The same Likert scale was used for all
the items, and the instrument was completed online.
Procedure
The same acquaintance network sampling procedure used
in Study 1 was employed to obtain the sample for Study 2.
No participants reported problems obtaining recruits and
no dropped online sessions or failures to complete were
observed.
Analysis
The 60 individual items did not depart drastically from a
normal distribution, as evidenced by reasonable skewness
levels (-2.07 to 2.02). EQS 6.2 (Bentler 2005) using
maximum-likelihood was again used to specify the con-
firmatory factor analyses. The scale of each latent factor
was set to 1 for identification, and item measurement errors
were estimated. The main CFA allowed the 15 motive
factors to be correlated, and each factor was tapped with
their 4 corresponding items as presented in Table 2. The
same fit indices were used to evaluate the models.
Results and discussion
Confirmatory factor analysis of the 60-item scale involving
the correlated 15-factor model yielded a model with sat-
isfactory fit indices, v
2
=3,183.66, df =1,605, p\.001,
CFI =.90, IFI =.90, RMSEA =.045 [90 % CI .043–
.047], SRMR =.053. Standardized factor loadings ranged
Table 7 Study 2: inter-factor correlations of the 60-item scale
Factor F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15
F1 Environmental
Inquisitiveness
–
F2 Illness Avoidance -.01 –
F3 Threat Avoidance -.02 .43 –
F4 Aggression -.07 .03 -.33 –
F5 Interpersonal Inquisitiveness .13 -.10 -.20 .56 –
F6 Appearance .06 .17 .23 .20 .03 –
F7 Mental .20 -.02 -.09 .52 .46 .34 –
F8 Physical .10 .09 -.07 .46 .30 .29 .60 –
F9 Wealth .02 .07 .04 .54 .34 .70 .63 .48 –
F10 Sex .05 -.21 -.26 .56 .40 .19 .40 .31 .39 –
F11 Commitment .38 .04 .11 -.28 -.03 .17 .03 .07 .00 -.07 –
F12 Altruism .36 .15 .02 -.13 .00 -.01 .13 .18 -.08 .00 .34 –
F13 Social Exchange .40 .11 .23 -.57 -.31 -.12 -.13 -.10 -.31 -.40 .54 .47 –
F14 Legacy .35 .28 .23 -.10 -.19 .16 .02 .04 -.04 -.16 .22 .29 .31 –
F15 Meaning .50 .09 .02 .12 .12 .05 .24 .24 .05 .13 .20 .26 .14 .42 –
Correlations of at least |.19| are significant, p\.001
Correlations are taken from the cross-validation confirmatory factor analysis
60 Motiv Emot (2014) 38:47–64
123

from .39 to .90, and were all significant, p\.001. The
majority of factor loadings were above acceptable values,
with only 6.7 % (4 of 60) of the items producing loadings
below .60 (see Table 6). Inter-factor correlations of the
CFA ranged from -.57 to .70 (see Table 7), all of which
remained below the .80 guideline recommended to deter-
mine factor discriminant validity (Brown 2006; Mahoney
et al. 1995). Tests of constraints, in which pairs of factors
were forced to be perfectly correlated, offered additional
evidence that none of the dimensions were statistically
isomorphic in sharing the same variance.
With the 60 items, the alternative models described in
Study 1 were undertaken for corroboration using this new
sample. The fully orthogonal 15-factor model, not allowing
factors to be correlated, was discovered to be poorly fitting,
v
2
=5,218.38, df =1,710, p\.001, CFI =.77, IFI =.77,
RMSEA =.065 [90 % CI .063–.067], SRMR =.165. Of the
recurring fitness domains, the correlated five-factor model
produced unacceptable fit indices, v
2
=8,934.81, df =
1,700, p\.001, CFI =.53, IFI =.53, RMSEA =.093
[90 % CI .091–.095], SRMR =.095, and so did the uncor-
related five-factor model, v
2
=9,281.33, df =1,710,
p\.001, CFI =.50, IFI =.51, RMSEA =.095 [90 % CI
.093–.097], SRMR =.129. The one-factor model was also
judged to be inferior in fit, v
2
=12,799.08, df =1,710,
p\.001, CFI =.27, IFI =.28, RMSEA =.115 [90 % CI
.113–.117], SRMR =.143. v
2
difference tests corroborated
that the oblique modelproduced a significant improvement in
fit over all alternative models, all ps\.001.
Finally, although the variables in both studies did not
depart drastically from a normal distribution, robust esti-
mation was also performed on the final 60-item scale, to
evaluate whether the fit of the models remained relatively
stable using this approach. Robust estimation, requested
through the EQS program, is considered the most
appropriate methodology in SEM if the assumption of
normality is violated, as it corrects for the extent of such
deviations in the model, but results should parallel that of
the normal theory approach if normality is largely satisfied
(Bentler 2005; Satorra and Bentler 1994). The models
analyzed with robust fit indices are presented in Table 8.
The fit indices for the models with 15 correlated factors in
Study 1 and 2 remain satisfactory with this correction, but
the alternative models were still found to be implausible
even with the adjustment. Considered together, these
findings provided strong cross-validation evidence in sup-
port of the correlated 15-factor model of 60 items.
General discussion
The objectives of the present study have been met. The
results of the CFA and the CFA cross-validation supported
both the underlying structure of Bernard et al.’s (2005a)
theory of motivation and the adequacy of the final, reduced
60-item Quick-AIM version used to measure individual
differences in motivation in 15 dimensions. Given the rig-
orous requirements of CFA and the relatively large sample
size, these results provide, to date, the best evidence in
support of Bernard et al. (2005a) evolutionary theory of
human motivation as revised (Bernard 2012). Thus, a crucial
step in support of the theory has been accomplished.
The original 15 dimension model, allowing for corre-
lations between dimensions, appears to fit the data sub-
stantially better than both the completely independent
model with none of the 15 dimensions correlated and the
unidimensional model. Consistent with the theory and its
revision, these results suggest that, while the 15 hypothe-
sized motive dimensions are largely independent, with
predominantly negligible and small correlations, moderate
Table 8 Robust fit indices for 60-item CFA
Model Model v
2
df p CFI IFI RMSEA [90 % CI]
Study 1
15 factors (oblique) 4,308.19 1,605 \.001 .93 .93 .035 [.033, .036]
15 factors (orthogonal) 8,986.48 1,710 \.001 .81 .81 .055 [.054, .056]
5 factor (oblique) 19,815.96 1,700 \.001 .53 .53 .087 [.086, .088]
5 factor (orthogonal) 20,851.72 1,710 \.001 .50 .48 .089 [.088, .090]
1 factor 29,099.67 1,710 \.001 .29 .29 .107 [.105, .108]
Study 2
15 factors (oblique) 2,701.65 1,605 \.001 .92 .92 .037 [.035, .040]
15 factors (orthogonal) 4,449.91 1,710 \.001 .79 .79 .057 [.055, .059]
5 factor (oblique) 7,934.22 1,700 \.001 .53 .53 .087 [.085, .088]
5 factor (orthogonal) 8,236.62 1,710 \.001 .51 .51 .088 [.086, .090]
1 factor 11,451.37 1,710 \.001 .29 .29 .106 [.104, .107]
SRMR not reported, as this fit index is not available using robust estimation (Bentler 2005). Scaled v
2
difference tests supported that the four
alternative models were significantly inferior to the oblique 15-factor model, all ps\.001
Motiv Emot (2014) 38:47–64 61
123

correlations remain between some of them. The factor
intercorrelations listed in Table 4(Study 1) and Table 7
(Study 2) are remarkably similar and suggest no strong
([.70) correlations between the dimensions. These corre-
lations that remain are logically consistent with respect to
their underlying constructs. For example, theoretically,
each of the competitive/status motives represents an inde-
pendently evolved strategy to address the same challenge
of obtaining a suitable mate. Likewise, each of the coop-
erative motives represents an independently evolved strat-
egy to address the challenge of pair-bonding and coalition
formation. Therefore, it is logical that the competitive/
status motives (Appearance, Mental, Physical, and Wealth)
are moderately correlated and mostly correlated with
Aggression, while the cooperative motives (Commitment,
Altruism, Social Exchange, and Legacy) are also moder-
ately intercorrelated. Two other correlations are also worth
noting: Threat Avoidance and Illness Avoidance (r=.43
and .43 in Study 1 and Study 2, respectively) and
Aggression and Social Exchange (r=-.59 and -.57 in
Study 1 and Study 2, respectively). These relationships are
also logical and suggest interesting directions for future
study.
It is not surprising that the 60-item Quick AIM fits the
data better than the 120-item B-AIM1R. A model with 15
dimensions is rather large for a CFA. It is quite difficult to
obtain good fit for a multi-dimensional model of this scope
and 120 items posed a greater challenge for good model fit.
Indeed, the present results produced both support for the
theoretical model and for more parsimonious subscales to
assess its dimensions. Although a satisfactory result, it also
raises the issue of the adequacy of systematically sampling
from a dimension—i.e., content validity, which is not able
to be assessed in any formal way. Are four highly inter-
correlated items sufficient to fully represent a dimension of
behavior? The answer partly depends on the conceptual
bandwidth of the dimension being considered, and the
present dimensions are theoretically domain-specific, and
of relatively narrow bandwidth. Perhaps, then, the 60-item
version is sufficient, because it consists of the items with
the highest factor loadings, which may better tap the core
conceptual elements in the dimensions.
Two important limitations of the present line of research
must be taken into account. First, the evolved nature of the
dimensions on which the theory is based is challenging to
study empirically precisely because it is in the past.
However, hypotheses in evolutionary psychology can and
should be investigated empirically (Confer et al. 2010).
The reliable and valid dimensions represented in the AIM-
Q’s current version of most refined scales should foster
experimental research. Early returns from such research
suggest that some of the motives, when manipulated, can
produce changes in the judgment of the attractiveness of
others, an important indicator of fitness. A second limita-
tion is that purposeful human behavior is a complex phe-
nomenon that is undoubtedly the result of many factors
operating simultaneously. The putative evolved motives
represented in the AIM-Q may reflect ultimate causal
factors that operate in human motivation today, but there
must also be room for learning, environmental cues,
incentives, goals, and other factors.
There are some other uses for this new Quick AIM
version of the AIM-Q. First, Bernard et al. (2005a) original
theory tentatively proposed neuropsychological circuits
that might be involved in mediating the 15 motive
dimensions. Therefore, the ability to measure the strengths
of motives could be of use in brain imaging studies. The
strength in a motive dimension should be related to the
activity level in the theory’s proposed neural circuits.
Theoretically, this could work both ways, also allowing
neuroscientists to match certain patterns of brain activity to
the 15 dimensions of motivation. Second, the AIM-Q may
be helpful to counselors and clinicians. We concur with
Fiske (2008) that the study of individual differences in
motivation is an integral part of personality psychology.
However, prior research suggests that the AIM-Q motives
are conceptually different from personality dimensions and
do not substantially overlap with the five factor model of
personality (Bernard 2010). Therefore, counselors may
wish to use the AIM-Q to assess people’s motivation, as
well as their personality, in order to better understand the
‘‘ …dynamics of action, the forces that move people to do
what they do…’’ (McAdams and Pals 2007). Third, the
AIM-Q permits psychologists to take the visceral, not
entirely conscious, processes that presumably derived from
human evolutionary history into account when studying
more conscious proximal causes of motivated behavior.
Bernard et al. (2005a) did not claim that these 15 are the
definitive motive dimensions. They proposed them as a
way to unite three vibrant areas of psychological science,
the two historical fields of motivation and individual dif-
ferences, and the newer field of evolutionary psychology,
in research. This approach provides a way to assess
motives and link them to other phenomena that are of
interest to researchers in diverse areas of the field. This
may result in a broader application of motivation science to
expanded areas of contemporary research.
Acknowledgments The authors thank Aaron Lukaszewski for his
valuable assistance and suggestions in the preparation of this
manuscript.
References
Amso, D., & Davidow, J. (2012). The development of implicit
learning from infancy to adulthood: Item frequencies, relations,
62 Motiv Emot (2014) 38:47–64
123

and cognitive flexibility. Developmental Psychology, 54,
664–673.
Bandura, A. (1977). Social learning theory. Oxford: Prentice-Hall.
Bem, D. (1967). Self-perception: An alternative explanation of
cognitive dissonance phenomena. Psychological Review, 74,
183–200.
Bentler, P. M. (2005). EQS 6 structural equations program manual.
Encino, CA: Multivariate Software (www.mvsoft.com).
Bernard, L. C. (2007a). Assessing individual differences in motiva-
tion: Convergent validity of the Assessment of Individual
Motives-Questionnaire [AIM-Q] and measures of aggression,
cognition, playfulness, and sexuality. Individual Differences
Research, 5, 158–174.
Bernard, L. C. (2007b). Sex and motivation: Differences in evolu-
tionary psychology-based motives. In P. W. O’Neal (Ed.),
Motivation of health behavior (pp. 65–84). Hauppauge, NY:
Nova Science Publishers.
Bernard, L. C. (2009). Consensual and behavioral validity of a
measure of adaptive individual differences dimensions in human
motivation. Motivation and Emotion, 33, 303–319.
Bernard, L. C. (2010). Motivation and personality: Relationships
between putative motive dimensions and the five factor model of
personality. Psychological Reports, 106, 613–631.
Bernard, L. C. (2011). Assessment of individual motives—Question-
naire manual. Retrieved from http://myweb.lmu.edu/lbernard.
Bernard, L. C. (2012). Evolved individual differences in human
motivation. In R. Ryan (Ed.), The Oxford handbook of human
motivation (pp. 381–407). New York: Oxford University Press.
Bernard, L. C., Mills, M., Swenson, L., & Walsh, R. P. (2005a). An
evolutionary theory of human motivation. Genetic, Social, and
General Psychology Monographs, 131, 129–184.
Bernard, L. C., Mills, M., Swenson, L., & Walsh, R. P. (2008).
Measuring motivation multidimensionally: Development of the
Assessment of Individual Motives-Questionnaire [AIM-Q].
Assessment, 15, 16–35.
Bernard, L. C., Walsh, R. P., & Mills, M. (2005b). The motivation
analysis test: An historical and contemporary evaluation. Psy-
chological Reports, 96, 464–492.
Bollen, K. A. (1989). Structural equations with latent variables. New
York: Wiley.
Boring, E. G. (1950). A history of experimental psychology.
Englewood Cliffs, NJ: Prentice-Hall.
Borsboom, D., Mellenbergh, G. J., & van Heerden, J. (2003). The
theoretical status of latent variables. Psychological Review, 110,
203–219.
Brown, T. A. (2006). Confirmatory factor analysis for applied
research. New York: Guilford Press.
Buss, D. M. (1991). Evolutionary personality psychology. Annual
Review of Psychology, 42, 459–491.
Buss, D. M. (1999). Evolutionary personality psychology. Annual
Review of Psychology, 42, 459–491.
Buss, D. M. (2009). How can evolutionary psychology successfully
explain personality and individual differences? Perspectives on
Psychological Science, 4, 359–366.
Buss, D. (2012). Evolutionary psychology: The new science of the
mind. Boston: Allyn & Bacon.
Buss, D. M., & Hawley, P. H. (Eds.) (2011). The evolution of
personality and individual differences. New York: Oxford
University Press.
Byrne, B. (2006). Structural equation modeling with EQS; Basic
concepts, applications, and programming (2nd ed.). Mahway,
NJ: Lawrence Erlbaum Associates.
Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant
validation by the multitrait-multimethod matrix. Psychological
Bulletin, 56, 81–105.
Carver, C. S., & Scheier, M. F. (1981). Attention and self-regulation:
A control theory approach to human behavior. New York:
Springer.
Carver, C. S., & Scheier, M. F. (1998). On the self-regulation of
behavior. New York: Cambridge University Press.
Cattell, R. B. (1957). Personality and motivation structure and
measurement. Yonerks-on-Hudson, NY: World Book.
Cattell, R. B., & Dreger, R. M. (1977). Handbook of modern
personality theory. New York: Hemisphere Publishing.
Cattell, R. B., Horn, J. L., Sweney, A. B., & Radcliffe, J. A. (1964).
Handbook for the motivation analysis test. Champaign, IL:
Institute for Personality and Ability Testing.
Cattell, R. B., Radcliffe, J. A., & Sweney, A. B. (1963). The nature
and measurement of components of motivation. Genetic Psy-
chology Monographs, 66, 49–211.
Chou, C.-P., & Bentler, P. M. (1990). Model modification in
covariance structure modeling. A comparison among likelihood
ratio, Lagrange Multiplier, and Wald tests. Multivariate Behav-
ioral Research, 25, 115–136.
Confer, J. C., Easton, J. A., Fleischman, D. S., Goetz, C. D., Lewis, D.
M. G., Perilloux, C., et al. (2010). Evolutionary psychology:
Controversies, questions, prospects, and limitations. American
Psychologist, 65, 110–126.
Costa, P. T., Jr, & McCrae, R. R. (1992). Revised NEO Personality
Inventory (NEO PI-R) and NEO Five-Factor Inventory (NEO-
FFI) professional manual. Odessa, FL: Psychological Assess-
ment Resources, Inc.
Crano, W. D., & Brewer, M. B. (2002). Principles and methods of
social research. Mahway, NJ: Lawrence Erlbaum Associates.
Deci, E. (1975). Intrinsic motivation. New York: Plenum.
Deci, E., & Ryan, R. (1987). The support of autonomy and the control
of behavior. Journal of Personality and Social Psychology, 53,
1024–1037.
Dick, D. M. (2011). Gene-environment interaction in psychological
traits and disorders. Annual Review of Clinical Psychology, 7,
383–409.
Digman, J. M. (1996). The curious history of the five-factor model. In
J. S. Wiggins (Ed.), The five-factor model of personality:
Theoretical perspectives (pp. 1–20). NY: Guilford.
Figueredo, A. J., Va
´squez, G., Brumbach, B. H., & Schneider, S.
M. R. (2007). The K-factor, covitality, and personality. Human
Nature, 18, 47–73.
Figueredo, A. J., Va
´squez, G., Brumbach, B. H., Sefcek, J., Kirsner,
B. R., & Jacobs, W. J. (2005). The K-factor: Individual
differences in life history strategy. Personality and Individual
Differences, 39, 1349–1360.
Fiske, S. T. (2008). Core social motivations: Views from the couch,
consciousness, classroom, computers, and collectives. In J.
Y. Shah & W. I. Gardner (Eds.), Handbook of motivation science
(pp. 3–22). NY: Guilford.
Goldberg, L. R. (1993). The structure of phenotypic personality traits.
American Psychologist, 48, 26–34.
Hagen, E. E. (2005). Controversial issues in evolutionary psychology.
In D. M. Buss (Ed.), The handbook of evolutionary psychology
(pp. 145–174). New York: Wiley.
Harriman, P. L. (1977). Handbook of psychological terms. Totowa,
NJ: Littlefield, Adams & Co.
Heckathorn, D. D. (1997). Respondent-driven sampling: A new
approach to the study of hidden populations. Social Problems,
44, 174–199.
Holt, E. B. (1931). Animal drive and the learning process. Oxford:
Holt.
Hu, L., & Bentler, P. M. (1998). Fit indices in covariance structure
modelling: Sensitivity to underparametrized model misspecifi-
cation. Psychological Methods, 3, 424–453.
Motiv Emot (2014) 38:47–64 63
123

Hull, C. L. (1943). Principles of behavior: An introduction to
behavior therapy. Oxford: Appleton-Century.
Jackson, D. N. (1997). Personality research form manual. Port Huron,
MI: Sigma Assessment Systems.
James, W. (1890). The principles of psychology (Vol. 1). New York:
Holt.
Locke, E. A., & Latham, G. P. (1990). A theory of goal setting and
task performance. Englewood Cliffs, NJ: Prentice-Hall.
Lukaszewski, A. W., & Roney, J. R. (2011). The origins of
extraversion: Joint effects of facultative calibration and genetic
polymorphism. Personality and Social Psychology Bulletin, 37,
409–421. doi:10.1177/0146167210397209.
MacCallum, R. C., & Austin, J. T. (2000). Applications of structural
equation modeling in psychological research. Annual Review of
Psychology, 51, 201–226.
MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996).
Power analysis and determination of sample size for covariance
structure modeling. Psychological Methods, 2, 130–149.
MacDonald, K. (2005). Personality, evolution, and development. In
R. L. Burgess & K. MacDonald (Eds.), Evolutionary perspec-
tives on human development (2nd ed., pp. 207–242). Thousand
Oaks, CA: Sage.
Mahoney, C. A., Thombs, D. L., & Howe, C. Z. (1995). The art and
science of scale development in health education research.
Health Education Research, 10, 1–10.
Mayer, J. D., Faber, M. A., & Xu, X. (2007). Seventy-five years of
motivation measures (1930–2005): A descriptive analysis.
Motivation and Emotion, 31, 83–103.
McAdams, D. P., & Pals, J. L. (2007). The role of theory in
personality research. In R. W. Robins, R. C. Fraley, & R.
F. Krueger (Eds.), Handbook of research methods in personality
psychology (pp. 3–20). NY: The Guilford Press.
McClelland, D. C., Atkinson, J. W., Clark, R. W., & Lowell, E. L.
(1976). The achievement motive (2nd ed.). NY: Irvington.
Miller, G. A., Galanter, E., & Pribram, K. H. (1960). Plans and the
structure of behavior. New York: Holt.
Murray, H. A. (1938). Explorations in personality: A clinical and
experimental study of fifty men of college age. Oxford, England:
Oxford Univeristy Press.
Nettle, D. (2006). The evolution of personality variation in humans
and other animals. American Psychologist, 61, 622–631.
Nielsen, M., & Day, R. H. (1999). William James and the evolution of
consciousness. Journal of Theoretical and Philosophical Psy-
chology, 19, 90–113.
Oettingen, G., Bulgarella, C., Henderson, M., & Gollwitzer, P. M.
(2004). Self-awareness, self-motives, and self-motivation. In R.
A. Wright, J. Greenberg, & S. S. Brehm (Eds.), Motivational
analyses of social behavior: Building on Jack Brehm’s contri-
butions to psychology (pp. 225–244). Mahwah, NJ: Erlbaum.
Paynter, C. A., Kotovsky, K., & Reder, L. M. (2010). Problem-
solving without awareness: An ERP investigation. Neuropsych-
ologia, 48, 3137–3144.
Penke, L., Denissen, J. J. A., & Miller, G. F. (2007). The evolutionary
genetics of personality. European Journal of Personality, 21,
549–587.
Platt, S. A., & Sanislow, C. A., I. I. I. (1988). Norm-of-reaction:
Definition and misinterpretation of animal research. Journal of
Comparative Psychology, 102, 254–261.
Reiss, S. (2000). Who am I? The 16 basic desires that motivate our
actions and define our personality. New York: The Berkley
Publishing Group.
Reiss, S., & Havercamp, S. (1996). The sensitivity theory of
motivation: Implications for psychopathology. Behaviour
Research and Therapy, 34, 621–632.
Reiss, S., & Havercamp, S. (1998). Toward a comprehensive
assessment of fundamental motivation: Factor structure of the
Reiss Profiles. Psychological Assessment, 10, 97–106.
Roff, D. A. (2001). The evolution of life histories: Theory and
analysis. NY: Chapman & Hall.
Ryan, R. (2012). The Oxford handbook of human motivation. New
York: Oxford University Press.
Satorra, A., & Bentler, P. M. (1994). Corrections to test statistics and
standard errors in covariance structure analysis. In A. von Eye &
C. C. Clogg (Eds.), Latent variable analysis: Applications for
developmental research (pp. 399–419). Thousand Oaks, CA:
Sage.
Shah, J. Y., & Gardner, W. L. (Eds.). (2008). Handbook of motivation
science. New York: Guilford Press.
Sih, A., Bell, A., & Johnson, J. C. (2004a). Behavioral syndromes: An
ecological and evolutionary overview. Trends in Ecology &
Evolution, 19, 372–378.
Sih, A., Bell, A., Johnson, J. C., & Ziemba, R. E. (2004b). Behavioral
syndromes: An integrative overview. The Quarterly Review of
Biology, 79, 241–277.
Silvia, P. J., & Duval, T. S. (2004). Self-awareness, self-motives, and
self-motivation. In R. A. Wright, J. Greenberg, & S. S. Brehm
(Eds.), Motivational analyses of social behavior: Building on
Jack Brehm’s contributions to psychology (pp. 57–75). Mahwah,
NJ: Erlbaum.
Skinner, B. F. (1938). The behavior of organisms: An experimental
analysis. Oxford: Appleton-Century.
Skinner, B. F. (1971). Beyond freedom and dignity. New York:
Knopf/Random House.
Sorrentino, R. M., & Higgins, E. T. (Eds.). (1986). Handbook of
motivation and cognition: Foundations of social behavior. New
York: Guilford Press.
Szyf, M., McGowan, P. O., Turecki, G., & Meany, J. (2010). The
social environment and the epigenome. In C. M. Worthman, P.
M. Plotsky, D. S. Schecchter, & C. A. Cummings (Eds.),
Formative experiences: The interaction of caregiving, culture,
and developmental psychobiology (pp. 53–81). New York:
Cambridge University Press.
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate
statistics (5th ed.). Boston: Allyn and Bacon.
Thorndike, E. L. (1911). Animal intelligence. Experimental studies.
Oxford: Macmillian.
Tingergen, N. (1951). The internal factors responsible for the
‘spontaneity’ of behaviour. New York: Clarendon Press.
Tolman, E. C. (1932). Purposive behaviour in animals and men.
London: Century/Random House.
Tooby, J., & Cosmides, L. (1990). On the universality of human
nature and the uniqueness of the individual: The role of genetics
and adaptation. Journal of Personality, 58, 17–68.
Tooby, J., & Cosmides, L. (1992). The psychological foundation of
culture. In J. H. Barkow, L. Cosmides, & J. Tooby (Eds.), The
adapted mind (pp. 19–136). New York: Oxford University Press.
Toyokawa, S., Jddin, M., Koenen, K. C., & Galea, S. (2012). How
does the social environment ‘get into the mind?’ Epigenetics at
the intersection of social and psychiatric epidemiology. Social
Science and Medicine, 74, 67–74.
Ullman, J. B. (2007). Structural equation modeling. In B. G. Tabach-
nick & L. S. Fidell (Eds.), Using multivariate statistics (5th ed.).
Allyn Bacon: New York.
Ullman, J. B., & Bentler, P. M. (2003). Structural equation modeling.
In J. A. Schinka & W. F. Velicer (Eds.), Handbook of psychology
(pp. 607–634). Hoboken, NJ: Wiley.
Watson, J. B. (1913). Psychology as the behaviorist views it.
Psychological Review, 20, 158–177.
64 Motiv Emot (2014) 38:47–64
123