Minkov indb

Notes 
1. In the early 1980s, Adler (1983) advised 
against the treatment of culture as a residual 
but stated that it could be viewed “as an inde-
pendent or as a dependent variable” (p. 37). At 
the turn of the 20th century, van de Vijver and 
Leung (1997a) had to inform their readers that 

---

18
◆
Understanding “Culture”
“culture is too global a concept to be meaning-
ful as an explanatory variable, however, and 
should be replaced by its constituents” (p. 3). 
Singelis et al. (1999) noted that cross-cultural 
studies in psychology had often been criticized 
precisely because culture was treated as a 
single package, although it can be unraveled 
into numerous variables, any of which might 
account for the observed differences between 
the populations that a researcher has studied; 
consequently, it is necessary to unpackage cul-
ture. Almost a decade later, Leung (2008) still 
deemed it necessary to give the same advice: 
“In other words, researchers need to unpack-
age culture into a set of elements.” (p. 60). 
Treating culture as a single categorical 
variable (for instance, “American” versus 
“Japanese”) and using it as an explanation for 
any phenomenon is as pointless and confusing 
as doing the same with other categorical vari-
ables, for instance, “man” versus “woman.” In 
fact, these are identification labels, not factors 
that can cause anything. If one finds any differ-
ence between a male population and a female 
population on a variable of interest, such as 
aggressiveness, ascribing the difference to being 
“male” versus “female” does not elucidate 
anything about the nature of that difference. 
Differences in aggression are not produced 
by different labels but by differences in genes, 
hormones, patterns of upbringing, and so on. 
Only studies of such characteristics, expressed 
as numerical variables, can shed light on dif-
ferences in aggression or other phenomena 
between individuals or groups. 
2. The low status of the social sciences 
was noted also by Magala (2005). 
3. In his treatise on cross-cultural analy-
sis, Parker (1997) advocated controlling for 
factors that are “(1) exogenous to the depen-
dent variable yet (2) independent to the theory 
under study” (p. 13). It is needless to say that 
selecting such factors would involve a lot of 
subjectivity since any theory that is still in the 
process of being studied empirically is inevi-
tably subjective. Being aware of this problem, 
Parker (1997) noted that each discipline within 
the social sciences often treats the others’ vari-
ables as exogenous to their variables of interest. 
4. Consider also the following statement 
about personality factors by Paunonen et al. 
(1996): “But those findings do not mean that 
other factors, equally real and equally impor-
tant, do not exist, be it in North American, 
European, or other cultures. The problem is 
that people have yet to provide a convincing 
search for those other factors. For a variety 
of reasons having to do not only with vari-
able selection but also with the methodology 
of factor analysis . . . , it is our belief that 
the number five is probably a lower bound to 
the true number of factors at this level of the 
personality hierarchy” (p. 351, italics added). 
The words real, exist, search, and true number 
suggest that these authors see personality fac-
tors as having an existence of their own and an 
unknown fixed number. These real factors are 
lurking in the dark and waiting for researchers 
to find them with appropriate search engines. 
5. The following example can serve as an 
illustration. Schwartz and Sagiv (1995) demon-
strated that Schwartz’s value structure theory 
was essentially supported at the individual 
level throughout the countries from which 
Schwartz’s samples were drawn. However, 
Schwartz and Sagiv also published national 
estimates of deviations from the hypothesized 
structure. One such estimate—“deviations of 
value locations” (Table 2, p. 99) correlates 
with Hofstede’s individualism index as follows: 
teachers’ samples 
–.68** ( n = 24) 
students’ samples 
–.60** ( n = 26) 
(Note: Here and throughout the book, ** 
stands for correlation significant at the .01 
level; * stands for correlation significant at the 
.05 level.) 
GLOBE’s in-group collectivism index (see 
9.17.) yields positive correlations of a similar 
magnitude with the deviation measures. This 
demonstrates that although Schwartz’s theory 
finds some universal empirical support, it is 
closest to the value structures in the minds of 
the respondents in the individualist nations. 
As Schwartz’s project evolved from the work 
of Milton Rokeach (Schwartz, 2011), it is not 
surprising that a Western perspective can be 
discerned in it. 
Of note, Hofstede, Hofstede, and Minkov 
(2010) acknowledged that their perspective 
was partly shaped by their Dutch and Bulgarian 
backgrounds.