Cultural Dimensions: Who Is Stereotyping Whom?

Cultural Dimensions: Who is Stereotyping Whom? Gabrielle Ford1, Paula Kotzè2 and Aaron Marcus3 1

University of KwaZulu-Natal, King George V Avenue, Durban, South Africa, Email: [email protected] 2 University of South Africa, P O Box 392, Pretoria, South Africa, Email: [email protected] 3 Aaron Marcus and Associates, Inc., 1196 Euclid Avenue, Berkeley, USA, Email: [email protected]

Abstract Research to establish what aspects of subjective culture influence usability, and to what extent, requires the use of a valid and relevant cultural model. The use of cultural dimension models in general, and Hofstede’s model in particular, has been widely criticized as being stereotypical and rigid, resulting in the rejection of this model as the basis for such research. In contrast to these criticisms, analysis and synthesis of the extant literature has revealed sufficient theoretical evidence to suggest that Hofstede’s cultural dimensions impact a system’s usability in terms of the cognitive load, user acceptance, objective usability, and the context of use. In addition, we present evidence that the problem of stereotyping is not caused from the use of the model, but rather as a result of the way in which the model is used in experimental design. Consequently, we conclude Hofstede’s cultural dimensions model is a relevant paradigm for further research into the effects of subjective culture on cross-cultural usability.

1

Introduction

The literature contains as many arguments against as supporting the accommodation of subjective culture into user interface design. Consequently a need exists for further research to establish the aspects of subjective culture that influence usability, and to what extent. Such research requires the use of a valid and relevant cultural model. The use of cultural dimension models in general, and Hofstede’s model in particular, has been widely criticized as being stereotypical and rigid [Fitzgerald, 2004; Bourges-Waldegg & Scrivener, 1998; Jagne et al., 2004; Light, 2003], resulting in the rejection of this model as the basis for this further research. A literature review was undertaken to identify and analyze the results of studies previously undertaken that were based on Hofstede’s cultural dimensions model [Ford, 2005]. Only a handful of studies [Straub et al., 1997; Smith and Chang, 2003; Anandarejen et al., 2002; Forer and Ford, 2003; Hall et al., 2003; Massey et al., 2001] were found. In contrast to the criticisms, these studies suggest that Hofstede’s cultural dimensions will influence the cognitive load on the users, and are related to usability from three different perspectives: user acceptance, context of use, and objective usability. Consequently, this paper argues that there is sufficient theoretical evidence to suggest that Hofstede’s cultural dimensions do influence usability, and is therefore a relevant paradigm for further research into the effects of subjective culture on cross-cultural usability. Section 2 reviews the concept of culture, approaches to culturalization, and criticisms against the use of cultural dimension models. Section 3 presents evidence that supports the use of Hofstede’s model [2001] as a basis for researching the impact and extent of subjective culture on usability. Section 4 examines the way in which Hofstede's cultural dimensions have been used in experimental design and concludes that it is the implementation of the model, rather than the model itself, that is causing the problem of stereotyping.

2

Culture

Culture is a complex concept that cannot merely be described: culture must be interpreted [van Peursson, in Evers and Day, 1997]. This complexity implies one must consider dimensions of culture. Metamodels of culture provide a high-level view of the overriding philosophies surrounding the concept of culture, by defining different layers of culture [Hoft, 1996; Masaaki, 2002]. Cultural models provide a detailed view of culture, by identifying a number of cultural dimensions that are used to organize cultural data [Marcus, 2004; Marcus and Baumgartner, 2004b].

2.1

Cultural models and dimensions

Hoft [1996] describes four models of culture, developed by Victor [1992], Hall [1959], Trompenaars [1993] and Hofstede [2001], which are summarized in Table 1. A best-of-breed set of dimensions derived from these and other models is presented in Marcus and Baumgartner [2004b] Table 1: Cultural models and their dimensions Victor [1992]

Hall [1959]

• • • • • • •

• • • • • •

Language Environment and Technology Social Organization Contexting Authority Conception Nonverbal Behavior Temporal Conception

Speed of Messages Context Space Time Information Flow Action Chains

Trompenaars [1993]

Hofstede [2001]

• • • • • • •

• • • • •

Universalism vs. Particularism Neutral or emotional Individualism vs. Collectivism Specific vs. Diffuse Achievement vs. Ascription Time Environment

Power Distance Masculinity vs. Femininity Individualism vs. Collectivism Uncertainty Avoidance Time Orientation

Hofstede [2001] focuses his model on determining the patterns of thinking, feeling and acting that form a culture’s mental programming. He conceptualized culture as ‘programming of the mind’ in the sense that certain reactions were more likely in certain cultures than in others, based on differences between the basic values of the members of different cultures [Smith et al., 2004]. As reflected in Table 1, Hofstede identified five cultural dimensions that can be used to distinguish among different cultures [ITIM, 2003]. All five of Hofstede’s dimensions relate to subjective culture, and many of these dimensions also appear in the other three models summarized in Table 1. Each of these dimensions is a dichotomy, in that there are two opposing sides to each dimension. A brief explanation of each of these dimensions is presented below [ITIM, 2003; Evers, 2001; Hoft, 1996; Marcus and Gould, 2000]: • Power distance: refers to the extent to which less powerful members expect and accept unequal power distribution within a culture. The two opposing sides to this dimension are high and low power distance. • Uncertainty avoidance: relates to the way in which people cope with uncertainty and risk. The two opposing sides to this dimension are high uncertainty avoidance and low uncertainty avoidance. • Masculinity vs. femininity: refers to gender roles, as opposed to physical characteristics, and is commonly characterized by the levels of assertiveness or tenderness in the user. The two opposing sides to this dimension are masculinity and femininity. • Individualism vs. collectivism: refers to the role of the individual and the group, and is characterized by the level of ties between an individual in a society. The two opposing sides are individualism and collectivism. • Time orientation: refers to people’s concerns with the past, present and future. The two opposing sides to this dimension are short-term orientation and long-term orientation.

2.2

The role of culture in human-computer interaction

We argue that as culture influences the way in which people interact in general, culture will also influence the way in which people will interact with computers. Using interactive systems to perform tasks requires communication between the system and the user. People learn patterns of thinking, acting and communicating from living in a specific social environment, normally typified by national culture [Massey et al., 2001]. As such, culture partially pre-determines a person’s communication preferences and behaviors. Communication style, which reflects how a person sends and interprets messages, represents the overall patterns and values of a culture. As the user interface is the means by which the user and the computer interact [Dix et al., 1998], it stands to reason that the user interface

should facilitate users to use their particular communication styles [Massey et al., 2001]. Consequently, global products and services need user interfaces that accommodate a diversity of communication styles to provide support for the cultural diversity of the users. Diversity in culture is particularly relevant for global web user-interface designs [Jagne et al., 2004; Barber and Badre, 1998; Chau et al., 2002; Del Galdo and Nielsen, 1993; Marcus, 2000; Smith and Chang, 2003; Hall et al., 2003]. Like client-server application user interfaces, the web user interface should facilitate ease of learning and use of the system, ensuring the information is conveyed in a cognitively efficient way [Chau et al., 2002]. From a cross-cultural usability perspective, the primary difference between traditional software and web-based user interfaces is that websites are constantly addressing different cultural audiences simultaneously [Chau et al., 2002]. Within the global information technology environment, cross-cultural usability of websites is concerned about making websites an effective means of communication between a global web site owner and a local user [Smith et al., 2004]. Because users differ across regional, linguistic and country boundaries, their expectations of websites are driven primarily by their local cultural perspectives. Consequently, user reactions become more predictable and understandable when the user’s cultural perspective is taken into account [Barber and Badre, 1998; Massey et al., 2001]. Websites need to display ‘culturability’, that is, designing the user interface to accommodate the cultural preferences and biases to increase the usability of the user interface and of the product/service [Barber and Badre, 1998]. However, there are still many arguments against the influence and consequent importance of culture in HCI. The first argument is based on grounds of cultural convergence [Bryan et al., 1994; Norton, 2001; Light, 2003]. Cultural convergence theories suggest that cultures have a tendency to become similar due to technology and global industrialization influences [Bryan et al., 1994]. “Culture may be embedded in technology, but technologies change culture” [Light, 2003]. Computer users are increasingly exposed to influences from the global economy, resulting in the transfer of knowledge, behaviors and standards between cultures [Norton, 2001]. This suggests that all computer users, irrespective of their country of origin, belong to a “computer/technology sub-culture” that uses a specific language, set of symbols, values, and protocols for behavior” [Norton, 2001]. The existence of such a sub-culture essentially refutes the need for accommodating cultural differences. For example, Mrazek and Baldaccini [1997] point out that the use of a trashcan to signify the deletion of files for European and Asian users is often thought to be a mistake, because the users may not recognize the icon. However, if users have previously been exposed to this icon, they will not only recognize it, but will also expect it. Therefore, exposure to another culture may cause expectations, which if not met, will confuse the users more than attempts to accommodate their cultural diversity. The convergence argument is counteracted by the theory of cultural divergence. This theory points out that conformance to global pressures will not necessarily override specific cultural values and practices [Bryan et al., 2004]. Instead, continuous pressures from other cultures will result in “an outburst of attitudes of defense of national and regional identities, and manifestations of the fear of a mixing of races, religions, customs, and habits” [De Souza and Dejean, 1999]. Consequently, sub-cultures are important, but they are superceded by the culture of the country [Marti and Muller, 2003]. This assertion is supported by Child’s [in Bryan et al., 2004] hypothesis that convergence and divergence should be synthesized: cultural convergence is most likely to be found in work structures, whereas divergence is more apparent in work practices. For example, Bryan et al.’s [2004] study found that the work done by information systems professionals is conceptually similar across cultures (convergence), but that technology does not always eliminate differences in cultural beliefs and practices (divergence). A second argument against the importance of culture in HCI relates to the purpose for which the user interface is being used. Chau et al. [2002] identified four main purposes for web user interfaces: information searching, hobbies, social communication, and e-commerce. According to the Wharton Virtual Test Market study [Chau et al., 2002; Fitzgerald, 2004], the most important factor affecting on-line buying behavior is how much individuals used the web to search for product information. Therefore, Fitzgerald [2004] contends that the factor that most influences attitudes and behavioral responses to websites, is the purpose for which the site is being used, rather than culture. However, Chau et al. [2002] point out that it is the culture of the users that dictate the different purposes for which websites are used. Their study found that US users used websites primarily for searching for information, whereas Hong Kong users viewed the Internet as a social interaction mechanism. Furthermore, the different purposes may lead users to have different impressions of the same sites. For example, website user interfaces for US users that do not provide an efficient mechanism for searching for product information will not be perceived as very useful. In contrast, Hong Kong users would be more accepting of website interfaces that provide a virtual community-like environment. Ultimately then, it does not matter what the purpose of the site is: “disregard cultural differences and your site will be doomed” [Forbes, in Chau et al., 2002]. Consequently we agree that cultural differences do matter [Massey et al., 2001], and that it is necessary to accommodate for cultural diversity in software products and Internet sites to meet the needs of the culturally diverse market. Marcus and Baumgartner [2004a] came to the same conclusion from

examining approximately a dozen global corporate websites and matching the differences found in these websites to the culture differences predicted by Hofstede’s model.

2.3

Approaches to culturalization

Culturalization, or preparing a product/service for use by diverse cultures, requires two steps: internationalization and localization. Internationalization involves identifying the culturally specific elements of the product, and localization involves substituting those culturally specific elements with a local content [Russo and Boor, 1993]. Traditionally, the approaches to culturalization seemed to focus primarily on objective cultural issues rather than subjective culture [Ford, 2005]. The objective cultural approach concludes that, when dealing with human-computer interaction, meaning is the central issue in culture [Bourgess-Waldegg and Scrivener, 1998]. This approach suggests that designers need only cater for cultural diversity by ensuring that the intended meaning of user interface representations, such as symbols, icons, and language, are translated to suit the target cultures, so that they are understood correctly. The culturalization process has concentrated primarily on this approach, which is based on the premise that it is the objective, rather than the subjective, cultural aspects that are important. However, researchers are finding this method inadequate [Jagne et al., 2004]. It has been argued that whilst objective culture is important, it is also necessary for the user interfaces to reflect the values, ethics, and morals of the target users [Russo and Boor, 1993], in order to make the users more comfortable and accepting of the user interfaces. These aspects relate to subjective culture [Dunckley and Smith, 2000], and go beyond the “surface manifestations of culture that have been widely accepted” [Kersten et al., in Smith et al., 2004, p. 89]. Carey [1998] points out that there is a difference between comprehension and acceptance. She suggests that it is not enough just to translate the representations, but that it is equally important that other cultural conventions feel comfortable and recognizable to the user, thereby increasing the usability, usefulness, and appeal of the user interface. Russo and Boor [1993] expand on this suggestion, pointing out that in addition to language, user interfaces need to reflect the values, ethics, and morals of the target users. More specifically, there is an inseparable link between knowledge and culture [Ciborowski, 1979]. For example, what may be believed to be an irrefutable fact in one culture may be deemed to be groundless superstition in another culture. In the context of user-interface design, knowledge is a determinant of user performance. As culture influences knowledge, and knowledge influences performance, this implies that culture also influences usability and performance. Del Galdo and Nielsen [1993] clearly support this by pointing out that there are three levels of internationalization, namely: 1. Displaying the native language, character set, and notations. 2. Translating the user interface and documentation so that it is understandable and usable. 3. Matching the user’s cultural characteristics, which goes beyond avoiding offensive icons and must accommodate the way business is conducted and the way people communicate. Essentially, this approach is based on the premise that culture is about how individuals behave and respond, their beliefs and values, and therefore it is also necessary to reflect subjective culture in the design of user interfaces [Dunckley and Smith, 2000]. Smith et al. [2004] also support this view for the culturalization of web user interfaces: they believe that “the process of internationalization and localization for web-based systems involve an analysis of both objective and subjective cultural issues” [p 83]. This argument highlights that culturalization should be addressed from both sides of the cultural spectrum: subjective cultural issues should be accommodated using a top-down approach, whilst objective cultural issues should be dealt with using a bottom-up approach. We suggest that subjective culture provides the context or conceptual model within which the user interface is designed, thus forming the higher-level, semantic layer of the user interface. The more detailed syntactic level should be governed by objective culture, covering issues such as language, text and date formats, and content [Ford, 2005].

2.4

Criticisms against the use of cultural dimension models

Addressing culturalization from both a subjective and objective cultural perspective requires subjective cultural user-interface design guidelines to accommodate subjective culture into the design of the interfaces. Cultural dimension models are one of four types of models currently being used for managing the subjective aspects of crosscultural website design [Fitzgerald, 2004]. Interest in cultural dimension models arose as a result of a need for models of cross-cultural software development [Fitzgerald, 2004]. However, the use of cultural models has been criticized based on discomfort with

stereotypes [Light, 2003; Bourges-Waldegg and Scrivener; 1998] and the inconclusive results obtained from studies done to evaluate and apply these models to HCI [Fitzgerald, 2004]. A stereotype is defined as a “fixed notion about persons in a certain category, with no distinctions made among individuals” [Hofstede, 2001, p 14]. Thus, the use of cultural models is considered inadequate because they are generalizations and therefore do not take into account the context in which they are applied [Bourges-Waldegg and Scrivener, 1998]. In addition, not everyone fits into the ‘cultural mean’, because there is a wide range of individual differences [Light. 2003; Jagne et al., 2004]. In particular, Hofstede’s cultural model, which is the most often quoted of the four cultural models discussed previously [Smith et al., 2004], has endured much criticism from the HCI and usability-related community. For example, Hall [in Jagne et al., 2004] says that “it would be wrong to take the characteristics of Hofstede and attempt to deduce from these how a particular cultural group would respond to a particular technology.” Hofstede’s model has been considered too stereotypical [Bourges-Waldegg and Scrivener, 1998] or rigid [Nocera and Hall, in Jagne et al., 2004]. In addition, previous attempts to apply Hofstede’s model to usability also resulted in conflicting and therefore inconclusive findings. For example, Gould et al. [2000] found Malaysian websites contain links on the home page to website administration, which correlates well with the high power distance reported [Hofstede, 2001] for Malaysia. However, this finding does not explain why low power distant countries, for example, the US, also contain such links on their websites. Consequently, until better proof of their relevance to website design is provided, Fitzgerald [2004] suggests cultural dimension models should be used with care. In contrast to the arguments put forward against the use of cultural models in general and against the use of Hofstede’s [2001] cultural model in particular, we believe the literature provides sufficient theoretical evidence to support the use of Hofstede’s cultural model as a basis for managing the subjective cultural aspects of user-interface design. Analysis of this model indicates that accommodating these dimensions into the design would decrease the cognitive load, thus enhancing usability and performance. In addition, a number of studies in the literature suggest that these dimensions are related to usability. This theoretical evidence is presented in the section that follows.

3

The Relationship between Hofstede’s Cultural Dimensions and Usability

An analysis of the limited number of studies found in the literature that focus on subjective culture indicates that subjective culture does influence usability. The importance of this relationship is evident in that subjective culture is shown to relate to usability in terms of the cognitive load, user acceptance, context of use, and objective usability..

3.1

Cognitive load perspective

Users interpret new information on the basis of their existing mental models [Honold, 2000]. Hofstede’s cultural model is focused on determining patterns of thinking, feeling and acting that form a culture’s mental model [Hoft, 1996]. This leads us to believe that Hofstede’s cultural dimensions do influence the cognitive resources required for performance to take place. In terms of the attraction and identification processes, we suggest that the metaphors used should be pertinent to the cultural dimension of the user, particularly if the metaphor is used as a way of representing a specific function of the user interface. For example, high power distant users would be more attracted to metaphors that focus on expertise, authority, official stamps and logos, whereas low power distant users would be more attracted to metaphors that focus on images of equality, such as playgrounds and public places [Marcus, 2004]. Similarly, cultural dimensions influence the reasoning, problem solving, and decision-making processes by: • Influencing the user’s perception of concrete data: The provision of concrete data may alleviate the complexity of the problem, thus facilitating problem solving during the analysis phase of the cognitive process. We suggest that the user’s perception of concrete data can be influenced by the user’s cultural dimension. For example, a high uncertainty avoidant user prefers detailed explanations and information, whereas a low uncertainty avoidant user prefers summarized data [Marcus, 2004]. • Maintaining the user’s concentration during reasoning and problem-solving: Reasoning and problemsolving require concentration. Any distractions could cause a loss in concentration, resulting in a need to start again, thus reducing speed, and sometimes accuracy, of performance. Slow response times also cause the user to become distracted from their task through boredom or irritation, causing loss of concentration. This is exacerbated in users that are naturally impatient or wanting to complete tasks quickly, such as masculine and short-term oriented users [Marcus, 2004]. • Reducing anxiety: Anxiety reduces the size of memory, as the user is partially absorbed by concerns that are beyond the problem-solving task [Shneiderman, 1998]. For example, high uncertainty avoidant users are generally more emotional and stressed than low uncertainty avoidant users, and may be so anxious

about having to learn how to navigate through the user interface that they are not concentrating on the problem itself. As another example, high power distant users may be worried about disappointing their superiors, thus putting themselves under more pressure and causing additional anxiety [Marcus, 2004]. Hofstede’s cultural dimensions also could affect the way in which users will respond to messages. For example, collectivist users would be uncomfortable with having to express personal opinions, while high power distant users would not wish to express an opinion that is in direct contradiction to their superiors. Similarly, individualist users would be uncomfortable with having to provide personal information, but would find that being unable to express a personal opinion unacceptable [Marcus, 2004]. This situation, in turn, will affect the functionality provided in the user interface. For example, a study conducted on students’ use of a hypertext product to analyze a poem found that students should be able to add their own viewpoints to the system in some cultures (low power distance and individualism), but in others (high power distance and collectivism) this capability would be unacceptable [Russo and Boor, 1993]. Because the product did not provide a function that allowed students to add their viewpoints, the system was considered unacceptable to some cultures. Therefore, the functionality provided in the user interface should be consistent with the users’ cultural dimensions, thus adhering to the related principle of task conformance, and providing users with an appropriate mechanism with which to respond. This functionality will reduce the possibility of a loss of speed and accuracy, increased training time, and reduction in user satisfaction.

3.2

User acceptance perspective

Many of the studies relating to subjective culture identified in the literature focus on examining the role of subjective culture on user acceptance of information technology, rather than on usability. These studies are based on the Technology Acceptance Model (TAM), which was originally developed by Davis [Davis, 1989] to explain why people accept or reject information technology. The model identifies perceived ease of use and perceived usefulness as the two basic determinants of user acceptance. Acceptance, as defined by the TAM literature, is therefore based on user perceptions of usability, rather than actual performance. The question that arises is whether ‘acceptance’ is related to usability, or whether these two concepts should be treated as two mutually exclusive attributes of a product. The definition of usability as stated in the ISO 9241 standard [1997] indicates that usability is both a belief-based and performance-based concept. Effectiveness and efficiency are undoubtedly performance based concepts, in that they are measured in terms of actual performance. Satisfaction is defined by this standard as the comfort and acceptability of the work system to its users. Bevan and Macleod [1994] explain that “measures of satisfaction describe the perceived usability of the overall system by its users and the acceptability of the system to the people who use it and to other people who are affected by its use. Measures of satisfaction can provide a useful indication of the user’s perception of usability, even if it is not possible to obtain measures of effectiveness and efficiency”. Satisfaction is therefore a belief-based concept that assesses the users’ perceived usability of the system. Therefore, according to the ISO definition [ISO 9241, 1997] of usability, perceived usability is a valid measure of usability. The differentiation between performance-based usability and belief-based usability suggests that designers should differentiate between usability and user experience goals [Preece et al., 2003]. This allows for the identification of two categories of valid, measurable usability parameters, namely objective usability and subjective usability [Bailey, 1996]. Objective usability measures assess how capable users are at using the system, and include measures related to effectiveness and efficiency. Subjective usability assesses how satisfied the users are with the system, and includes measures of acceptance and comfort. The TAM literature shows that if users are satisfied with the perceived usability of the system, they will accept the system, thus indicating that the system is subjectively usable. The above discussion leads us to conclude that usability comprises subjective usability (satisfaction) and objective usability (effectiveness and efficiency). Consequently, a relationship between subjective culture and usability can exist as a relationship between subjective culture and objective usability, or as a relationship between subjective culture and subjective usability, or both [Ford, 2005]. Evidence that Hofstede’s cultural dimensions impact user acceptance, and therefore on subjective usability, was found in four studies reported in the literature. These studies are discussed briefly below. • Smith and Chang’s [2003] study reported that Chinese users preferred sites that displayed high power distant, high uncertainty avoidant, masculine and individualistic characteristics. The researchers expressed surprise at the users’ preference for individualistic user interfaces, because Hofstede’s study found the Chinese to be a predominantly collectivist society. In addition, it was found that the contributions to variance by these dimensions were different. Power distance contributed 22% to the variance in user preference, whilst masculinity/femininity and individualism/collectivism had lower contributions of 9%

•

•

•

3.3

and 2% respectively, and uncertainty avoidance virtually no contribution at 0.01%. The findings of this study suggest that while there is a relationship between culture and usability, only some of the cultural dimensions may have a significant impact on usability. These findings suggest the users’ subjective cultural profiles (as defined in terms of Hofstede’s cultural dimensions) influence the users’ preferences and acceptance of interfaces. The Straub, Keil, and Brenner [1997] study reported that the users’ subjective cultural profile influences their acceptance of e-mail as a communication medium. Specifically, it was found that users with a cultural profile of high uncertainty avoidance, high power distance, collectivism and masculinity (such as the Japanese), would reject email as a communication media as it is not information rich and does not support social presence. In contrast, users with a different cultural profile, such as American and Swiss users, would accept such communication media. This suggests that the users’ subjective cultural profile influences their acceptance of particular technologies to perform particular tasks. The Massey, Hung, Montoya-Weiss and Ramesh [2001] study reported that cultural differences in terms of uncertainty avoidance, context and individualism/collectivism affect communication behaviour and hence perceptions and preferences. The Anandarejen, Igbaria, and Anakwe [2002] study assessed the extent to which Hofstede’s [2001] cultural dimensions influenced user acceptance, specifically in the context of a collectivist society. They hypothesized that: - Individuals from a collectivist culture would use computers not because of their perceived usefulness or the enjoyment to be derived, but rather because of the perceived social pressure from their peers. - Computer playfulness would not influence acceptance for high uncertainty avoidant users, as playfulness involves creativity as well as unstructured experimentation. - Organizational support would influence acceptance for collectivist and feminine users, as these societies value cooperation and harmony. The study reported that cultural tendencies towards high uncertainty avoidance, high power distance, collectivism and masculinity disassociate usefulness from acceptance. In addition, perceived enjoyment was found to have no direct effect on acceptance, and individually based training did not enhance computer skills and therefore had no effect on perceived ease of use. In contrast, it was found that organizational support was positively related to perceived ease of use in a collectivist society. Thus, the findings not only confirm that there is a relationship between subjective culture and usability, but that the variables proposed to influence user acceptance is dependent on the subjective cultural context of the intended users.

The objective usability perspective

Objective usability assesses how capable users are at using the system, and include measures related to the accuracy, completeness and speed with which users complete specific tasks. Empirical evidence that subjective culture influences objective usability was found in the study performed by Forer and Ford [2003]. The study reported that users with a specific cultural profile achieved higher accuracy and faster speed when using a user interface that matched their cultural profile than when using a user interface with an opposing cultural profile. This suggests that for at least one cultural profile, Hofstede’s dimensions influence the objective usability of user interfaces. We now look at the relationship between subjective culture and usability from a use-context perspective.

3.4

The context of use perspective

Context of use is categorized into the contexts of the users, the task, the environment and the equipment [Bevan and Macleod, 1994]. The ISO [1997] definition of usability highlights that information technology is used within a specific context, and it is important that this context be taken into consideration when measuring usability. In terms of the user context, “individual differences between users such as those due to cultural differences and prejudices may influence the use of a product in the real world” [Bevan and Macleod, 1994]. Hall, Lawson, and Minocha [2003] argue that Hofstede’s cultural dimensions provide the context within which different solutions to the same problem can be identified. For example, a common problem faced by designers of web user interfaces is to find a way that will attract users to the site. For individualist cultures, one way to solve this problem would be to focus on the goals of the individual. In contrast, focusing on collective agendas would be a more suitable solution to the same problem for collectivist societies [Marcus, 2000]. This problem-solving context can be extended to the other cultural dimensions identified by Hofstede as well. Thus, these dimensions can provide

a high-level context of use within which the content and features of the interface should be designed. Furthermore, Gould (as reported by Light [2003]) points out that subjective cultural dimensions serve as design opportunities. This suggests that subjective culture influences the task context of use, as users with different cultures will want to perform different tasks appropriate to their respective cultural traits.

4

Who is Stereotyping Whom?

One main criticism against the use of cultural dimension models is that such models tend to stereotype users, as discussed in section 2. Although Hofstede [2001] identified a dominant cultural profile for each country, not every citizen of that country has the same cultural profile. Cultural boundaries do not necessarily coincide with national boundaries [Duncker, 2000]. National states often comprise multiple cultures and ethnicities, e.g., in South Africa. Therefore, it is possible that a selected sample of citizens of Japan, Guatemala and South Africa would share the same cultural profile, in contrast to the dominant cultural profiles identified for their countries of origin. Thus, assuming that a selected sample of users will have the same cultural profile as the dominant cultural profile of their country of residence or birth is falling into the trap of ecological fallacy, which Hofstede warns against. The studies quoted in section 3.2 do not appear to have identified the cultural profiles of the test users. This was confirmed in the case of the Smith and Chang study: personal correspondence with the lead author confirmed that users were chosen based on their nationality (Chinese). Smith and Chang [2003] expressed surprise that their test subjects preferred a web site displaying individualism characteristics, as this preference is in contrast to that expected from a collectivist culture such as the Chinese. However, as the cultural profiles of the test users were not identified, it is possible that the dominant cultural profile of the test subjects was different from the country’s cultural profile, which could explain the surprising results for the individualism/collectivism dimension. Although this factor does not influence the results of the study in terms of the relative impact of the cultural dimensions on user acceptance, it does not provide convincing evidence that matching the cultural profile of the user interface to the cultural profile of the users increases acceptance and usability. In addition, we interpreted the Anandarejen et al. [2002] findings as evidence that there is a relationship between subjective culture and usability, and the influence of some user acceptance variables is dependent on the subjective cultural profile of the users. However, if the cultural profiles of the test subjects were not identified, there is actually little evidence that the reported findings are valid for the cultural dimensions indicated. The Straub et al. [1997] study potentially suffers from the same flaw. The study provided evidence that users who are high uncertainty avoidant, high power distant, collectivist, and masculine would reject the use of email, once again suggesting that there is a relationship between subjective culture and usability. However, as the cultural profiles of the test subjects were not identified, this evidence is questionable. The same holds true for the Massey et al. [2001] study. The lack of assessing cultural profiles of the test subjects may well be a contributing factor to the conflicting evidence that such research has provided in the past, which is yet another criticism leveled against the use of cultural dimension models. The cultural profiles of the test subjects used in these experiments have not been identified and therefore are not shown to be equivalent to the assumed cultural profile that is being tested. This is a prime example of the model being stereotyped, rather than doing the stereotyping. Consequently, we argue it is not the cultural dimension models, but rather it is the way in which the cultural dimension models are implemented in research methodologies that are causing the problem of stereotyping.

5

Conclusions

In contrast to the criticisms against using cultural dimension models, the synthesis of the cited studies provides strong theoretical evidence to support the use of Hofstede’s model, and by inference, other cultural dimension models, as a basis for researching the impact of subjective culture on usability. In addition, we argue that perhaps the way in which researchers use such models causes stereotyping, rather than the developers of the models themselves: omitting to identify cultural profiles of individual test users is a prime example of such stereotyping. Consequently, we believe the assumptions surrounding the validity of cultural dimension models should be reassessed, and such reassessment should be sensitive to the interdisciplinary framework within which humancomputer interaction needs to operate.

References Anandarejen, M., Igbaria, M. and Anakwe, U.P. 2002. IT acceptance in a less-developed country: a motivational factor perspective. International Journal of Information Management 22, 47-65. Bailey, R. 1996. Human Performance Engineering. London: Prentice Hall PTR. Barber, W and Badre, A. 1998. Culturability, the merging of culture and usability. In Proceedings of the 4th conference on Human Factors and the Web. NJ, USA: Basking Ridge. Bevan, N. and Macleod, M. 1994. Usability measurement in context. Behaviour and Information Technology, 13: 132–145. Bourges-Waldegg, P. and Scrivener, S.A.R. 1998. Meaning, the central issue in cross-cultural HCI design. Interacting with Computers, (9)3: 287-309. Bryan, N.B., McLean, E.R., Smith, S.J. and Burn, J. 1994. The structure of work perceptions among Hong Kong and United States IS professionals: a multidimensional scaling test of the Hofstede cultural paradigm. In A Duane and P Finnegan (Eds.), Proceedings of the 1994 computer personnel research conference on Reinventing IS: managing information technology in changing organizations (pp 219-230). New York, NY, USA: ACM Press. Carey, J.M. 1998. Creating Global Software: A Conspectus And Review. Interacting with Computers, 9: 449-465. Chau, P.Y.K., Cole, M., Massey, A.P., Montoya-Weiss, M. and O’Keefe, R.M. 2002. Cultural Differences in the Online Behavior of Consumers. Communications of the ACM, 45(10): 138–143. De Souza, M. and Dejean, P. 1999. Interculturality and design: Is culture a block or an encouragement to innovation. In Proceedings of the International Conference of Design Research: Design Cultures. Sheffield, England: Sheffield Hallam University and The European Academy Of Design. Ciborowski, T.J. 1979. Cross-Cultural aspects of Cognitive Functioning: Culture and Knowledge. In A.J. Marsella, R.G. Tharp, and T.J. Ciborowski (Eds), Perspectives on Cross-Cultural Psychology. New York, NY: Academic Press Inc. Davis, F. D. 1989. Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information technology. MIS Quarterly, 13(3): 319-340. del Galdo, E and Nielson, J. 1996. International User Interfaces. John Wiley and Sons. Dix A, Finlay, J, Abowd, G and Beale, R. 1998. Human-Computer Interaction. Hemel Hampstead: Prentice Hall International (UK). Duncker, E. 2002. Cross-cultural usability of the library metaphor. In Proceedings of the second ACM/IEEE-CS joint conference on Digital libraries (pp 223 – 230). New York, NY, USA: ACM Press. Dunckley, L. and Smith, A. 2000 Cultural Dichotomies in User Evaluation of International Software. In D. Day, E. del Galdo, and G. Prahbu, (Eds.), Proceedings of IWIPS’00, Designing for Global Markets 2. Baltimore, MD: Backhouse Press. Evers, V. and Day, D. 1997. The Role of Culture in Interface Acceptance.In Mende S. Howard, J. Hammond and G. Lindgaard, Proceedings of the Human Computer Interaction INTERACT’97 Conference (pp 260 – 267). Sydney: Chapman and Hall. Fitzgerald, W. 2004. Models for Cross-Cultural Communications for Cross-Cultural Website Design, National Research Council Canada, Institute for Information Technology. Ford, G. 2005. Researching the Effects of Culture on Usability. MSc Dissertation, University of South Africa. Forer, D. and Ford, G. 2003. User performance and user interface design: Usability heuristics versus cultural dimensions, in Proceedings of the South African Computer Lecturer’s Association, June- July 2003. Gould, E.W., Zakaria, N. and Yusof, S.A.M. 2000. Applying culture to website design: a comparison of Malaysian and US websites. In Proc. IEEE prof. comm.. society’s internat. prof. comm. conf. and Proc. 18th annual ACM internat. Conf. on Computer documentation: technology and teamwork (pp 161-171). Piscataway, NJ, USA: IEEE Educational Activities Department. Hall, P., Lawson, C. and Minocha, S. 2003. Design Patterns as a Guide to the Cultural Localisation of Software, in Proceedings of the IWIPS 2003 conference, Berlin. Hofstede, G. 2001. Culture’s consequences. (2 ed) Sage Publications. Hoft, N. 1996. Developing a Cultural Model. In E del Galdo and J. Nielson (Eds.), International User Interfaces. New York: John Wiley and Sons.

Honold, P. 2000. Culture and Context: An Empirical Study for the Development of a Framework for the Elicitation of Cultural Influence in Product Usage. Internat. Journal of Human-Computer Interaction, 12(3&4): 327–345. ISO9241. 1997. ISO 9241-1: Ergonomic requirements for office work with visual display terminals. Berlin: Beuth. ITIM. 2003. Geert Hofstede Cultural Dimensions. Retrieved January 12, 2005 from the World Wide Web: http://www.geert-hofstede.com/geert_hofstede_resources.shtml. Jagne, J., Smith, S.G., Duncker, E. and Curzon, P. 2004. Cross-cultural Interface Design Strategy, Technical Report: IDC-TR-2004-006, Interaction Design Centre, , Middelsex University. Kurosu, Masaaki (2003). “Theory of Layers of Culture.” Proc. HCII, 2003. Light, A. 2003. Culture and Usability: Will Design Patterns ease Problems of Context?, in Usability News. http://www.usabilitynews.com/news/article1185.asp. Marcus, A. 2000. Cultural Dimensions and Global Web User Interface Design: What? So What? What Now? in South African Human-Computer Interaction Conference, May 2000. http://www.chi-sa.org.za/abstracts.htm Marcus, A. and Gould, E.W. 2000. Crosscurrents: cultural dimensions and global Web user-interface design. Interactions, 7(4): 32-46. Marcus, Aaron. 2004. "User-Interface Design and Culture," Chapter 3 in Aykin, Nuray, ed., Usability and Internationalization of Information Technology, Lawrence Erlbaum Publishers, New York, 51-78, 2004. Marcus, Aaron, and Baumgartner,Valentina-Johanna. 2004a. "Mapping User-Interface Design Components vs. Culture Dimensions in Corporate Websites," Visible Language Journal, MIT Press, 1-65, 2004. Marcus,Aaron, and Baumgartner, Valentina-Johanna. 2004b. "A Practical Set of Culture Dimension for Evaluating User-Interface Designs" in Proceedings, Sixth Asia-Pacific Conference on Computer-Human Interaction (APCHI 2004), Royal Lakeside Novotel Hotel, Rotorua, New Zealand, 30 June-2 July 2004, 252-261. Marti, J. and Muller, C. 2003. Cross-Cultural Communication and Comprehensibility on Websites. In V. Evers, K. Röse, P. Honold, J Coronado, D.L. Day, (Eds.), Proceedings of the IWIPS 2003 Conference (pp 57 – 64). Berlin, Germany: University of Kaiserslautern. Massey, A.P., Hung, Y.C., Montoya-Weiss, M. and Ramesh, V. 2001. When culture and style aren't about clothes: perceptions of task-technology ‘fit’ in global virtual teams , in Proc. 2001 International ACM SIGGROUP Conf. on Supporting Group Work Boulder, Colorado, USA, pp 207 - 213 ACM Press New York, NY, USA Mrazek, D and Baldaccini, C. 1997. Avoiding cultural false positives. Interactions, 4(4): 19-24. New York, NY, USA: ACM Press. Norton, D. 2001. Implementation of an Electronic Report Viewing Application for Multi-cultural Users. In Proceedings of the South African Human Computer Interaction Conference. CHI-SA. Preece, J., Rogers, Y. and Sharp. H. 2002. Interaction Design: Beyond Human-Computer Interaction. John Wiley and Sons, Inc. Russo, P. and Boor, S. 1993. How Fluent is your Interface? Designing for International Users. In Proceedings of the INTERCHI ’93 Conference on Human Factors in Computing Systems: INTERACT ’93 and CHI’93 (pp 342347). New York, NY, USA: ACM Press. Shneiderman, B. 1998. Designing the User Interface: Strategies for Effective Human-Computer Interaction. Addison-Wesley. Smith, A. and Chang, Y. 2003. Quantifying Hofstede and Developing Cultural Fingerprints for Website Acceptability, In Proceedings Of The IWIPS 2003 Conference, Berlin, Germany, July 2003, V. Evers, K. Röse, P. Honold, J Coronado, D.L. Day, Eds. University of Kaiserslautern. Smith A., Dunckley, L., French, T., Minocha, S. and Chang, Y. 2004. A Process Model for Developing Usable Cross-Cultural Websites. Interacting with Computers, 16: 63–91. Straub, D., Keil, M. and Brenner, W. 1997. Testing the technology acceptance model across cultures: A three country study, Information and Management, Vol. 33, No. 1, pp. 1-11