Comments 541

is an Agreeableness-bloated8 assessment of the personality CONCLUSION

space that fails to adequately capture the network of relations
among personality constructs at different hierarchical levels. These problems (and others highlighted in this issue) rein-
force that HEXACO-PM is not a good representation of the
personality construct space; HEXACO-PI is not an instru-
ment we currently have confidence in.

HEXACO, Dark Personalities, and Brunswik Symmetry

DELROY L. PAULHUS and PATRICK KLAIBER

University of British Columbia, Vancouver, BC Canada
[email protected]

Abstract: The authors have laid out a number of reasons for favoring the HEXACO model over the Five-Factor
model. We add that, in terms of compatibility with the Dark Tetrad model, the HEXACO is clearly superior. Whereas
the Five-Factor Model shows a motley pattern of linkages, the H-H factor provides a parsimonious higher-level con-
struct that subsumes established ‘dark’ personalities. Thus they dovetail nicely—another advantage of the HEXACO.
We couch this advantage in terms of the Brunswik Symmetry principle: Associations between measures must be eval-
uated at the same level of analysis. © 2020 European Association of Personality Psychology

We applaud the authors for the programmatic fashion in Acknowledging differing levels of analysis has an impor-
which they developed and validated the HEXACO. This tant corollary called Brunswik Symmetry (e.g. Figueredo
six-factor model of personality has proved to be a valuable et al., 2015). In brief, the principle holds that associations be-
practical tool as well as a challenge to Big Five advocates. tween two psychological measures (e.g. Pearson correla-
Of special note is the authors’ introduction of a new factor tions) are impaired if they fall at different levels of the
—Honesty-Humility (H-H). Its addition has led to a recon- hierarchy. That principle is especially important in personal-
sideration of how the earlier five factors should be oriented ity measurement because the construct hierarchy can be
in personality space. tapped at any one of a number of levels (see Ashton et al.,
We have a special interest in the H-H factor because of its 2004; Saucier & Ostendorf, 1999).
relevance to so-called ‘dark’ personalities: that is, narcissism, In Point 4 of the target article, the authors agree that the
Machiavellianism, psychopathy, and sadism (e.g. HEXACO factors should not be used when researchers are
Paulhus, 2014). To anticipate, we believe that H-H subsumes interested in a specific trait. In Point 5, the authors do not
but does not preclude the latter. specifically apply that argument to the dark personalities. In-
stead, they express concern about researchers who tack the
dark variables on to the Big Five. For various reasons—
Brunswik Symmetry being paramount—we strongly agree
OVERARCHING PERSONALITY CONSTRUCTS with that concern. We do not agree, however, with the con-
VERSUS LOWER LEVEL TRAITS verse argument: that, with the advent of H-H, dark variables
are now superfluous (Hodson et al., 2018).
The trade-off between broad-band and narrower, high fidelity The Brunswik principle also implies that constructs such
measures has been acknowledged throughout the history of as Agreeableness or H-H cannot act as a core or mediator for
personality assessment (e.g. Paunonen et al., 1999; lower level traits. The shell cannot mediate its constituents:
Wiggins, 1973). Although narrower measures provide more its meaning is distilled from them. Mediators must operate
precise measurement, inventories of broad-band measures at the same level of analysis as the variables they mediate
have dominated empirical research because of their compre- (Figueredo et al., 2015).
hensiveness and efficiency: among these are Eysenck’s PEN,
the Big Five, and the HEXACO.
8
THE ORIGIN AND ADVANTAGES OF STUDYING
Ashton and Lee describe bloated specifics as arising “only when lots of THE DARK TETRAD (D4)
nearly redundant variables are included in a factor analysis” (p. 29). They as-
sert that HH is not a bloated specific because its factor loadings are consis-
tent in magnitude to that of the remaining factors. However, Cattell (1978, Whereas broad personality factors such as the H-H emerge
p. 319) correctly notes that “artificially high intercorrelations [among items from comprehensive lexical analyses, lower level traits such
or subtests] and spurious claims to high broad factor loadings are often
due to making up highly similar items which introduce an undetected as the D4 originate in focused investigations of variables
‘bloated specific’.” showing convergence of self-report, observer-report, and

© 2020 European Association of Personality Psychology Eur. J. Pers. 34: 511–590 (2020)
DOI: 10.1002/per
542 Comments

behavioral modes of measurement. In each case, its validity narcissism has its strongest associations with extraversion
has been buttressed by psychometric, laboratory, I-O, and and openness (Book et al., 2016; O’Boyle et al., 2015).
clinical research. The subclinical variants have shown re- By contrast, the six-factor solution offered by the
markable parallels to their clinical–forensic counterparts. In HEXACO model provides a clear pattern with all four vari-
short, Ashton and Lee are incorrect when declaring that there ables aligning with the H-H factor (Book et al., 2016; Meere
is no empirical basis for focusing on the three (now four) & Egan, 2017; Johnson et al., 2019; Međedović &
constructs. Although others have been nominated, these four Petrović, 2015; Zettler et al., 2020). This parsimony offers
have the strongest theoretical and empirical basis (Paulhus more theoretical and empirical clarity than the scattered links
et al., 2020). with the Five-Factor Model.
Readers interested in the lengthy historical background of In fact, Hodson et al. (2018) have shown that the first la-
each tetrad member are directed to Jones and Paulhus (2009) tent factor of the Dark Triad lines up with the latent factor of
for Machiavellianism; Campbell and Miller (2011) for nar- H-H. The authors went further to argue that researchers
cissism; Lebreton and colleagues (2006) for psychopathy; should simply use the (reversed) H-H scale instead of the
and Foulkes (2019) for subclinical sadism. The substantial triad. Clearly, that recommendation went too far, but the cor-
overlap in these literatures—as well as their measures—had respondence of latent factors is evidence that the higher-level
originally spurred Paulhus and Williams to investigate the H-H subsumes the lower-level Dark Tetrad (and more).
similarities and differences. Possible differentiation required The authors agree that the facets chosen for H-H were
simultaneous measurement and extensive research—soon somewhat arbitrary (Point 4). Nonetheless, most resemble
forthcoming (Furnham et al., 2013). Interestingly, the reverse-keyed versions of tetrad traits, which are firmly
bottom-up and top-down investigations were occurring at rooted in empirical traditions. Thus, in most cases, links with
about the same time. Emergence of the H-H factor was soon other constructs are more easily interpreted using tetrad traits
accompanied by the revelation that it subsumed the Dark than the H-H facets.
Triad (Lee & Ashton, 2004). In sum, the D4 and HEXACO models are not in compe-
tition but are complementary. The HEXACO model helps in-
tegrate dark personalities into personality space, whereas the
THE HEXACO VERSUS FIVE-FACTOR MODELS tetrad traits provide well-defined lower-level traits falling
neatly under H-H. Overall, we are sympatico with the au-
Several studies have sought to locate the dark constellation in thors’ favoring of their six-factor HEXACO model over the
overarching personality space. Under the assumption that the Five-Factor Model. In terms of compatibility with the Dark
Five-Factor Model was exhaustive, some have attempted to Triad and Tetrad constellations, the HEXACO is clearly su-
project dark variables onto that structure. Unfortunately, that perior. In promoting the HEXACO in previous writings, it
mapping task proved messy: for example, psychopathy has is unfortunate that the authors expressed disdain for the dark
been linked to all five (Paulhus & Williams, 2002), Machia- personality distinctions (Hodson et al., 2018, pp. 127–128).
vellianism to neuroticism and low agreeableness (DeShong Au contraire, we feel that they dovetail nicely—another ar-
et al., 2015; Jakobwitz & Egan, 2006). By contrast, gument favoring the HEXACO over the Five-Factor Model.

Why … the HEXACO Model Should Be Preferred as a Basic Personality Structure

MARCO PERUGINI
University of Milan-Bicocca, Milan, Italy
[email protected]

Abstract: I argue that personality structure should be grounded in psycholexical studies as this provides a most solid
foundation. Personality structure should be factorially simple and should be distinguished from personality content.
In turn, both should not be given per se explanatory status and should be distinguished from personality processes.
Based on these considerations and the theoretical and empirical arguments put forward by Ashton and Lee, I agree
with them that one should prefer the HEXACO model as a basic personality structure. © 2020 European Association
of Personality Psychology

WHY PERSONALITY STRUCTURE SHOULD BE psycholexical studies. In one of their last points (14), they
GROUNDED IN PSYCHOLEXICAL STUDIES … briefly explain why it matters. My only disagreement with
them is that it should have been the first point, just to signal
Ashton and Lee (2020a) highlight that their proposed how fundamental is this issue. First, a psycholexical study
HEXACO personality structure is well-grounded in defines a broad set of features (e.g. personality-descriptive

© 2020 European Association of Personality Psychology Eur. J. Pers. 34: 511–590 (2020)
DOI: 10.1002/per