Grid,Group,and Grade

Introduction

A well-known and often-used framework for the analysis of culture in the political–administrative world is Cultural Theory (CT), also known as the Grid–Group theory, developed by Mary Douglas (among others, 1970, 1978, 1992, 1996) and elaborated on by many other scholars (among others, Coyle & Ellis, 1994; Douglas & Ney, 1998; Douglas & Wildavsky, 1982; Hood, 1998; Thompson, Ellis, & Wildavsky, 1990; Thompson, Grendstad, & Selle, 1999; Verweij & Thompson, 2006). In this article, we will critically assess and explore attempts to operationalize and measure culture, in terms derived from Grid–Group theory, at the cross-national level.

At the core of the Douglasian approach to culture are two dimensions of sociality—Grid and Group—that harken back to the work of Durkheim, and four related cultural types (or cultual biases, or ways of life)—individualism, hierarchy, egalitarianism, atomism—that resonate with many other classifications of culture (Thompson et al., 1990). The resulting two-by-two Grid–Group typology has inspired two basic types of cultural analysis.

The first stays close to the anthropological tradition, transferred by the intellectual mother of the theory. This can take a form ranging from classic fieldwork to contemporary discourse analysis in its many versions (e.g., Gyawali, 2001; Hendriks, 1999; Hoppe & Peterse, 1993; Lodge & Wegrich, 2011; Mars, 1982). The emphasis is on the four types of culture, which are used as analytical coordinates, “idealtypes,” with which to compare empirical reality. Grid and Group are part of the multilayered analytical coordinates, in addition to other elements. Individualism as an idealtype, for instance, is assumed to be congruent with particular social relations (low Group/low Grid), particular values (equality of opportunity, choice, freedom), views of man (self-seeking), and views of nature (benign; Hendriks, 1999; Thompson, 2008; Verweij, 2011).

The second type of analysis is less concerned with qualitative “verstehen” (understanding) of culture and more with quantitative measurement, usually in extensive, large-scale, survey research designs (e.g., Boyle & Coughlin, 1994; Caulkins & Peters, 2002; Dake, 1990; Grendstad, 1999; Gross & Rayner, 1985; Kahan, 2012; Lockhart, 2011). The emphasis is on the two name-giving dimensions of the theory, Grid and Group, which are the focus of operationalization and subsequent measurement. Cultural assessment, here, is derived from a set of “axes,” on which units of analysis have a particular position, while in the other approach cultural typification connects to multilayered coordinates with an “onion” type of makeup.

Various scholars tried to measure Grid and Group and test implications of the theory along these empirical dimensions, although it is known that Douglas herself was rather wary of the detailed operationalization and measurement of Grid and Group. Her cultural typology was no more and no less than “a nice little typology that goes a long way in understanding the world around us” (Douglas, 1992, p. 137). And indeed, in this vein, many “worlds around us” have been effectively assessed in rich cultural detail, mainly in single-case, binary-case, or other small-N types of research.

Nevertheless, we think there are three good reasons to seek refinement of the other, “axis-oriented,” type of cultural analysis, starting from the Grid and Group dimensions. First, if any serious attempt to refine a theory deserves a fair chance to see how far it can come, then surely this one—elaborating on central dimensions of an influential theory—does, even though its main theoretician was not in favor of it. Second, Mary Douglas as well as other proponents of her CT have asserted that it is usable at different levels of analysis, those being either individuals, groups, or nations (Mamadouh, 1999; Swedlow, 2011). Especially in large-scale, cross-national research—the focus of this article—the use of measurable dimensions seems inevitable. Third, grave problems and challenges surround the operationalization and measurement of Grid and Group, which demand attention of the type that we seek to provide here.

In this article, therefore, we try to contribute to the discussion on the operationalization of Douglasian CT by answering the following questions: (a) What should be the starting point of operationalization: the four cultural types or the two cultural dimensions of Grid–Group Theory? (b) What are the societal cultural traits that Grid and Group represent, if dimension-based operationalization is the better alternative for large-scale cross-cultural research? (c) What are the challenges in operationalizing Grid and Group at the cross-national level? (d) How can we refine and extend the Grid and Group dimensions to properly measure and adequately explain cultural variances across countries? In the following sections, we attempt to answer these questions.

Operationalization of CT

Regarding the operationalization of CT, the first and foremost question is what should be operationalized to begin with: the proposed fundamental dimensions of sociality (Grid and Group) or the proposed essential “ways of life” (the four types that are also called “worldviews,” “cultural biases,” and “form of solidarity” in the CT literature)?

Inspired by the approach of Wildavsky and Dake (1990), most cultural theorists utilize survey items to operationalize four cultural biases. The most known and used instrument for operationalizing CT is a questionnaire designed by Karl Dake (1990, 1991) for measuring four worldviews. Some scholars indicate that Dake’s measures fail to display scale reliability and internal validity in empirical tests (Kahan, 2012; Marris, Langford, & O’ Riordan, 1998; Rippl, 2002). Rippl (2002) argues that “Dake’s instruments are inadequate measures of Cultural Theory” because some “correlations were found that do not conform to the theoretical assumptions of cultural theory” (pp. 153-154).

For example, the empirical tests on three measurements that utilized Dake’s questionnaire show strong positive correlations between hierarchy and individualism. This violates the theoretical assumption because these two worldviews have dissimilarity on both Grid and Group dimensions and expectedly they should have a negative correlation. Therefore, it indicates that Dake’s instrument is problematic for a consistent operationalization (Rippl, 2002). Using a different questionnaire but a similar approach, Coughlin and Lockhart (1998) also operationalized four worldviews. Similarly, their measurement suffers from a problem of nonconformity of correlations between worldview measures (Rippl, 2002).

Olli (2012) has more recently published an elaborate research on the survey-based operationalization of the four cultural biases. He concludes that “Cultural Theory still works rather poorly in surveys, and I believe I have pushed the limits of what can be done with Dake and Wildavsky’s approach to measuring cultural bias” (Olli, 2012, p. 504). However, he emphasizes that his “critiques of the assumptions apply only to analyses that rely on individual-level measurements of the four cultural biases, not the grid–group as two fixed dimensions. A different measurement model changes what one can potentially find” (Olli, 2012, p. 430).

It is proposed, and we support the idea, that for having a more valid and systematic measurement of CT, one should first operationalize the Grid and Group dimensions and then aggregate them to measure four cultural types (Boyle & Coughlin, 1994; Kahan, 2012; Kahan, Braman, Gastil, & Slovic, 2007; Rippl, 2002). This approach of developing the cultural types on the basis of the underlying dimensions seems more consistent with the two-dimensional typology introduced in CT. For large-scale cross-national research, it seems to be the more promising alternative.Before discussing the challenges with the existing operationalization of Grid and Group, we first briefly elaborate on the conceptions, definitions, and measures introduced for these two dimensions in the CT literature.

Definitions and Measures of Grid and Group

Grid and Group are two bipolar cultural dimensions that were introduced to “account for the distribution of values within a population” (Douglas, 2007, p. 2) and accordingly classify different ways of life or worldviews. Each dimension is supposed to represent some specific cultural traits that cannot be explained by another dimension.

Although there are various ways of expressing these general thoughts, most scholars using the theory would agree that Grid denotes the extent to which people’s thoughts and actions are regulated by position-related rankings, rules, and role prescriptions, while Group denotes the extent to which an individual’s life is absorbed and sustained by group membership (Coyle & Ellis, 1994; Douglas, 1992).

Grid and Group are not the only bipolar dimensions of culture in social science. There are several other cross-cultural theories that introduce or extract cultural dimensions conceptually or empirically. Comparison of the concepts and definitions of these many dimensions of culture reveal the similarity and differences of dimensions. Maleki and De Jong (2014) provide a summary and clustering of cultural dimensions, including Grid and Group. In their proposal of clustering the cultural dimensions based on the conceptual similarities and comparable definitions, Group is clustered with some other known cultural dimensions like individualism (vs. collectivism; Hofstede, 2001; Triandis, 2002), universalism (Minkov, 2007; Parsons & Shils, 1951; Triandis, 2002), and in-group collectivism by GLOBE (Global Leadership and Organizational Behavior Effectiveness) project (House, Hanges, Javidan, Dorfman, & Gupta, 2004). These are similar conceptual, as well as empirical, constructs to represent (low vs. high) Group. The three empirical constructs by Hofstede, Minkov, and GLOBE are highly correlated and virtually measure the same cultural attributes of the Group dimension.

Grid is clustered with known cultural dimensions like power distance (Hofstede, 2001; House et al., 2004), vertical (Triandis, 2002) and hierarchy (Schwartz, 1999). These dimensions represent role asymmetry and the extent to which hierarchy is expected and accepted in different societies, the attributes that the Grid dimension aims to present as well. Although the empirical dimensions measuring power distance/hierarchy would not be completely converged, they can be considered good estimators for operationalization of the Grid dimension in cross-national level.

For any cultural dimension to be measured, it should be clearly identified what cultural traits and attributes the dimension represents. However, Grid and Group tend to be defined in such broad and diverse terms that many different cultural values could be connected to one and the same dimension. The most general description is usually not the problem. Difficulties arise when Grid and Group are to be defined in ways that make them measurable. Gross and Rayner (1985) are among the first, and few, to provide guidelines for measuring Grid and Group. As illustrated in Table 1, they introduced different measures for each dimension of Grid and Group.

OPEN IN VIEWER

Table 1.

Measures and Definitions of Group and Grid Suggested by Gross and Rayner (1985).

Measures	Definitions
Group
Proximity	Measure of closeness of group members, frequency of interactions
Transitivity	Likelihood that if Member 1 interacts with Member 2 and 2 interacts with 3, then 1 will interact with 3
Frequency	Proportion of time a group member spends in some activity with other members
Scope	Diversity of a member’s interactive involvement in group activities
Impermeability	Likelihood that a nonmember who satisfies membership requirements will actually gain membership
Grid
Specialization	Amount of possible roles a group member assumes in a given time span
Asymmetry	Measure of lack of symmetry in role exchanges among group members
Entitlement	Proportion of ascribed versus achieved roles in the group
Accountability	Amount of member interactions in which one is dominant and the other subordinate

Source. Caulkins and Peters (2002).

With regard to the measures and definitions of Group and Grid in Table 1, we identify some ambiguities that complicate the operationalization of the two dimensions. The measures for Group lack some essential features of interrelatedness that are normally incorporated in individualism versus collectivism as specified and validated in cross-cultural research (Hofstede, 2001; House et al., 2004). For instance, one of the core values for collectivism is loyalty and commitment to the group. It is not clear whether proximity measures can include loyalty or not. Another feature of collectivist culture is in-group sharing, which in turn connects to another collectivist attribute of reciprocity. Living with parents until (even late) marriage and, reciprocally, living of parents with (or close to) their children are examples of collectivist sharing. Again, it is not clear whether or not these collectivist features can be assigned to frequency and proximity (as specified in Table 1).

Most of the Group measures specified in Table 1 can be applied to people who join groups and clubs for volunteering. Contrary to common belief, joining groups and clubs is not a reliable indicator of collectivism. Voluntary activities are more popular among people in individualistic cultures than collectivistic ones because individualists see volunteering as a way of socializing and avoiding being alone (we will elaborate on this later). Thus, the Group measures in Table 1 are not particularly defined to describe the cultural orientation of “grouping with insiders” or “in-group collectivism,” but they can also be interpreted for cultural trait of “grouping with outsiders” or “out-group collaborativeness.” Therefore, two distinct interpretations of Group are recognizable. In this article, we call the former “in-Group” and the latter “out-Group” and we argue that these are two distinctive dimensions of culture.

Grid’s measures by Gross and Rayner (1985) have less ambiguity and more consistency with the cultural attributes of similar dimensions in cross-cultural research: power distance and hierarchy (Hofstede, 2001; House et al., 2004; Schwartz, 1999). However, as we will show further on, some scholars place other cultural orientations like traditionalism/religiosity, gender egalitarianism, or indulgence on the Grid dimension, which increases complexity again.

Given these considerations, we critically examine the measures and survey items used in the literature for the operationalization of Grid and Group in cross-national research. We will demonstrate that Grid and especially Group are operationalized based on very different interpretations of either Grid or Group, and in some cases, these dimensions are overloaded with cultural traits irrelevant to their basic definitions. Connecting to other cross-cultural studies that empirically measure cultural dimensions, we integrate three cultural dimensions of the GLOBE project in our study. These dimensions are pertinent to Group and Grid and their definitions; question items (provided in the appendix) are comparable with the measures in Table 1.

Challenges of Cross-National Operationalization of Grid and Group

Most of the empirical measurements of CT have been made within a nation (Marris et al., 1998; Shin, Chey, & Kim, 1989) and particularly in the United States (Gastil, Braman, Kahan, & Slovic, 2011; Kahan et al., 2007). In recent years, few studies aimed to operationalize Grid and Group dimensions at the cross-national level. Most of them have used the World Values Survey (hereafter WVS) database and its questionnaire (Chai, Liu, & Kim, 2009; Grendstad, 1999; Melton, 2003; Torsello, 2013) and some have used other cross-cultural databases (e.g., Caulkins [1999] uses the Human Relations Area Files’ Probability Sample Files).

We argue that inconsistency and arbitrariness are seen in selecting WVS items for operationalization of Grid and Group in a sense that different studies measure different interpretations of these dimensions. In some cases, items are even irrelevant to the conceptual definitions of Grid and Group. Chai et al. (2009) used 22 survey items from WVS as measures for Grid and Group (11 for each). Their selection of items has been questionable and authors have not performed any systematic method to examine the validity of their selection. Although they argued that “use of factor analysis is not readily compatible with deductive theories like Grid–Group theory” (Chai et al., 2009, p. 200), there is no convincing explanation of how their operationalization can be systematic and reliable.We recognized that some of their selective items have been irrelevant to definition of Grid and Group.

Similarly, Melton (2003) used the WVS items to empirically test the Grid–Group theory. He administered a survey to five Groups in an American university, based on both WVS questions and Gross and Rayner’s measures. He compared the results of these two instruments and concluded that WVS indicators could be proper tools for operationalization of Grid–Group theory. Similar to Chai et al. (2009), Melton’s selection of WVS items for measuring Grid and Group is questionable. In our estimation, some of the selected items are not congruent with the definition of Grid and Group.

Grendstad (1999) also put effort into operationalizing Grid and Group for 12 Western European countries. In his article, which is an attempt to extract the political–cultural map of Europe by using Grid–Group theory, Grendstad (1999) used four cultural items from WVS database. He selected two items for each dimension, performed a factor analysis to validate his assumption, and extracted two factors representing Grid and Group dimensions. Although the factor analysis is a systematic approach for examining the relevance of items to cultural dimensions, a few items in his analysis are problematic and endanger the sampling adequacy. Moreover, although the factor analysis confirmed the independence of two pairs of items and extracted two exclusive factors, the conceptual relevance of adopted items for operationalizing Grid and Group dimensions are debatable.

Grendstad has reported scores of Grid and Group for Western European countries in his article and that is why his operationalization has been used by other researchers, particularly in some scholarly publications of comparative politics in recent years (Heijstek-Ziemann, 2014; Lockhart, 2011). In the next section, we critically assess these measurements and compare Grendstad’s operationalization of Group and Grid with GLOBE’s cultural dimensions to illuminate the challenges in interpretations of Group and Grid at the cross-national level.

Operationalization of Group: In-Group Collectivism or Out-Group Collaborativeness?

Grendstad (1999) utilized two items of “membership in voluntary organizations” and “work in voluntary organizations” as measures for Group. Here, again the aforementioned question arises regarding what meaning of Group has been considered in this operationalization. Being a member of a club/organization may not be a proper measure for representing high Group inclination. Indeed, we can theoretically argue that people in individualistic culture, who have less group ties with extended family or friends, would more frequently join voluntary organizations to make themselves busy and socialized. However, people in high Group cultures may have less extra time to spend in voluntary organizations because they have a lot of collective commitments and in-group gatherings. For instance, Spain, Italy, and Greece display stronger collectivistic orientation among Western European countries (we will later present and discuss this in Figure 4) while they have a very low rate of membership in voluntary organization.

A study by Rippl (2002) for examining the measurement instrument of the Grid–Group model indicated that the item measuring “joining clubs of any kind” has a very low loading on the Group dimension among a sample of German students. We will show later that “volunteering” is highly associated with the cultural trait of team working or (out-group) collaborativeness.

Comparing with one of the existing cross-cultural measurements, namely, GLOBE’s dimensions of culture (see the appendix for GLOBE’s question items), we evaluate Grendstad’s (1999) measurement and study what interpretation of Group he has operationalized. His European cross-cultural map, shown in Figure 1, indicates that Northern European countries, namely, Norway, Sweden, Denmark, Iceland, as well as the Netherlands, score high Group and southern countries, namely, Spain and Italy, score low Group. This is completely contrary to the scores by other established cross-cultural studies, in which the Northern European countries score high individualistic (low Group), whereas Spain, Portugal, and Italy are relatively collectivistic (high Group). Considering face validity, the latter is much more true to life. But what is the reason for this discrepancy?

OPEN IN VIEWER

Figure 1.

Grendstad’s European cultural map (1981 & 1990) based on the Group–Grid model (the circles indicate the 1990 position).

As we discussed above, Grendstad used the WVS items of “voluntary involvement” in his study to measure the Group dimension. We indicated that high voluntary activity is much more popular in individualistic cultures than in collectivist societies because in the latter the connections and interactions of people in their in-group are so much that there is less room for spending time with an out-group through voluntary activities. Thus, we expect that the involvement in voluntary organizations would be higher in individualistic (low Group) cultures. In support of this proposition, we found a high, significant correlation of .61 (p< .05) between a Gallup survey item measuring “the percentages of those who volunteered their time to an organization” (Legatum Prosperity Index, 2011) and Hofstede’s individualism for 16 Western European countries.

We discussed earlier that we should distinguish between two different perceptions of in-Group and out-Group. This ambivalence of Group, seen in other cross-cultural literatures as well, originates from the lack of distinction between in-group collectivism and out-group collaborativeness (Maleki & De Jong, 2014). Whereas the former is a cultural construct to represent the interrelatedness of people in their in-group (sometimes also called familism), the latter is related to collaborative activities and team working mostly with out-groups.

Some cross-cultural scholars consider these two attributes to be two related facets of collectivism (Hofstede, 2001), while in practice it is not the case. GLOBE project (House et al., 2004) has made an effort to separate these two types of collectivism via introducing two exclusive cultural dimensions of “in-group collectivism” and “institutional collectivism.” The latter is associated with the spirit of team working while the former represents the cultural feature of interpersonal ties (see the appendix for the question items). An empirical study in China also indicates the difference between in-group collectivism and out-group cooperativeness (Koch & Koch, 2007). Caulkins (1994), in the chapter titled “Norwegians: Cooperative Individualists,” illuminates how the Norwegian culture accommodates the coexistence of cooperation and individualism.

GLOBE’s work is a valuable effort to make a distinction between these two cultural features both conceptually and empirically. Understanding the difference between these two constructs is crucial for a meaningful operationalization of Group dimension in CT. We found that there is a high, significant correlation of .77 (p< .01) between GLOBE’s “institutional collectivism” and the Gallup measure of “volunteering” for 13 Western European countries. Thus, it can be argued that what Grendstad (1999) has measured was not the Group dimension in its meaning of collectivism (in-Group) but his operationalization of Group resembles a facet of the cultural dimension of “collaborativeness” (out-Group). Figure 2 presents high correlation of .82 between Grendstad’s Group dimension (scores for 1990) and GLOBE’s institutional collectivism for 9 European countries in common. This corroborates our assertion about his operationalization.

OPEN IN VIEWER

Figure 2.

The relation between Grendstad’s Group (scores for 1990) and GLOBE’s institutional collectivism.

In general, we can recognize a kind of “cultural bias” in definition as well as operationalization of the Group dimension by scholars socialized in relatively individualistic cultures in the western hemisphere. For instance, there is no item related to interpersonal ties, family, parents, or friends in Kahan’s questionnaire items for measuring Group; instead, all items are related to the relation between government and citizens (Kahan, 2012). This perception of Group is very different from the alternative perception of individualism versus collectivism based on interpersonal ties. An American bias, in which high versus low Group is associated with left versus right wing and socialism versus capitalism, seems to be dominant in Kahan’s measures. Kahan (2012) himself acknowledges that “we devised our cultural worldview measures . . . to understand variance in perceptions of risk within the US public” (p. 737) and he asserts that these measures might perform poorly in other contexts. Similarly, none of two selected items by Grendstad (1999), a Norwegian scholar, are relevant to interpersonal ties but both are related to joining to voluntary organizations.

Assessing the WVS items selected for measuring Group by three Eastern Asian scholars (Chai et al., 2009), we learned that they selected three items directly related to the importance of family, parents, and friends beside other items related to the private–public dichotomy. Influenced by relatively collectivistic cultures (Korea and China), they are quite likely to be more sensitive to these features of the Group dimension. In another attempt to measure Group using WVS items, an Italian scholar selected one item, among five, associated with the loyalty to friends (Torsello, 2013). The notion that Italian culture is more collectivist than American and Nordic culture, and less collectivist than East Asian culture, might again account for the difference in operationalizing Group. Thus, there are various indications that the operationalization of cultural bias is affected by cultural bias itself, ironically congruent with Douglas’s theory.

There are more challenges of operationalization in addition to the ones already illustrated by Grendstad. Among the selected items for the Group dimension by Chai et al. (2009), the variable, “trust people,” seems not so relevant to the Group dimension. The authors seem to assume that a higher level of trust is associated with high Group. However, other cross-cultural scholars (Hofstede, 2001; Minkov & Hofstede, 2014) have demonstrated that the issue of trust is not related to the Group concept but is conceptually relevant to another cultural construct called “uncertainty avoidance.” Besides, in the ranking of interpersonal trust index, using the same item from WVS, many countries with high individualist culture (low Group) enjoy high interpersonal trust (Medrano, 2008). Our calculation shows that interpersonal trust index has a significant positive correlation (.49, N = 65) with Hofstede’s individualism. This is in complete opposition to Chai and associates’ assumption that “trust” has a positive interrelatedness with high Group.

Among other WVS items used by Chai et al. (2009) for operationalizing Group, some measure the preference for the control of government on society and economy (e.g., “who should run business/take responsibility: government vs. private”). They assumed that the emphasis on the role of government is an indicator for high Group. It implies that welfare states would have a higher Group dimension in comparison with more capitalistic countries. It is again questionable whether this is a pertinent indicator for measuring Group, given the definition by Gross and Rayner (1985).

Furthermore, in Melton’s, (2003) operationalization of Group, one item is about “the preference for new ideas vs. preference for those that have stood the test of time” (p. 140). We argue that the relevance of this item to the Group dimension is dubious. To the best of our knowledge, this item is also better matched to the cultural dimension of uncertainty avoidance (Hofstede, 2001).

Operationalization of Grid: Power Distance or Religiosity?

Regarding the Grid dimension, Grendstad (1999) used two items: “discuss political matters” and “persuading others to share your views” (p. 468). He asserted that Grid “refers to negotiation, autonomy and networking.” It can be argued that these features are not directly relevant to the definition of Grid, which refers to the acceptance of hierarchy and differentiation and asymmetry of roles (Gross & Rayner, 1985). Nevertheless, these two items are not totally irrelevant to Grid definition; however, it is debatable to what extent they really measure Grid. To evaluate this, we utilize GLOBE’s power distance as an operationalization of Grid. The content analysis of the questionnaire used for measuring GLOBE’s power distance shows that all Grid’s measures, indicated in Table 1, are covered by GLOBE’s question items (see the appendix for GLOBE’s question items). Figure 3 shows the relation between GLOBE’s power distance and Grendstad’s Grid dimension (scores for 1990) for nine European countries in common. As can be seen they have a moderate correlation of .57.

OPEN IN VIEWER

Figure 3.

The relation between Grendstad’s Grid (scores for 1990) and GLOBE’s power distance.

Chai et al. (2009) used the items “importance of religion/God,” “justifying of homosexuality/prostitution/abortion/divorce,” and “equal job opportunity for men and women” from the WVS database as measures for the Grid dimension. The last item is about “gender egalitarianism,” which is related to an independent cultural construct (House et al., 2004; Maleki & De Jong, 2014). The first two items are associated with the cultural trait of religiosity/traditionalism, which is also distinct from the Grid dimension. Caulkins (1999), in a pilot study employing factor analysis of variables from 60 cultures, shows that the items related to religiosity (ideology) are independent from the factor representing Grid dimension.

It is widely assumed that there is a plausible association between power distance (or Grid) and religiosity but empirical evidence indicates that no strong correlations exist between them. In some cases, societies with high religiosity have high power distance but the reverse is not necessarily true. For instance, among European countries, while Portugal has a high score of religiosity as well as a high score of power distance, France has a high power distant culture while its religiosity is very low. On the contrary, Ireland has high religiosity but low power distance. Moreover, the United States is an example of having low hierarchy/power distance but virtually high religiosity while China is totally opposite. Thus, these two cultural features cannot be combined in one dimension. Given this, using cultural items related to religiosity as measures for Grid dimension, like what Chai et al. (2009) and Torsello (2013) did, is questionable. Furthermore, in Melton’s (2003) operationalization, one of the WVS items used for measuring Grid is related to “the acceptance of enjoying complete sexual freedom” (p. 141). Again, it can be argued that this item is less relevant to the definition of Grid and more connected to the cultural dimension of religiosity/traditionalism or even more to another distinct dimension called “indulgence,” extracted by Minkov (2007) and adopted by Hofstede as an exclusive dimension of culture (Hofstede, Hofstede, & Minkov, 2010).

All in all, we observe that the existing measurements of both Grid and Group suffer from inconsistency and arbitrariness. This does not mean that the survey-based method of measuring culture should be rejected and disqualified per se. Survey methods and statistical tests are established instruments that allow ongoing improvement. Many dynamic social phenomena are measured and evaluated in this way. The main challenge of using these instruments for measuring culture is fundamental to measuring social phenomena in general: safeguarding the validity and reliability of measures. Regarding validity, Grid–Group CT needs to answer the question of what cultural values and conceptual definitions are attributed to Grid and Group; accordingly, proper measures and survey items need to be selected. Grid–Group theory needs to clarify what the Group dimension represents: in-group collectivism or out-group collaborativeness, and what the Grid dimension represents: power distance or religiosity.

Answering these questions is more important for measuring culture at the cross-national level than at the national level because within a specific society some cultural orientations might appear either to be highly interrelated to other cultural orientations or to be totally intangible and irrelevant within that society, while at the cross-national level these cultural differences do matter.

Measures of Cultural Values or Situational Attitudes?

Normally, it is not expected to observe very sharp changes in the deep cultural patterns of a society over a short period of time. Situation-specific attitudes may change and fluctuate relatively fast but underlying cultural values change more gradually. If Grid and Group are indeed fundamental cultural dimensions, then they should be operationalized using value items rather than attitudinal questions. Grendstad (1999), for instance, tried to measure changes of Grid and Group from 1980 to 1990. Considering a minimum of 0.10 as a remarkable change in his Grid and Group scale suggested that in many Western European countries Grid has had higher variability than Group (see Figure 1). In some countries, like Spain, Ireland, and the Netherlands, strong fluctuations in Group are reported as well.

In our estimation, these variations are related to the items selected for measuring these two dimensions. Items on discussing politics, which have been used for measuring Grid, can be considered situational items that may change from time to time. One may discuss political matters a lot around election time and not before or after that. Asking about “belonging to voluntary organizations” and “currently doing unpaid voluntary work” may also measure situational attitudes and not cultural values on the Group dimension. Melton (2003) also used items asking about political activism, namely, signing petitions, joining boycotts, and attending demonstrations, for measuring Grid. It can be argued that these items are also indicators of situational attitudes rather than cultural values.

The findings from another operationalization of Grid and Group support our line of reasoning. Chai et al. (2009) used items from two waves of WVS to operationalize and compare Grid and Group over time. The items selected by them are more relevant to cultural values, rather than situational attitudes. Accordingly, the scores of Grid and Group for 23 countries in two waves of surveys are consistent and the variation of each dimension is low (see Figures 5 and 6 in Chai et al., 2009). This is in accordance with the expectation of low fluctuation of cultural values over a decade.

We do not assert that cultural values and situational attitudes are fully independent but we argue that they present different kinds of social phenomena; proper indicators for measuring the two need to be different. Indeed, the level of variation of an index over time might show whether it measures cultural values or situational attitudes.

Our assumption, in accordance with most students of culture, that cultural values change gradually is not incompatible with the approach taken by some authors in CT who put great emphasis on the dynamic interplay between the four cultural types. Most prominently, Thompson (2008) argues that the four cultural types are repeatedly splitting and coalescing, increasing and decreasing. In our view, the interplay between cultural biases can be highly changeable indeed, while the underlying value patterns and dimensions—which are played and tossed with—are highly durable. Against this background, we propose to better distinguish between cultural values and situational attitudes while selecting items for measuring Grid and Group.

How Many Dimensions of Culture?

So far, our discussion reveals the challenges of operationalizing multifaceted dimensions such as Grid and Group. In principle, there is no problem if different, adjacent cultural values are attributed to one and the same dimension; “cooperation” and “interpersonal ties” could be assigned to the Group dimension, and “role-asymmetry” and “religiosity” could be linked to the Grid dimension. In making sense of empirical phenomena, however, the overloading of Grid and Group could lead to confusion in understanding cultural diversity—especially across nations, as we demonstrate in this study.

Cultural dimensions are constructed to help in mapping differences among groups of people. When a set of interdependent values is differentiated from others, a new dimension is needed to represent them. Dimensionalization is a kind of classification of cultural values. A classification can be broader or narrower and its number of categories depends on different criteria, some objective and some subjective. The proper number of dimensions should be determined based on the conceptual theory as well as the existing diversity in reality. A parsimonious classification can be useful and appropriate for a specific purpose but it might be nonexplanatory for other purposes. Grid–Group theory made a parsimonious model at the expense of explanatory power of having more, distinctive cultural dimensions.

The number of cultural dimensions is always a matter of controversy. Whereas some cross-cultural scholars believe that having more dimensions is better for explaining the variance of differences among communities (Triandis, 2001), others believe that the number of dimensions should be as few as possible for the sake of parsimony (Douglas, 1999; Hofstede, 2001). Although there is always a trade-off between the principle of parsimony and explanatory power, a middle approach might have the benefit of the both. Working with two big-dimensions of culture is simple and preferable but can be nonexplanatory of subtle differences, especially when we utilize big-dimensions for explaining cross-national differences among countries with a similar cultural background. Big-dimensions can only explain big differences.

New dimensions are required for explaining some variances that cannot be explained with the help of existing dimensions. So a new construct (dimension) must be introduced; otherwise, a latent cultural trait might be embedded into existing dimensions. This overburdens a dimension and can even make it a combination of contradictions.

We already signaled that Chai et al. (2009) define the Grid dimension as an amalgamation of “respect authority” (or acceptance of hierarchy), “gender inequality,” and “religiosity.” If these three measures do not have a conceptual relevance and significant correlations with each other, combining them in one big-dimension can be seriously misleading. When we have a group of people with high inclination to accept hierarchy and at the same time with high gender equality and low religiosity (e.g., France), then how can one combine these different aspects and make one construct to judge about the community’s culture? Thus, if different attributes of culture are not conceptually relevant and statistically correlated to each other, binding them into one dimension is problematic.

Grid and Group theory is not the only parsimonious, two-dimensional cultural theory in the field of cross-cultural studies. Another famous bi-dimensional cultural model is that empirically extracted by Inglehart from the WVS database (Inglehart & Baker, 2000). There is a principal difference between Douglas’s theory and Inglehart’s model; while the former is a theory-based model, the latter is an empirically driven model. Nevertheless, both believe that a major variance of values and attitudes among societies is embedded in and explained by their two dimensions of culture. In a study of major cross-cultural theories, it is asserted that Inglehart’s big-dimension, namely, survival versus self-expression, is the combination of three separate cultural dimensions, namely, individualism, power distance, and indulgence (Maleki & De Jong, 2014). This big-dimension might be useful to explain some variances for a specific purpose but the problem arises when someone needs to compare a particular cultural trait, for instance, power distance or indulgence, among societies. That is why in recent years, Minkov (2007) tried to break down Inglehart’s big-dimension of self-expression and extract new dimensions out of it.

In the next section, we argue that the Grid–Group model suffers from the lack of, at least, a third cultural dimension, without which CT cannot explain some major cross-national differences, and sometimes intra-national differences as well. We assert that this third cultural orientation cannot be incorporated into Grid and Group either conceptually or empirically.

Grade: A Missing Dimension of CT

Group and Grid represent two important cultural orientations that differentiate societies or groups of people. Group, in the meaning of what is called collectivism in other cross-cultural theories, is one of the most elaborated dimensions of culture in the literature. Grid, or hierarchy, or power distance, as labeled by other cross-cultural scholars, is also a well-known dimension in cross-cultural studies.

CT scholars argue that Group and Grid aim to answer two central questions that confront individuals in all societies: “Who am I?” and “What shall I do?” (Wildavsky, 1987, p. 6). That is, “Am I an individual or a member?” and “Shall I behave as a free equal or a regulated ordinal?” But there is another identity question that should be responded to, namely, “How shall I deal?” with the follow-up question suggested by cross-cultural research, “Shall I compete or compromise?” This reveals a crucial cultural dimension, not covered by Grid or Group, needed for a fuller understanding of cross-cultural variation, as we will substantiate. This cultural orientation distinguishes societies/groups in which people are more focused on dominance, achievement, and excellence versus those societies/groups in which harmonious, agreeable, and relaxed relations are key. Trying to find a label alliterating with Grid and Group, we name this third dimension “Grade.”

High Grade culture represents societies in which dominance and mastery are valued, whereas low Grade cultures are more in favor of harmony and compromise. High Grade culture emphasizes grades of achievement and the importance of standing out. Low Grade culture is less focused on exceptional excellence and more interested in acceptable averages. Table 2 presents some features related to high versus low Grade cultures.

OPEN IN VIEWER

Table 2.

Some Features Related to High Versus Low Grade.

High Grade: Standing out	Low Grade: Settling for
High mastery	Low mastery
Competition	Compromise
Dominant	Reclining
Restless	Laid-back
Ambitious	Contented
Adventurous	Relaxed
Quantity of achievement	Quality of life
Excellence, maximizing	Average, satisfying

In the CT literature, competition is loaded on either Grid or Group. For instance, Hampton (1982) and Mars (1982) put competition on the Grid dimension, while Dake and Thompson (1999) loaded competition on the Group dimension. Douglas (1970) assigned this cultural value to the features of an “individualist worldview” (low Grid and low Group) which is also called “competitive individualism” in the literature (Douglas, 1999; Wildavsky, 1987). This implicitly assumes that people with high Grid or high Group cultures have less inclination toward competition and self-assertion. However, the empirical evidence does not confirm this proposition. For instance, at the cross-national level, Japan displays a high Grid and a competition-oriented culture. South Korea has a high Group/high Grid culture as well as a highly competitive culture. Thus, we can have “competitive hierarchy” or even “competitive atomism” in practice.

Grade is compatible with the third dimension of “activity,” which was once suggested by James Hampton (1982) after his attempt, together with Mary Douglas, to give the Grid/Group dimensions an operational definition. Grade is also reminiscent of the third dimension, “manipulation” or “Grip,” which was suggested at one point (Thompson, 1982; Thompson et al., 1990), but not further developed as such, to catch positive and negative attitudes to dominance and the exertion of power. Thompson introduced Grip to justify the existence of a fifth way of life (autonomy or hermit), which was not originally in Douglas’s typology. Grip is not an orthogonal axis similar to Grid and Group, but is “concomitant of the other two” (Mamadouh, 1999, p. 399). Having an extra, independent dimension will double the number of cultural types, or ways of life, from four to eight. This invalidates the “impossibility theorem,” which asserts that “five and only five ways of life” are viable (Thompson et al., 1990, p. 3).

In his PhD thesis, Pepperday (2009) argues for the necessity of employing a third dimension in CT. His proposed dimension, called competition, is compatible with our Grade dimension. He provides a valuable review of other social interaction theories that have implicitly or explicitly proposed a third dimension (Pepperday, 2009). He indicates that “geometrically, three dichotomized dimensions yield eight types,” but he draws a similar conclusion as Thompson: “four of them are not viable and do not arise” (Pepperday, 2009, p. iii). This means that his third dimension is also a concomitant and not an independent dimension.

The advantage of Grade over theoretically compatible items such as “activity,” “grip,” and “competition” is that it can be pinpointed and substantiated, both conceptually and empirically. In the dimensional approach of CT, the expansion of cultural types (or ways of life) due to the presence of a third dimension is not in and of itself undesirable. The number of cultural types is not set in stone and can be revised if additional viable types emerge. After all, cultural dimensions and cultural types are man-made constructs that are meant to help us in explaining social relations and behavioral patterns.

In a recent symposium on CT, Verweij, Luan, and Nowacki (2011) presented suggestions for testing CT in future research. They indicate that there is an affinity between CT and the analysis of social interaction systems developed by Robert Bales and associates. They asserted that Bales’s analyses have many features in common with CT. They notify that Bales (2002) measures the conflicting values and manifested behaviors using three bipolar dimensions: friendly versus unfriendly (conforming to high vs. low Group), rejection of authority (resembling low vs. high Grid), and dominant versus passive. The latter seems to be the very missing dimension in CT that we discussed above. This dimension resembles the cultural features of Grade. The lack of this dimension would limit the explanatory power of CT.

This cultural dimension has been introduced and operationalized in other cross-cultural theories as well (see Maleki & De Jong, 2014). Hofstede’s masculinity versus femininity has some features in common with Grade. Mastery orientation, operationalized cross-nationally by Schwartz (1999), measures more specifically this cultural dimension. Schwartz (1999) defines mastery as “a cultural emphasis on getting ahead through active self-assertion (ambition, success, courage, competence)” (p. 28).

Schwartz’s mastery does not have any significant correlation either with Grendstad’s Grid and Group measures or with other empirical dimensions measuring collectivism and power distance. This corroborates our proposition that Grade is an independent dimension whose cultural values cannot be represented by Grid and Group.

As an illustration for cross-national comparison, comparable with Grendstad’s mapping, Figures 4 and 5 present three-dimensional mappings of Western European countries. They illustrate two different interpretations of the Group dimension, in-Group and out-Group. Scores of GLOBE’s in-group collectivism, institutional collectivism, and power distance (House et al., 2004), and Schwartz’s mastery (Licht, Goldschmidt, & Schwartz, 2007) are used as empirical operationalization of in-Group, out-Group, Grid, and Grade respectively. Comparing these two figures, we see how the different interpretations of Group can result in different cultural mappings of countries. Moreover, the figures also indicate the differences in the Grade dimension across Western European countries.

OPEN IN VIEWER

Figure 4.

In-Group/Grid/Grade map of Western European countries (based on the empirical scores of GLOBE and Schwartz).

OPEN IN VIEWER

Figure 5.

Out-Group/Grid/Grade map of Western European countries (based on the empirical scores of GLOBE and Schwartz).

The lack of this dimension can also be felt in some applications of CT in the study of democracy. Hendriks (2010), in his book Vital Democracy, discusses the affinity between two dimensions of democracy, namely, integrative versus aggregative and direct versus indirect democracy, and two dimensions of “political culture” as well as “societal culture.” For political culture, he uses two bipolar dimensions of power distance (vs. power equality) and contest (vs. convergence). For societal culture, he utilizes the Grid and Group dimensions. He argues that the two dimensions of democracy display “elective affinity” with the two dimensions of societal culture and political culture. If elective affinity is narrowed down to conceptual similarity, then the two dimensions of political culture and societal culture should be connected. Grid and power distance should present the same cultural orientation and the same should go for Group and convergence. However, the latter is debatable. The measures distinguishing high and low Group (see Table 1) seem to be not closely related to the distinction between contest and convergence. If the third dimension of Grade would be available, then a better connection to the cultural attributes of contest versus convergence could be made.

Moreover, this dimension, for instance, can significantly contribute to explaining variance in models of democracy in different countries. In a recent study, Maleki and Hendriks (2014) present a significant association between the mastery dimension and the adopted model of democracy (i.e., consensual or majoritarian) in action. They theoretically argue and empirically demonstrate how cross-national variance in mastery (or Grade) predicts the inclination of different countries to different models of democracy.

All in all, we propose that a third independent dimension is required to explain the variance of culture; without Grade, CT scholars would have to overload the Grid and Group dimensions in dubious ways.

Conclusion

Summing up and referring back to the four initial questions, this article suggests some considerations for operationalizing Douglasian CT for cross-national research:

This study gives some reflections on operationalization of CT at the cross-national level. First, we argued why for a systematic operationalization of CT, particularly for cross-national research, one should measure Grid and Group dimensions instead of direct measuring of four ways of life, which are supposed to be idealtypes extracted from the combination of Grid and Group axes.

We debated that, in some works, different definitions and cultural values—some of which are not interrelated, are used in operationalization of Grid and Group. This can be confusing if it is not clarified what exact meanings and measures are assigned to Grid and Group. Having a very broad definition of a dimension makes its operationalization arbitrary and selective. Thus, both validity and reliability of such operationalization would be questionable.

We established that two different interpretations of Group are evident in the literature. These two perceptions of Group, called in-Group and out-Group, are related to two distinct cultural dimensions, namely, [in-group] collectivism and [out-group] collaborativeness. This is a crucial specification, regardless of the level of measurement. We assert that the puzzling emergence of two exclusive facets of Group in operationalizing this dimension (Boyle & Coughlin, 1994; Hampton, 1982) can be resolved by understanding this differentiation.

We also discussed the importance of differentiation between cultural values and situational attitudes in operationalizing dimensions of culture. Although there is no doubt that cultural values do change over time, it is widely emphasized that they would change slowly and gradually, whereas situational attitudes might change fast. In any effort for measuring culture, it is very important to distinguish indicators that measure cultural values and situational attitudes.

We believe that the main message of CT and the approach of involving cultural differences for explaining variances in sociopolitical institutions and outcomes are valuable and promising. However, we should not restrict ourselves only to qualitative applications of the theory, nor to the two-dimensional model introduced by Douglas. As we argued above, the two dimensions of Grid and Group cannot cover some cultural variances between as well as within societies, and hence, we should introduce and add other dimension(s) when required. This is also applicable to other parsimonious models of culture like the two-dimensional model proposed by Inglehart.

It is understandable that a cultural model might dismiss a dimension in a specific context in which there is no societal variance in that excluded dimension. It is also conceivable that a cultural bias could be seen in the definition and interpretation of a cultural dimension (e.g., Group) operationalized by different scholars socialized in different cultural backgrounds. But when the application of a model goes beyond a specific context, specification of each dimension and adoption of new dimension(s) would be unavoidable. We think CT should consider this if it is supposed to be utilized in cross-national studies. That is why we argued that the third dimension for representing the cultural orientation of competition, which we named Grade, can be complementary to the two-dimensional model of CT.

Finally, we suggest that other existing dimensions of national culture, measured and extracted by other cross-cultural scholars who are mostly affiliated with cross-cultural psychology, can be utilized as a kind of operationalization of CT in cross-national level, at least until a systematic and robust operationalization of Grid, Group, and Grade are available. In a recently published article, some Cultural Theorists also argue that they have identified “the remarkable overlap between concepts and theories in cross-cultural psychology and Douglas’ approach” (Verweij et al., 2014, p. 83).

We are aware of challenges of the existing dimensions of national culture. They implicitly ignore the variances of culture within nations because they use the average tendency of individuals for measuring cultural orientation of a nation. In any comparative study, one has to select a unit of analysis, either a small group of people or a larger population. The aim of cross-cultural comparative research is to compare the average tendency or dominant cultural orientation of different groups or populations. We are aware of the variations underlying all averages, which is neglected in this procedure. This is one of the major limitations of any cross-national analysis working with averages. Moreover, these theories consider cultural values as static constructs and say less about the cultural changes over time. Another challenge to these measurements of national culture is that sometimes operationalization of a cultural dimension (e.g., power distance) by different scholars are not converging.Some of these limitations are inescapable for any cross-national study and some should be treated and improved.

There is no doubt that we need to have more systematic and comprehensive operationalization of cultural dimensions. For now, however, we can utilize those existing cultural dimensions that have been extracted based on a systematic analysis and specified measures. It is time to reconcile the different cross-cultural studies and make benefit from the variety of efforts, gathering them under the broader umbrella of cultural theory. We have begun to realize this purpose in the field of comparative politics and obtained interesting findings in democratization studies.