Factor analysis of color preferences

[Oleg M. Goryunov]

Abstract

This study empirically examined the individual color preferences by factor analysis. A
total of 64 worldwide participants in the age range 18-65 years evaluated 320 colors
with Random Color Test. The data were compressed to one general color factor and to
five factors explaining 29% and 54% of total variance respectively. Significant
correlations between the factors were found. The applying of scientific, well-developed
method of current psychometrics to color preferences creates a new branch for
investigation of relationships between personal traits and color preferences factors
assessments.

Keywords: color preferences; factor analysis; Random Color Test; general color factor;
GCF.

Factor analysis of color preferences

Psychologists long studied the individual color preferences (CP). The number of colors
that they researched typically was less than thirty. Such small sets of colors
motivated to find relationships between CP and something external to CP, but not
between CP themselves.
To conduct a factor analysis of CP themselves, it was necessary to use a perceptually
uniform color space, for example, CIELAB (CIE 1976 L*a*b*), i.e. an approximation to
the absolute color space in which the perceptual difference between colors is directly
related to distances between colors as represented by points in the color space (Volz,
2001; Buxbaum & Pfaff, 2005).
Following this necessity, the author increased the number of colors, distributed evenly
in the CIE 1976 L*a*b* color space, up to 320. Hereupon it became possible to find the
CP factors (CPF). The found CPF were compared with factors of Big Five model (Costa &
McCrae, 1992) and with the general factor of personality (GFP) (Musek, 2007).

Method

Participants

The sample consisted of a total of 64 worldwide respondents, who sent voluntary their
data through Internet. Male participants were 26, female ones - 33 (in some cases,
information on the gender or/and age was omitted). Participants' number (age) were: 5
(18-19), 17 (20-24), 10 (25-29), 8 (30-35), 4 (36-39), 5 (40-49), 4 (50-59), 2 (60-65),
9 (no data).
The participants passed a Random Color Test.

Measures

The Random Color Test was developed by the author in 2012. It was a computer program
to display on screen 320 unique colors in a random order. A participant evaluated,
choosing a number from the set {0, 1,... 7}, how much one liked a color. Then the
program displayed next random color.

Results

Principal component analysis

The principal component analysis with varimax rotation was conducted with a program GNU
PSPP (http://www.gnu.org/software/pspp/).
One-component and five-component solutions were examined. The first explained 29% of
the total variance, the second - 54%. From every component five the most loaded
variables were chosen for the principal one-factor analysis with over 12 cases per
variable.

Principal Factor Analysis

A program Factor was used for the Unweighted Least Squares factor analysis
(http://psico.fcep.urv.es/utilitats/factor/index.html).
The factorability of the chosen variables was examined. For every set of variables
several criteria for the factorability were used. The items correlated at least .45
with the other items of the set, suggesting good factorability. The Kaiser-Meyer-Olkin
measure of sampling adequacy was at least .8, and Bartlett's tests of sphericity were
significant (chi-square (10) > 148.5, p < .000000).
All one-factor analysises of the six sets were checked parallel analysises based on
Minimum Rank Factor Analysis (Timmerman & Lorenzo-Seva, 2011). Advised number of
dimensions for every set of variables was 1.
Loadings of the one-factor solutions were at least .52. Reliability estimates (Mislevy
& Bock, 1990) were at least .88.
Using factor scores, the factors correlations were found. The correlations (p < .01)
between general color factor (GCF), based on one-component solution, and four CPF were:

.67 - Extraversion;

.57 - Neuroticism;

-.52 - Openness to experience;

.39 - Agreeableness.

The correlation between GCF and Conscientiousness color factor was -.26 (p < .05). The
CPF were presumably identified in terms of the Big Five model of personality.

Discussion

The results of the study prove existence of CPF and the significant correlations
between GCF and the other five CPF. The GCF differs from the GFP:

GFP E -N O A C (Rushton, Bons, & Hur, 2008);

GCF Ec Nc -Oc Ac -Cc (the subscript character
"c" labels the CPF).

Looking on the first three "constituent parts" of GCF - Extraversion, Neuroticism,
Conservativeness (-Openness) one could interpret GCF as a Conservativeness - Liberality
vector. Whereas the GFP unites the socially desirable labels of the Big Five. The
future researches are needed to explain the GFP-GCF differences.
The main advantage of the CPF assessments is using scientific, well-developed method of
the current psychometrics. The other advantage, provided that the psychological meaning
of colors is concealed (the substantiation of identification of the CPF is not given
for this reason), is minimizing the impact of social desirability on the tests results.
The randomness of displaying of colors, especially in combination with retesting,
strengthens this advantage. The concealment of the psychological meaning of colors is
possible, if it be only one, i.e. international, center of color testing with a
centralized data base. Thereby the optimal using of human features could be achieved.

References

Buxbaum, Gunter, & Pfaff, Gerhard (2005). Industrial Inorganic Pigments. Wiley-VCH.
ISBN 3-527-30363-4.

Costa, P.T., Jr., & McCrae, R.R. (1992). Revised NEO Personality Inventory (NEO-PI-R)
and NEO Five-Factor Inventory (NEO-FFI) manual. Odessa, FL: Psychological Assessment
Resources.

Mislevy, R.J., & Bock, R.D. (1990). Bilog 3: Item Analysis and Test Scoring with Binary
Logistic Models: Scientific Software.

Musek, J. (2007). A general factor of personality: Evidence for the Big One in the
five-factor model. Journal of Research in Personality, 41, 1213-1233.

Rushton, J. Philippe, Bons, Trudy Ann, & Hur, Yoon-Mi. (2008). The genetics and
evolution of the general factor of personality. Journal of Research in Personality,
42(5), 1173-1185. doi: http://dx.doi.org/10.1016/j.jrp.2008.03.002

Timmerman, M. E., & Lorenzo-Seva, U. (2011). Dimensionality Assessment of Ordered
Polytomous Items with Parallel Analysis. Psychological Methods, 16, 209-220.

Volz, G. Hans (2001). Industrial Color Testing: Fundamentals and Techniques. Wiley-VCH.
ISBN 3-527-30436-3.