Exploring the Roots of Culture Using Spatial Autocorrelation

Abstract

This article examines the distribution of some well-established dimensions of national culture within geographic and ecosocial space. Using spatial autocorrelation to quantify the relationships between geographical location, distance from the equator, physical climate, religion, and national development and 38 cultural dimensions, we find a substantial degree of spatial organization; some cultural dimensions are as spatially organized as temperature or rainfall. Overall, we find that culture as measured by aggregated personal values covaries to about the same extent with geographical proximity, national development and religion, and significantly less with the physical climate and distance from the equator. It is also found that there are significant differences in the spatial patterning of GLOBE societal-level values and the personal value measures of previous researchers. GLOBE values are also strongly organized by geographical proximity and religion but unlike personal value measures only weakly organized by level of national development. GLOBE practices are not strongly spatially organized at all suggesting that practice in a nation does not evolve simplistically from national values or vice versa. Religion, a major organizing variable for how people believe things should be in society, has little relationship with how people believe things are. We conclude that the approach taken by the project GLOBE is a valuable contribution to our understanding of national culture and that spatial autocorrelation is an exploratory method of analysis that is underused in cross-cultural research.

Keywords

national culture physical climate national development religious profile geographical proximity spatial autocorrelation GLOBE

Introduction

In this article, we investigate five geographical and ecosocial factors that previous authors have found or reasoned to be associated with variation in national culture, namely, geographical proximity, distance from the equator, the nature of the physical climate, level of national development, and religious profile. Although major studies have adopted a variety of clustering approaches to overcome the problems created by non-independence of observations, the suspicion remains that the reported effect sizes may be biased and conclusions unsound (see our discussion of Galton’s problem later in this article). Therefore, the purpose of this article is to validate previous findings by quantifying the strength of the relationships between these contextual factors and some of the major dimensions of national culture using a different and more defensible method of analysis and a wider range of studies and dimensions.

Unlike previous research in the area, we use a measure of spatial autocorrelation (Moran’s I) to quantify these relationships. Moran’s I measures the degree of autocorrelation between variables distributed in some contextual space and as such enables the test of the general hypothesis that nations that are similar in some contextual variable X (for example, climate or religious profile) are similar in some cultural dimension Y (for example, individualism or societal cynicism). Moran’s I provides direct estimates of such contextual effects, without requiring the a priori clustering of nations. Furthermore, assuming that samples of nations are similar, Moran’s I enables comparison of the strength of the relationships across studies as well as within them. Hence, we have included a wider range of studies and dimensions of national culture than previous work.

The structure of the article is as follows. First, we outline some of the evidence from previous work that suggests that our five selected context variables are associated with the national cultural differences and develop some general hypotheses. Next, we discuss the nature of Galton’s problem and the reasons why previous conclusions may be unsound. Then, we present spatial autocorrelation and Moran’s I as an alternative method of analysis. Finally, we use Moran’s I to undertake an analysis and discuss the results.

Geographical and Ecosocial Influences on National Culture

Geographical proximity

From a theoretical point of view, there are good reasons for expecting countries who are geographical neighbors to be more similar than those far apart. Societies may impose their beliefs or practices on one another, or adopt beliefs and practices from each other voluntarily. In either case, spatial proximity is likely to increase the degree of interaction and interchange between societies, producing more convergence between cultures that are geographically close than between those that are far apart. Furthermore, populations may migrate and short-range migration would tend to increase the cultural similarity of nearby regions.

Of course, it would be absurd to claim that geographical distance alone can explain the similarities and differences between national cultures. Indeed, as pointed out by Inglehart and Baker (2000), some nations that are far apart, for instance the United Kingdom and Australia, or Spain and Uruguay, share similar cultures because of shared historical connections and large-scale immigration. Nevertheless, there are good reasons to expect a higher degree of cultural similarity between social groups that are physically close then in those that are far apart.

Cross-cultural psychologists have often observed that the constructs they study are organized geographically. In an influential article, Ronen and Shenkar (1985) reviewed eight studies that clustered countries by the similarity of work values and concluded, “it is apparent . . . that countries tend to group together geographically” (p. 444). A similar conclusion was reached by Schwartz (1999) who used the co-plot technique to locate 44 national cultures on a two-dimensional map so that nations with similar cultural profiles appeared close together, and those with dissimilar profiles appeared far apart. The results indicated the existence of broad cultural groupings of nations that were largely defined by geographical proximity. Similarly, Georgas, van de Vijver, and Berry (2004) noted that when nations are clustered according to ecosocial indices, they form cultural zones consisting of nations that share cultural similarities and that tend to be “. . . in general, geographically contiguous” (p. 78).

Hypothesis 1: Such is the pervasive impact of geographical proximity that we would expect distance between nations to have a significant impact on many of the cultural dimensions included in this study. Close geographical neighbors should be found to have similar cultures.

Physical climate

For many years, social scientists have argued for the impact of the physical environment on cultural development. Harsh environments—extreme hot or cold and those lacking in rainfall—demand greater concern with the maintenance of body temperature, shelter, and the provision of food and water.

Previous authors have suggested that some cultural traits are adaptations to environmental conditions. For example, it has been shown that warmer climates promote increased emotional expression (Pennebaker, Rime, & Blankenship, 1996; Robbins, DeWalt, & Pelto, 1972), elevated levels of aggression and hostility (Anderson, 1989), and more immediacy behaviors such as smiling, physical closeness, and touching (Andersen, 1985). Van de Vliert, Huang, and Parker (2004) have also demonstrated relationships between the physical climate and altruism and subjective well-being. Gilmore (1990) has argued that a value of masculinity (i.e., strong gender differentiation and a high regard for strength and fearlessness) is more likely to develop in harsh climates where survival depends on the ability to battle courageously against a hostile environment.

Of specific relevance to studies considered here, Georgas et al. (2004) report significant effect sizes for ecology (a composite of precipitation and temperature) for Schwartz’s (1999) autonomy factor, Hofstede’s (2001) power distance and individualism, and Smith, Dugan, and Trompenaars’ (1996) involvement and commitment scales. Also, Bond et al. (2004) report a significant positive relationship between dynamic externality and average daytime temperature.

Van de Vliert, Schwartz, Huismans, Hofstede, and Daan (1999), in a 136 nation study, found a curvilinear relationship between thermal climate and masculinity. Nations in temperate climates were found to be more masculine than those in either hotter or colder climes. They argue that for warm-blooded humans it is extreme hot or cold climates that cause climatic stress. Van de Vliert (2007) also finds a curvilinear relationship between Inglehart & Baker’s (2000) survival versus self-expression dimension and thermal climate which is moderated by affluence. Understandably, poorer countries in harsh climates endorse survival values whereas richer countries in such climates are more likely to value self-expression. Van de Vliert (2011) finds similar curvilinear relationships between his Thermal Demand Scale and measures of compatriotism, nepotism, and familism that are moderated by affluence. These measures of in-group favoritism are strongest in lower income countries with demanding hot or cold climates, moderate in countries with temperate climates irrespective of income per head, and weakest in higher income countries with demanding hot or cold climates. Of relevance to our current study, Van de Vliert used GLOBE’s in-group collectivism (practice) as his measure of familism, and so we would expect to find a curvilinear relationship between in-group collectivism (practice) and thermal climate.

Hypothesis 2: Nations with similar physical climates will be found to be similar in terms of the cultural dimensions of autonomy (Schwartz, 1999), power distance, individualism, and masculinity (Hofstede, 2001), involvement and commitment (Smith et al., 1996), dynamic externality (Bond et al., 2004), survival versus self expression (Inglehart & Baker, 2000), in-group favoritism (Van de Vliert), and in-group collectivism practices (GLOBE).

Distance from the equator

Disregarding the impact of land mass and altitude, a number of authors have used latitude or distance from the equator as a proxy for a linear Hot–Cold Temperature Scale. For example, Hofstede (2001) noted that some of his dimensions of cultural values varied systematically with distance from the equator and suggested that “. . . a country’s geographic position is a fundamental fact that is bound to have a strong effect on the subjective culture of its inhabitants . . .” (p. 116). He found that power distance was inversely related to latitude whereas individualism was positively related. Hofstede and Hofstede (2005) also report a moderate negative correlation between masculinity and latitude and suggest that feminine societies are somewhat more likely in colder climes because of the need for cooperation and shared roles between the sexes.

Hypothesis 3: Nations similar in latitude will be found to be similar in terms of the cultural dimensions of power distance, individualism, and masculinity (Hofstede, 2001).

Level of national development

Level of national development and measures of socioeconomic development have also frequently been found to be associated with national values. Modernization theorists (e.g., Bell, 1973) have argued that economic development influences national values. Thus, as a society moves from pre-industrial, through industrial and post-industrial stages of development, culture reflects differing concerns. Inglehart and Baker (2000) reasoned and provided evidence that traditional versus secular rational and survival versus self-expression dimensions would reflect the change from preindustrial to industrial and then industrial to post-industrial societies respectively. Georgas et al. (2004) concluded that affluence was positively associated with self-enhancing values such as individualism (Hofstede, 2001), egalitarian commitment (Smith et al., 1996), and self-expression (Inglehart & Baker, 2000), and negatively associated with power distance (Hofstede, 2001).

Hofstede (2001) reports a strong relationship between the level of economic development and individualism, and that richer countries score lower on power distance. One would expect Schwartz’s (1999) hierarchy and egalitarianism and conservatism and autonomy scales to be similarly clustered by economic development. Gouveia and Ros (2000) evaluated the associations between social and economic variables, and measures of individualism based on Hofstede’s (2001) and Schwartz’s (1999) cultural dimensions. They found that the Hofstede (2001) model was better explained by economic factors such as Gross National Product (GNP), agricultural activity, and inflation, whereas the Schwartz (1999) model was better explained by social indicators such as birth rate, human development, and illiteracy.

Bond et al. (2004) report significant negative correlations between dynamic externality and both GDP per capita and the Human Development Index.

Hypothesis 4: Nations similar in level of national development will be found to be similar in terms of the cultural dimensions of survival versus self-expression and traditional versus secular rational (Inglehart & Baker, 2000), individualism, and power distance (Hofstede, 2001), egalitarian commitment (Smith et al., 1996), hierarchy, egalitarianism, conservatism, and autonomy (Schwartz, 1999), and dynamic externality (Bond et al., 2004).

Religious profile

Huntington (1996) identified eight major civilizations largely based on different religions that he argued have persisted for centuries. Inglehart and Baker (2000) used Huntington’s civilizations as a guide to draw boundaries among groups of nations and concluded that both economics and religion make significant contributions to the development of national values: economic development moves societies in a common direction but distinctive cultural zones based around religious values persist. “A history of Protestant or Orthodox or Islamic or Confucian traditions gives rise to cultural zones with distinctive value systems that persist after controlling for the effects of economic development” (Inglehart & Baker, 2000, p. 49). They concluded that the historically dominant religion of a nation impacts on both survival versus self-expression and traditional versus secular rational values.

Georgas et al. (2004) using a wider range of cultural variables than Inglehart & Baker (2000) also found religion and economics to be associated with cross-national differences. In contrast to Inglehart & Baker, they found the effect of religion to be greater than that of economic development. Specifically, they found that power distance and uncertainty avoidance (Hofstede, 2001), hierarchy (Schwartz, 1999), and commitment and involvement (Smith et al., 1996) were associated with religious profile. In contrast to earlier findings, Georgas, Berry, van de Vijver, Kagitcibasi, and Poortinga (2006), in a 30-nation study, found the effect of economic development to be greater than that for religion.

Bond et al. (2004) report a strong correlation between dynamic externality and a variety of religious aspects. Also, Hofstede and Bond (1988) report a strong link between Confucianism and long-term orientation (Dynamic externality).

Hypothesis 5: Nations similar in religious profile will be found to be similar in terms of the cultural dimensions of survival versus self-expression and traditional versus secular rational (Inglehart & Baker, 2000), power distance, uncertainty avoidance, and long-term orientation (Hofstede, 2001), hierarchy (Schwartz, 1999), commitment and involvement (Smith et al., 1996) and Dynamic Externality (Bond et al., 2004).

Having briefly reviewed the findings of previous studies on the impact of contextual variables on national culture let us turn to an issue that has confronted them all, namely, Galton’s problem.

Galton’s Problem

In an analysis of international data sets that included Hofstede’s (2001) cultural dimensions and aspects of the World Values Surveys, Eff (2004) found that 76% of the 864 tests for autocorrelation were significant at the .01 level. Galton’s problem (Naroll, 1965) is a significant issue for cross-cultural research. In essence, this states that due to the likelihood of cultural borrowing and common descent social phenomena are unlikely to be independent and unless steps are taken to control for this, valid inferences cannot be made. Many statistics assume independence of observations.

Ordinary least squares requires that the residuals in the estimated model not be correlated with each other. Violation of this property causes the estimated standard errors of the coefficients to be biased so that one cannot trust the t statistics and one therefore cannot make hypothesis tests regarding the estimated coefficients. (Eff, 2004, p. 2)

Dow and Eff (2008) used Moran’s I (see below) to estimate the extent of non-independence in the Standard Cross-Cultural Sample (SCCS) probably the most widely used data set in cross-cultural research. They conclude,

The finding of such high levels of autocorrelation in the SCCS clearly indicates that the failure to adequately account for the potentially devastating effects of non-independence on statistical inference seriously limits the in-depth interpretation of almost all global statistical analysis in cross-cultural studies to date. (p. 167)

In contrast, Ember and Ember (2009) consider that Galton’s problem is not a serious problem in the sense that there are solutions to the problem of nonindependence of observations.

One potential solution is to eliminate multiple cases from the same cultural or geographical area. Thus, the SCCS (Murdock & White, 1969) using judgmental criteria as a basis for selection, and the Human Relations Area Files (HRAF probability sample files) using a random sample, provide just one case per cultural area. Although Dow & Eff’s (2008) analysis of the SCCS suggests that the use of judgmental criteria as the basis of selection of cases has not overcome Galton’s problem, the random selection of cases in the HRAF probability sample files, as advocated by Ember and Ember (2000), accords with the requirements of statistical independence and thus reduces the possibility of non-independence biasing the results. Although the procedure reduces the analytical sample, selecting repeated random samples from a larger list of societies, cultural areas or nation states with replicated results adds to the credibility of the findings.

A second potential solution is to aggregate cases into clusters, such that cases within a cluster are similar and cases in different clusters are dissimilar, and to examine the cultural variations between clusters. Inglehart and Baker (2000), for example, plotted their sample of nations on a two-dimensional map using scores on the cultural dimensions of traditional versus secular rational and survival versus self-expression as x and y axes. They then superimposed boundaries defining four levels of economic development. As the areas defined by the economic zones were contiguous, Inglehart and Baker (2000) concluded there was a systematic relationship between culture and economic development. One weakness of this approach is that the choice of four economic zones is rather arbitrary. Would the organization be so clear-cut with five rather than four economic zones? Georgas et al. (2004) however, used a more rigorous approach (cluster analysis) to identify groups of nations with similar ecosocial indices, and then estimated the influence of cluster membership on a number of cultural dimensions. Analysis of variance was used to examine the extent of variation between clusters, the critical statistic being the effect size (proportion of variance explained) by cluster membership. Such an approach may overcome Galton’s problem, but the question remains as to how independent these groups and clusters are. A further limitation of this approach is that the observed effect size depends on the clustering pattern that emerges, changing the number of clusters or changing the number of nations in a cluster can change the observed effect size. It is therefore arguable that their effect size for say ecology (based on five clusters) is not directly comparable with their effect size for religion (based on four clusters). Even where the underlying independent variables are the same, different cluster solutions will of themselves produce different effect sizes.

A more elegant way of treating the problem of non-independence of observations is to use spatial statistics to estimate its influence or to control for its effects (Dow, 2007). Spatial statistics have been extensively employed by geographers and in a wide variety of other scientific contexts, for example, in epidemiology by Rosenberg, Sokal, Oden, and DiGiovanni (1999), in political science (e.g., Beck, Gleditsch, & Beardsley, 2006), economics (e.g., Moscone & Knapp, 2005), genetics (e.g., Heywood, 1991), ecology (e.g., Baskent, 1999), sociology (e.g., Loftin & Ward, 1983), criminology (e.g., Baller, Anselin, Messner, Deane, & Hawkins, 2001), and neuroscience (e.g., Hafting, Fyhn, Molden, Moser, & Moser, 2005).

Apart from the cross-cultural study of IQ (Gelade, 2008) and aspects of Eff’s (2004) and Dow and Eff’s (2008) articles referred to above, spatial statistics have rarely been used in cross-cultural psychology. In this article, we use spatial statistics to assess the extent of non-independence of national cultures within different types of ecosocial space. Here, rather than view non-independence as a problematic source of bias to be controlled for, we view it as a cultural phenomenon to be quantified and explored.

Spatial Data and Spatial Autocorrelation

By spatial data, we mean a set of locations, {l₁ . . . . ., l_n} and a set of observations which vary by location {x (l₁) . . . . . ., x(l_n)}. Traditional statistics relies on the assumption of independence between observations, but because independence cannot be assumed for spatial data, traditional statistical techniques are not appropriate.

The assessment of spatial autocorrelation is one of the most widely used techniques within spatial statistics, and the most widely used measure of spatial autocorrelation is Moran’s I (Moran, 1950). Moran’s I quantifies the relationship between location and some attribute characteristic of that location and can be envisaged as a spatial analogue of the familiar Pearson correlation coefficient. The formula for Moran’s I can be written:

I = \frac{N \sum_{i} \sum_{j} w_{ij} z_{i} z_{j}}{s_{0} \sum_{i} z_{i}^{2}}

where N is the number of locations; w _ij is the element in a weights matrix w describing the spatial relationship between locations i and j; z_i and z_j are observations at locations i and j, respectively, expressed as deviations from the mean; and S₀ is a normalizing factor equal to the sum of the elements of the weights matrix, that is, S₀ = ∑_i∑_jw _ij . For details, see Cliff and Ord (1973, 1981), and Upton and Fingleton (1985).

Moran’s I is structurally similar to the familiar Pearson correlation coefficient; both are cross-product statistics whose numerator contains the term z_i times z _j . It can easily be seen that if near neighbors are similar, this term will tend to be positive, because both values will tend to be either simultaneously above or simultaneously below the mean; while if near neighbors are dissimilar, the observed values are likely to fall either side of the mean so that the product of z_i and z _j will tend to be negative. I generally takes the values between −1 and +1, although in certain circumstances these bounds may be exceeded. Positive values of I indicate positive spatial autocorrelation or clustering (for example, nations that are geographically close together have similar levels of individualism). whereas negative values indicate the presence of negative spatial autocorrelation or a nonrandom “checkerboard” distribution (nations close together have different levels of individualism) Values of I close to zero indicate no spatial autocorrelation and a random patterning (nations close together sometimes display similar levels of individualism and sometimes do not).

There are a number of alternative ways of constructing the weights matrix. By convention, the diagonal elements of w, the weights matrix, are set to 0, but the off-diagonals may be encoded in a variety of ways depending on the data being modeled. This is important because different encodings of w lead to different numerical values for I. It is therefore only possible to directly compare values of I for a given encoding of w. In this study, for comparability, we used a four nearest neighbors weights matrix throughout which is discussed further under the Method section.

The spatial autocorrelation methods described above can be used to map cultural variation within any conceptual space. Weights matrices are generic proximity structures, and any conception of distance between a pair of objects can be implemented as a weights matrix, and used as the basis for computing spatial autocorrelations. Consequently a weights matrix can be created for each of our five contextual variables and Moran’s I used to quantify the degree of spatial autocorrelation between the contextual variables and cultural dimensions. If, say, religious ideology is a root cause or consequence of culture then one would expect countries of similar religious profiles to share some similar values.

As spatial autocorrelation reflects any form of spatial patterning (linear or curvilinear), failure to find the expected relationships questions the earlier methods and conclusions. As the coefficients are comparable within studies, they indicate the relative degree of association between the geographical and ecosocial factors and the dimensions of national culture. And, to the extent that samples of nations and variables are comparable across studies then between-study comparisons can be made. For example, if Triandis and Hofstede’s individualism scales are measuring similar cultural values, then one would expect similar spatial patterning with the context variables.

Method

Measures of national culture

National measures of power distance, individualism, uncertainty avoidance, and masculinity for 50 countries were taken from Hofstede (2001, p. 500). National measures of Schwartz’s (1999) cultural values were taken from the Schwartz Value Survey (Israel Social Sciences Data Centre, 2005). For comparison with Georgas et al. (2004), we also conducted a principal components analysis to extract two factors from the Schwartz values which we called Autonomy and Hierarchy. Our factors look very similar to those obtained by Georgas et al. (2004) Our Autonomy factor was bipolar with strong positive loadings for affective and intellectual autonomy and a strong negative loading for embeddedness (conservatism). The second factor had strong +ve loadings for mastery and hierarchy with strong −ve loadings for harmony and egalitarianism. Georgas et al.’s two factors explained 71% of the variance (31 countries) and our two factors explained 77% of the variance (using a larger sample of Schwartz countries).

Societal cynicism and dynamic externality scores were taken from Bond et al. (2004), and scores for survival versus self-expression and traditional versus secular rational were taken from the Inglehart-Welzel map of cultures (Inglehart & Welzel, 2005). Scores for Triandis’ individualism were taken from Diener, Diener, and Diener (1995) and scores for the GLOBE Study societal cultural practices and values were taken from House, Hanges, Javidan, Dorfman, and Gupta (2004). Scores for the loyal utilitarian versus involvement and conservatism versus egalitarian commitment dimensions described in Smith et al. (1996) were kindly provided by Peter Smith. And finally, scores for in-group favoritism for 178 countries were taken from Van de Vliert (2011)

Geographical and ecosocial indices

Geographical distance was measured as the great circle distance (km) between the capital cities of each pair of countries. Distance from the equator was the shortest great circle distance between the capital cities and the equator.

The indicators of climate space were Van de Vliert’s measure of thermal climate demands (Van de Vliert, 2009) and 30-year averages (1961-1990) of precipitation, cloud cover, and vapor pressure, taken from the Tyndall Centre data set TYN CY 1.1 (Mitchell, Hulme, & New, 2002). The indicators were converted to z scores prior to calculating the Euclidean distances, so that each indicator contributed equally to the distance in climate space.

The indicators of socioeconomic space were education, life expectancy, and per capita GDP, which are the indicators used to construct the United Nations Human Development Index. Data for 2003 was taken from the Human Development Report 2005 (United Nations Development Programme, 2005). As for the climate matrix, these indicators were converted to z scores for the purpose of constructing distances between nations.

The primary data for the religious matrix were national populations for 15 different religious categories (for 2005) obtained from the Association of Religion Data Archives (Grim and Finke, nd.). The religious categories were, Bahai, Buddhist, Chinese Universist, Christian, Confucianist, Ethnoreligionist, Hindu, Jain, Jewish, Muslim, Shintoist, Sikh, Spiritist, Taoist, and Zoroastrian. The number of adherents per 100,000 of the population was computed for each nation, and Pearson’s Φ² was used to index the similarity between nations.

Benchmark variables

Temperature and precipitation for the same set of countries as included in the particular study were calculated as benchmark variables. Annual values of temperature (°C) and precipitation (mm) averaged over the period 1961-1990 were taken from the Tyndall Centre data set TYN CY 1.1 (Mitchell, Hulme, & New, 2002). As nations close together geographically are known to be similar in temperature and rainfall, these two variables serve as reference points against which the magnitudes of the cultural autocorrelations can be compared.

Analysis

Moran’s I was computed for each national dimension, using weights matrices representing respectively, geographical proximity, similarity in latitude (distance from the equator), similarity in climate, national development, and religious profile. Moran’s I was also calculated between geographical proximity and temperature and precipitation to act as benchmark variables.

In this work, we used a nearest neighbor matrix in which the 4 nearest neighbors were assigned the value 1 and all other nations coded 0. Our choice of encoding for the weights matrices was guided by simplicity and the desire for comparability between matrices, but was not the only possibility. We judged that fewer neighbors would constitute a too narrow definition of proximity, and that more would be too broad. Other choices are possible. Experiments on Hofstede’s (2001) dimensions with two and eight neighbors showed that as expected using fewer neighbors increases the autocorrelations, and using more decreases them; the same overall patterns persisted, but with eight neighbors the differences between dimensions were generally flattened. Another choice would be to use a continuous function of proximity rather than to dichotomize the scores. Again, experiments showed that an exponential decay function gave similar results to those presented here (these are given in Gelade & Dobson, 2009). We are therefore reasonably assured that our results are not simply an artifact of the particular encoding function we chose.

The Quadratic Assignment Procedure (QAP; 5,000 permutations) was used to determine the one-tailed significance of the spatial autocorrelation coefficients (Hubert & Schultz, 1976).

Results

The spatial autocorrelations between the five geographical and ecosocial context variables and the cultural dimensions included in this study are given in Table 1. Comparison between the different samples is somewhat complicated by the fact that different researchers used different sets of countries. All the samples covered approximately the same area of the globe, except that Smith et al.’s (1996) sample did not extend as far west as the others. There were also differences in regional emphasis. Hofstede’s (2001) sample of countries for example did not include any countries within the former Soviet Union, and societies in Africa and the Middle East were underrepresented. However, the inclusion of benchmark variables (temperature and precipitation) allows some indication of the degree of between-sample comparability. Differences in the spatial characteristics of the samples would be expected to cause differences in the observed autocorrelations for the benchmark variables. In fact, for the national samples, Moran’s I for temperature ranges between .47 and .74, and for precipitation between .52 and .73, differences that are not negligible, but which imply at least some degree of comparability between the samples.

Table 1.

Spatial Autocorrelations (Moran’s I) for Geographical and Ecosocial Proximities

Cultural system and dimension	Benchmark variables	Geographic proximity	Similarity in latitude	Similarity of climate	Similarity of development	Similarity of religious profile
Schwartz (2005) N = 73
Temperature	.59***
Precipitation	.53***
Harmony		.25**	.12	.19 **	.10	.28***
Conservatism		.50***	.44***	.43 ***	.64***	.63***
Hierarchy		.43***	.19*	.25**	.28***	.46***
Mastery		.12*	.14*	−.11	−.02	.04
Affective autonomy		.42***	.42***	.40***	.49***	.55***
Intellectual autonomy		.60***	.45***	.48***	.55***	.49***
Egalitarian commitment		.55***	−.04	.18**	.40***	.26***
Autonomy factor		.48***	.48***	.45***	.54***	.60***
Hierarchy factor		.37***	.08	.04	.21**	.28***
Hofstede (2001) N = 50
Temperature	.64***
Precipitation	.65***
Power distance		.27**	.37***	.33 ***	.36***	.35***
Individualism		.58***	.59***	.49 ***	.70***	.59***
Uncertainty avoidance		.31***	.19*	.11	.17*	.07
Masculinity		.13*	.08	−.03	.07	−.06
Long-term orientation (N =23)		.45***	.03	−.06	.08	.46***
Inglehart & Welzel (2005) N = 78
Temperature	.59***
Precipitation	.52***
Survival versus self-expression		.55***	.00	.21**	.65***	.41***
Traditional versus secular rational		.68***	.58***	.56***	.35***	.45***
Bond et al. (2004) N = 42
Temperature	.47***
Precipitation	.69***
Dynamic externality		.44***	.31**	.30***	.53***	.49***
Societal cynicism		.06	−.10	−.17	.00	.05
Triandis (Diener, Diener, & Diener, 1995) N = 53
Temperature	.62***
Precipitation	.60***
Individualism		.51***	.37***	.35***	.64***	.64***
Smith, Dugan, and Trompenaars (1996) N = 42
Temperature	.67***
Precipitation	.56***
Conservatism versus egalitarianism commitment		.36***	.08	−.03	.39***	.25***
(continued)
Utilitarian versus loyal involvement		.50***	.43***	.65***	.20*	.29***
Van de Vliert (2011) N = 178
Temperature	.74***
Precipitation	.73***
Ingroup favoritism		.43***	.19***	.23***	.61***	.24***
Average		.41	.25	.24	.36	.35

p ≤ .05. **p ≤ .01. ***p ≤ .001.

Table 1 reveals a considerable degree of organization within geographical as well as ecosocial space. Excluding the Schwartz (1999) factors to avoid double counting, 78% of the ecospatial autocorrelations we report are significant at the .05 level or higher. This is not surprising as the contextual variables were selected because previous authors had found them to be associated with dimensions of national culture. The means are significantly different on a repeated measures ANOVA (Wilkes γ = .433, p < .001) and post hoc tests find that climate and latitude are significantly weaker organizing structures than geography, development, and religion, although the differences between geography, religion, and development are not significant.

To grasp the substantive meaning of these results, note that the largest spatial autocorrelations are similar in size to those for temperature and precipitation; in other words, the relationship between the contextual variables and cultural dimensions can be as strong as the relationship between geographical space and temperature or rainfall.

A small minority of cultural dimensions show limited geographical or ecosocial organization. These are societal cynicism, mastery, and masculinity. According to Bond et al. (2004) in cynical countries “citizens believe that they . . . are suppressed by powerful others and subjected to the depredations of willful and selfish individuals, groups and institutions” (p. 566). Bond et al. speculate that the determinants of this dimension might be a national history of civil and international warfare, of colonization by foreign powers, of major political conflicts, or of frequent economic disasters—to which we would add incompetent or repressive governments and institutions. If so, we would expect levels of societal cynicism to be specific to individual countries, and thus largely independent of geographical or ecosocial factors. However, there seems no compelling reason why these arguments should apply to mastery and masculinity. Georgas et al. (2004) did not include Bond et al.’s study in their research and did not separate out Schwartz’s (1999) mastery, but did find as we have here weak effects for Hofstede’s (2001) masculinity across a wide range of contextual variables.

Geographical proximity

As can be clearly seen most cultural dimensions display a significant degree of geographical organization (average I = 0.41). As was hypothesized, because of common descent and cultural borrowing geographical proximity has a general and significant impact upon national culture. If one wants to know the likely culture of a nation, one port of call should be the culture of its nearest neighbors.

Of course it is not the geographical proximity of capital cities per se that results in similarity of culture but that proximity promotes interaction between nations through mechanisms such as conquest, immigration, trade, tourism, and so on. Furthermore, countries close together are likely to have similar climates, religious profiles and levels of national development. Our findings support Hypothesis 1 and the conclusions of Ronen and Shenkar (1985), Schwartz (1999), and Georgas et al. (2004) that countries of similar culture tend to be geographically contiguous.

Similarity of physical climate

Hypothesis 2 is largely supported. Schwartz’s (1999) affective and intellectual autonomy, Hofstede’s (2001) power distance and individualism, Smith et al.’s (1996) utilitarian versus loyal involvement, Bond et al.’s (2004) dynamic externality, Inglehart & Baker’s (2000) survival versus self-expression, Van de Vliert’s in-group favoritism, and GLOBE’s in-group collectivism (practice) are all significantly spatially organized by physical climate. However, we find no evidence for the spatial organization of Hofstede’s (2001) masculinity or Smith et al.’s (1996) conservatism versus egalitarian commitment dimension.

We find evidence for spatial organization by climate for a number of dimensions that were not hypothesized. Like Georgas et al. (2004) we find no spatial patterning with climate for Schwartz’s (1999) hierarchy factor, but we do find a significant relationship for the hierarchy value which given its association with power distance is perhaps not surprising. Unlike Georgas et al. (2004), we find a highly significant spatial patterning for Inglehart and Baker’s (2000) traditional versus Secular Rational Scale for physical climate. The difference between the findings of the studies is quite marked. Georgas et al. (2004) report a nonsignificant effect whereas we find Moran Is of 0.56 for climate (and .58 for distance from the equator both p < .001). Although the composition of measures of ecology and climate differ between the two studies, it is more likely that the difference results from either a difference in the sample of nations used (Georgas et al. (2004) used 29 nations from Inglehart [1997], whereas we used 78 nations from Inglehart & Welzel, 2005) or from the difference in method used.

In this research, we, as others, have used averaged annual measures of precipitation, cloud cover and vapor pressure as part of our measure of climate. Recent work questions the psychological significance of such annual measures of climate. For example, Tavassoli (2009) provides evidence that seasonal effects are important in the origins of culture. Van de Vliert argues that for warm blooded humans it is the departure from 22°C that creates climatic stress and his Thermal Demand Scale, which we have used in this research, considers both winter and summer temperature deviations. Van de Vliert’s underpinning rationale seems to be sound, however, the constructed scale does lead to some counter-intuitive scores for thermal demand. It is surprising to find that United Arab Emirates, Libya, Kuwait, and India are considered more temperate climates than the Netherlands, France, Ireland, or United Kingdom.

Similarity of latitude

Table 1 reveals that as hypothesized Hofstede’s (2001) measures of power distance and individualism are spatially patterned with respect to distance from the equator. However, we find no significant patterning for masculinity and distance from the equator.

Previous authors have used distance from the equator as a proxy for temperature: nations close to the equator are hot and those far away are cold. This line of reasoning neglects the impact of altitude, land mass, and ocean and atmospheric currents on temperature. For example, both Amsterdam and Warsaw are the same distance from the equator (52°N) but with significantly different summer and winter temperatures and climates. Distance from the equator has also been equated with the harshness of the environment: harsh environments are those close to the equator (hot) or those far away (cold) with temperate climates in between. However, this line of reasoning seems to neglect the fact that the equator while hot is the site of the world’s tropical rainforests which offer a plentiful supply of food. The harsh environment of, say, the Saharan desert is 20°N of the equator, and the seas off Iceland and Norway are teaming with fish, in contrast to the food available in the Gobi desert some 15° to the south. In short, distance from the equator seems to be a poor metric for assessing either temperature or the harshness of the environment. Despite this, the spatial patterning of cultural values with latitude is extremely similar to that with physical climate; only Inglehart & Baker’s (2000) survival versus self-expression dimension reveals any significant difference.

National development & religious profile

We report the findings for the spatial patterning of similarity of level of national development and religious profile together. With only a few exceptions, if a cultural scale reveals significant patterning for national development, it also has a significant patterning for religious profile. The results in Table 1 reveal, as previous authors, a very significant association for national development and religion with national culture. The average Moran’s I for both is 0.35, and most of the cultural dimensions are influenced significantly by both. Only Schwartz’s (1999) harmony and Hofstede’s (2001) long-term orientation appear to be exceptions both being associated with religion but not with national development.

The results for both national development and religion largely support the specific findings of previous authors (Hypotheses 4 & 5). Thus, we find that Inglehart & Baker’s (2000) traditional versus secular rational and survival versus self expression, Hofstede’s (2001) individualism and power distance, Smith et al.’s (1996) conservationism versus egalitarian commitment, Bond et al.’s (2004) dynamic externality, and, with the exception of mastery, all of the Schwartz (1999) scales are significantly spatially organized with respect to national development and religious profile. Nations at similar levels of development and with similar religious profiles possess similar cultures.

The only significant difference between our findings and those previously is that we do not find a significant association between Hofstede’s (2001) uncertainty avoidance and religion, whereas Georgas et al. (2004) do find a significant effect. Hofstede (2001) postulates that in countries with a higher need for uncertainty avoidance that religions that stress absolute certainties and are intolerant of other religions will be found. He concludes that uncertainty avoidance and religion do appear to be meaningfully related but that other influences are at work as well. The reason why we fail to find an association between religion and uncertainty avoidance may be for one or a combination of three factors, namely, differences in the sampling of nations (Georgas et al. (2004) sample 66 whereas we have only included 50 nations), differences in the construction of the religious profile index (Georgas et al. (2004) use only 7 religious categories whereas our index comprises 15 categories), or difference in method of analysis (Georgas et al. (2004) use ANOVA whereas we have used spatial autocorrelation). We conclude that the association between uncertainty avoidance and religion is empirically unproven and theoretically unclear.

GLOBE values and practices

There are important differences between House et al.’s (2004) GLOBE measures of cultural dimensions and the other cultural dimensions included in this research. First, GLOBE measures two cultural manifestations “how things should be” which they label as values and “how things are” which they label as practices. These two aspects are measured by simply replacing “should be” with “are/is” in the same question item. Second, GLOBE focuses their questions at a societal level, in essence, “what kind of society should this be/is this.” GLOBE therefore aggregates individual beliefs about how things should be in a society (values) and individual beliefs about how things are (practices). This approach is in contrast to those studies that aggregate personal values to represent national culture. For example, Schwartz (1999) “the guiding principles in my life” or Hofstede (2001) “how important is it to you” or Inglehart and Baker (2000) “respondent favors more respect for. . . .” Although personal values and beliefs about how a society should be are likely to overlap, they are not the same thing. This is probably a reason why House et al. (2004) report mixed convergent validity for their scales when compared with the scales of Hofstede (2001) and Schwartz (1999).

The spatial analysis of GLOBE values and practices for the five geographical and ecosocial indices are given in Table 2. Comparison of the autocorrelations for GLOBE values in Table 2 with those for the other values studied in Table 1 suggests that the GLOBE values show a rather weaker degree of ecosocial organization. The average I for GLOBE values is lower than that in Table 1 for the all the contextual variables studied (mean I for GLOBE is 0.21 compared with 0.32 for Table 1, t test, two-tailed p < .01). Only religion displays a similar degree of spatial patterning in the two tables. National development has much less association with the GLOBE values than it does with the values of the other studies. The most likely reason for the difference between GLOBE and the other studies is that they are measuring different aspects of national culture. For example, there are significant differences between the spatial organization of power distance and uncertainty avoidance as measured by Hofstede (2001) and GLOBE. We conclude that peoples’ beliefs about how things should be in a society are more influenced by the society’s religion than by its level of national and economic development, whereas the personal values of individuals in a society are more or less equally influenced by both.

Table 2.

Spatial Autocorrelations (Moran’s I) for GLOBE Values and Practices

Cultural system and dimension	Benchmark variables	Geographic proximity	Similarity in latitude	Similarity of climate	Similarity of development	Similarity of religious profile
GLOBE (House, Hanges, Javidan, Dorfman, & Gupta, 2004) N = 58
Temperature	.59***
Precipitation	.64***
Values
Assertiveness		.24**	−.06	−.01	.06	.11
Institutional collectivism		.31***	.20*	.42***	.21**	.21**
In-group collectivism		.44***	.11	−.06	−.01	.39***
Future orientation		.45***	.38***	.40***	.35***	.38***
Gender egalitarianism		.39***	.13	.10	.25**	.59***
Humane orientation		−.08	−.04	.04	.02	.00
Performance orientation		.34***	.03	−.03	−.02	.27***
Power distance		.05	.13	−.06	.02	.32***
Uncertainty avoidance		.36***	.31***	.37***	.70***	.43***
Average		.28	.13	.15	.18	.30
Practices
Assertiveness		.11	−.04	−.07	−.12	−.01
Institutional collectivism		.38***	−.07	.14*	−.05	.15*
In-group collectivism		.35***	.21*	.36***	.70***	.38***
Future orientation		.09	−.06	.07	.19**	.14*
Gender egalitarianism		.11	.12	.13	.15*	.04
Humane orientation		.13	.06	−.07	.26**	.05
Performance orientation		.16*	−.14	.04	.01	.05
Power distance		.05	.01	.09	.27***	.02
Uncertainty avoidance		.22**	.02	.04	.38***	.28***
Average		.18	.01	.08	.20	.12

p ≤ .05. **p ≤ .01. ***p ≤ .001.

Table 2 also reveals a significant difference between GLOBE values and practices. The average Moran’s I for practices is 0.12 compared with 0.21 for GLOBE values (t test, two-tailed p < .05). With the noticeable exception of the association between national development and in-group collectivism, GLOBE practices are significantly less spatially organized than GLOBE values. This is an interesting finding given that the same respondent is being asked the same question twice with “should be” simply replaced with “are/is.” In contrast to values, how things are in a society is often only weakly associated with the culture of the nearest geographical, climatic, or religious neighbors. Religion, in particular, appears to have far less influence on how things are in a society than how things should be (t test, two-tailed p < .05).

Comparison of GLOBE practices and values also reveals numerous instances where the practice and the value of the same dimension are organized in different ways. For example, the practice of in-group collectivism (defined by House et al. (2004) as the degree to which individuals express pride, loyalty, and cohesiveness in their families) is strongly associated with national development, but the corresponding value is not. Here, we might suppose that changing levels of economic activity produce corresponding changes in collectivist behaviors (or vice versa), although the underlying value orientations remain unchanged and independent of the level of development. National and economic development would appear to have a significant practical effect on the family in a society, for example, in the extent to which aging parents generally live at home with their children, but not on whether this practice is valued.

Likewise, the value of gender egalitarianism is strongly organized with respect to both religion and geography, but the corresponding practice is not. The explanation could be that countries with similar religious profiles have similar teachings with regard to how men and women in society “should be” treated. However, the realization of such values at a societal level is dependent on the ability of the religions to influence governments and policy makers. Similarly, near neighbors have similar views on how the sexes should be treated presumably because of the interchange of ideas and peoples across national boundaries; however, the presence or absence of national policies regarding, say, equal opportunities, are determined by governments or other dominant groups within the nation state.

Finally, we note that performance orientation value (defined by House et al 2004 as the degree to which a collective encourages and rewards group members for performance improvement and excellence) is not clustered according to national development but is significantly clustered according to religion. The corresponding practice is also not associated with economic development. This questions how these dimensions should be interpreted.

Discussion

This research has demonstrated that most dimensions of national culture are organized within geographical and ecosocial space. Countries that are geographically close, those with similar climates and religious profiles and those with similar levels of national development often have similar cultural values. Although this general idea is not particularly new, this article is, to our knowledge, the first to offer a quantitative account of the organization of national values in which geographical and ecosocial space are considered on the same footing. A key finding is that overall, national cultural dimensions are about equally dependent on location, development, and religion, but overall, less strongly dependent on the physical climate as measured here. We find that some dimensions are as strongly spatially organized as temperature and rainfall and that most show lesser, but still significant, degrees of organization.

For the most part the findings of previous authors, namely, Inglehart and Baker (2000), Schwartz (1999), Hofstede (2001), Smith et al. (1996), Bond et al. (2004), and Georgas et al. (2004) regarding the association between geographical and ecosocial variables and specific dimensions of national culture are supported. The notable exceptions to previous findings are that we find no association between egalitarian commitment (Smith et al., 1996) or masculinity (Hofstede, 2001) and physical climate and distance from the equator, but do find a significant association for Inglehart and Baker’s (2000) traditional versus secular rational dimension. Furthermore, we do not find any association between Hofstede’s (2001) uncertainty avoidance and religion. The likely reasons for these different findings are differences in the samples of nations used, differences in the constructed indices of the contextual variables, or differences in the method of analysis used. In contrast to previous authors, we have used spatial autocorrelation to analyze the data sets and thus have avoided the pitfalls associated with non-independence of observations. As Eff (2004) has demonstrated, this is a significant issue for international data sets of social and cultural phenomena and despite the ingenuity of previous designs may well have resulted in biased effect sizes and inappropriate inferences.

A few dimensions of national culture, namely, Schwartz’s (1999) mastery, Hofstede’s (2001) masculinity, and Bond et al.’s (2004) societal cynicism, revealed little spatial patterning with any of the contextual variables used in this study.

Two interesting findings come from the inclusion of GLOBE values and practices in this research, both of which underline the value of measuring the values and practices of a nation separately. First, this research finds that GLOBE values are differently and significantly less spatially organized than the other value dimensions included in this study. Those studies that have used aggregated personal values to represent national culture such as Hofstede (2001), Schwartz (1999), and Inglehart and Baker (2000) reveal significant and more or less equal association with national development and religion. GLOBE values, which do not measure personal values but rather individual beliefs about how a society should be, are, in contrast, primarily organized only by religion. We infer that people’s beliefs about how a society should be are primarily influenced by the religious profile of that society, although what people in a society consider important for themselves is, in addition, influenced by the level of the nation’s development. Perhaps questions such as “how important is it to you” or “the respondent favors” engage instrumental values linked to individual needs. Certainly, one would expect cultural dimensions that tap individual needs to be related to the development of the nation.

Second, we find that GLOBE values and practices reveal different patterns of spatial organization. GLOBE practices are typically, less strongly organized than GLOBE values. We find that the practice and value of the same dimension frequently reveals very different degrees of spatial organization and that religion, the major organizing variable for GLOBE values, has little association with GLOBE practice—how things are in a society. This is despite the fact that the same respondents are being asked the same questions with “should be” being replaced with “are or is.” This is an important finding that suggests that practice does not evolve simplistically from national values or vice versa. Practice and value may be more independent of each other than other authors have implied; indeed, in some cases the “collective programming of the mind,” as measured by the aggregated personal values of teachers, middle managers, and students (typical respondents to the surveys), bears little resemblance to practice in a nation. This is hardly surprising, for such groups do not frame policy and practice in a society: they reflect and enact it. Policy and practice is determined by the dominant groups in society, especially governments, who may have a different agenda and indeed values. To us, this seems a plausible explanation as to why values (how things should be) typically transcend national boundaries, but practices (how things are) often do not. Values are relatively enduring characteristics of broad supra-national populations, derived from interaction, shared history, and teaching, whereas practices reflect factors such as government legislation, policies, and resourcing decisions, influences which tend to be demarcated by national boundaries. To our mind, this analysis questions the adequacy of measuring national culture by aggregated personal values alone. National culture should also be accessed, as House et al. (2004) have done, by directly asking the question “What kind of society is this?”

Finally, some observations on the strengths and limitations of Moran’s I used to investigate differences in national culture. First, it is a statistic that is appropriate for investigating association between nonindependent observations. Second, it reflects nonrandom spatial patterning in the variables under investigation, and because the “independent,” that is, contextual, variable is a similarity measure, it can detect similarities in the “dependent” (i.e., focal) variable arising from shared multidimensional attributes among the units being analyzed. In this sense, it is a powerful test of association that bypasses the need to cluster the analytical units to identify relationships. If there is a relationship between the focal and contextual variables spatial autocorrelation will reveal it and the QAP procedure can be used to test the significance of the relationship. However, like any measure of association, it is limited in the conclusions that can be made, which are of the form “nations similar in X tend also to be similar in Y.” No inference of cause can be made, and the obtained relationship between X and Y may be the result of the influence of an omitted variable Z. However, being free of the assumption of independence of observations, it can be used to undertake comparative analyses of the relative significance of contextual factors within and between studies, and to explore data sets of cross-cultural variables to generate hypotheses.

The results reported in this research reveal, like Eff (2004) and Dow and Eff (2008), that many cross-cultural variables are clustered within ecosocial space (many significantly so) and therefore prey to Galton’s problem. Our focus in this article has been on spatial autocorrelation and more specifically on Moran’s I. However, there is a range of spatial statistics available to help analyze autocorrelated data. For example, Loftin and Ward (1983) and Dow, Burton, Reitz, and White (1984) develop spatial regression techniques, and Eff (2004) demonstrates how spatial autocorrelation can be used to specify a least squares model. We suggest that spatial autocorrelation, and spatial statistics generally, are neglected and underused methods in cross-cultural research.

Footnotes

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Andersen

P. A.

(1985). Nonverbal immediacy in interpersonal communication. In Siegman

A. W.

Feldstein

(Eds.), Multichannel integrations of nonverbal behavior (pp. 1-36). Hillsdale, NJ: Erlbaum.

Anderson

C. A.

(1989). Temperature and aggression: Ubiquitous effects of heat on occurrence of human violence. Psychological Bulletin, 106, 74-96.

Baller

R. D.

Anselin

Messner

S. F.

Deane

Hawkins

D. F.

(2001). Structural covariates of U.S. county homicide rates: Incorporating spatial effects. Criminology, 39, 561-588.

Baskent

E. Z.

(1999). Controlling spatial structure of forested landscapes: A case study towards landscape management. Landscape Ecology, 14, 83-97.

Beck

Gleditsch

N. S.

Beardsley

(2006). Space is more than geography: Using spatial econometrics in the study of political economy. International Studies Quarterly, 50, 27-44.

Bell

, (1973). The coming of post-industrial society. New York, NY: Basic Books.

Bond

M. H.

Leung

Tong

K-K.

de Carrasquel

S. R.

Murakami

Yamaguchi

. . . Stetovska

(2004). Culture-level dimensions of social axioms and their correlates across 41 cultures. Journal of Cross-Cultural Psychology, 35, 548-570.

Cliff

A. D.

Ord

J. K.

(1981). Spatial processes, models, and applications. London, UK: Pion.

Cliff

A. K.

Ord

J. K.

(1973). Spatial autocorrelation. London, UK: Pion.

10.

Diener

(1995). Factors predicting the subjective well-being of nations. Journal of Personality and Social Psychology, 69, 851-864.

11.

Dow

M. M.

(2007). Galton’s problem as multiple network autocorrelation effects. Cross-Cultural Research, 41, 336-363.

12.

Dow

M. M.

Burton

M. L.

Reitz

White

D. R.

(1984). Galton’s problem as network autocorrelation. American Ethnologist, 11, 754-770.

13.

Dow

M. M.

Eff

E. A.

(2008). Global, regional and local network autocorrelation in the standard cross-cultural sample. Cross-Cultural Research, 42, 148-171.

14.

Eff

E. A.

(2004, February). Spatial and cultural autocorrelation in international datasets. Murfreesboro, TN: Department of Economics and Finance Working Paper Series, Middle Tennessee State University.

15.

Ember

C. R.

Ember

(2009). Cross-cultural research methods (2nd ed.). Lanham, MD: Altamira.

16.

Ember

C. R.

(2000). Testing theory and why the “unit of analysis” problem is not a problem. Ethnology, 39, 349-363.

17.

Gelade

G. A.

(2008). The geography of IQ. Intelligence, 36, 495-501.

18.

Gelade

G. A.

Dobson

(2009). The cross-cultural geography of nations: Using the spatial autocorrelation coefficient to test Tobler’s first law. Retrieved from http://www.cassknowledge.com/node/4052

19.

Georgas

Berry

J. W.

van

Vijver

F. J. R.

Kagitcibasi

Poortinga

Y. H.

(2006). Families across cultures: A 30-nation psychological study. Cambridge, UK: Cambridge University Press.

20.

Georgas

van de Vijver

F. J. R.

Berry

J. W.

(2004). The ecocultural framework, ecosocial indices, and psychological variables in cross-cultural research. Journal of Cross-Cultural Psychology, 35, 74-96.

21.

Gilmore

D. D.

(1990). Manhood in the making. New Haven, CT: Yale University Press.

22.

Gouveia

V. V.

Ros

(2000). Hofstede and Schwartz’s models for classifying individualism at the cultural level: Their relation to macro-social and macro-economic variables. Psicothema, 12, 25-33.

23.

Grim

B. J.

Finke

(nd.). Cross-national data: Religion indexes, religious adherents, and other data (Datafile). Retrieved from http://www.thearda.com/Archive/Files/Downloads/INTL2003_DL.asp

24.

Hafting

Fyhn

Molden

Moser

E. I.

(2005). Microstructure of a spatial map in the entorhinal cortex. Nature, 436, 801-806.

25.

Heywood

J. S.

(1991). Spatial analysis of genetic variation in plant populations. Annual Review of Ecology and Systematics, 22, 335-355.

26.

Hofstede

(2001). Cultures consequences: Comparing values, behaviours, institutions and organizations across nations (2nd ed.). Thousand Oaks, CA: SAGE.

27.

Hofstede

Bond

M. H.

(1988). The Confucius connection: From cultural roots to economic growth. Organizational Dynamics, 16, 4-21.

28.

Hofstede

G. J.

(2005). Cultures and organizations: Software of the mind (2nd ed.). New York, NY: McGraw-Hill.

29.

House

R. J.

Hanges

P. J.

Javidan

Dorfman

P. W.

Gupta

(Eds.). (2004). Culture, leadership, and organizations. The GLOBE study of 62 societies. Thousand Oaks, CA: SAGE.

30.

Hubert

Schultz

(1976). Quadratic assignment as a general data analysis strategy. British Journal of Mathematical and Statistical Psychology, 29, 190-241.

31.

Huntington

S. P.

(1996). The clash of civilizations and the remaking of world order. New York, NY: Simon & Schuster.

32.

Inglehart

(1997). Modernization and postmodernization. Princeton, NJ: Princeton University Press.

33.

Inglehart

Baker

W. E.

(2000). Modernization, cultural change, and the persistence of traditional values. American Sociological Review, 65, 19-51.

34.

Inglehart

Welzel

(2005). Modernization, cultural change, and democracy: The human development sequence. New York, NY: Cambridge University Press.

35.

Israel Social Sciences Data Centre. (2005). Schwartz Value Survey (SVS) (Data set 0789). Jerusalem, Israel: Hebrew University of Jerusalem

36.

Loftin

Ward

S. K.

(1983). A spatial autocorrelation model of the effects of population density on fertility. American Sociological Review, 48, 121-128.

37.

Mitchell

T. D.

Hulme

New

(2002). Climate data for political areas. Area, 34, 103-112.

38.

Moran

P. A. P.

(1950) Notes on Continuous Stochastic Phenomena. Biometrika, 37 (1), 17-23.

39.

Moscone

Knapp

(2005). Exploring the spatial pattern of mental health expenditure. Journal of Mental Health Policy and Economics, 8, 205-217.

40.

Murdock

G. P.

White

D. R.

(1969). Standard cross-cultural sample. Ethnology, 8, 329-369.

41.

Naroll

(1965). Galton’s problem: The logic of cross-cultural research. Social Research, 32, 428-451.

42.

Pennebaker

J. W.

Rime

Blankenship

V. E.

(1996). Stereotypes of emotional expressiveness of northerners and southerners: A cross-cultural test of Montesquieu’s hypotheses. Journal of Personality and Social Psychology, 70, 372-380.

43.

Robbins

M. C.

DeWalt

B. R.

Pelto

P. J.

(1972). Climate and behavior: A biocultural study. Journal of Cross-Cultural Psychology, 3, 331-344.

44.

Ronen

Shenkar

(1985). Clustering countries on attitudinal dimensions: A review and synthesis. Academy of Management Review, 10, 435-454.

45.

Rosenberg

M. S.

Sokal

R. R.

Oden

N. L.

DiGiovanni

(1999). Spatial autocorrelation of cancer in Western Europe. European Journal of Epidemiology, 15, 15-22.

46.

Schwartz

S. H.

(1999). A theory of cultural values and some implications for work. Applied Psychology: An International Review, 48, 23-47.

47.

Schwartz

S. H.

(2005) Schwartz Values Survey, Israel Social Sciences Data Center.

48.

Smith

P. B.

Dugan

Trompenaars

(1996). National culture and the values of organizational employees: A dimensional analysis across 43 nations. Journal of Cross-Cultural Psychology, 27, 231-264.

49.

Tavassoli

(2009). Climate, psychological homeostasis and individual behaviors across cultures. In Wyer

R. S.

Chiu

C-Y.

Hong

Y-Y.

(Eds.), Understanding culture: Theory, research and application (pp. 211-221). Psychology Press.

50.

United Nations Development Programme. (2005). Human development report. Retrieved from http://hdr.undp.org/reports/global/2005/

51.

Upton

Fingleton

(1985). Spatial data analysis by example. New York, NY: John Wiley.

52.

Van de Vliert

(2007). Climatoeconomic roots of survival versus self-expression cultures. Journal of Cross-Cultural Psychology, 38, 156-172.

53.

Van de Vliert

(2009). Climate affluence and culture. New York, NY: Cambridge University Press.

54.

Van de Vliert

(2011). Climato-economic origins of variation in ingroup favoritism. Journal of Cross-Cultural Psychology, 42, 494-515.

55.

Van de Vliert

Huang

Parker

P. M.

(2004). Do colder and hotter climates make richer societies more, but poorer societies less, happy and altruistic? Journal of Environmental Psychology, 24, 17-30.

56.

Van de Vliert

Schwartz

S. H.

Huismans

S. E.

Hofstede

Daan

(1999). Temperature, cultural masculinity and domestic political violence: A cross-national study. Journal of Cross-Cultural Psychology, 30, 291-314.