Kevin Cheng  

Culture and Websites: Not Believing in Aaron Marcus’s Dimensions of Culture and Global Web Design

September 24th, 2004 by Kevin Cheng :: see related comic

Internationalization has become a very popular topic around web design. Designers are becoming more aware of the global scale of websites and are taking into account different language character sets, date formats and currencies. The more subtle effects of culture, however, are less evident. In an attempt to study these factors, Aaron Marcus and Emilie W. Gould (pdf) discuss how Hofstede’s cultural dimensions of power-distance (PD), individual vs. collectivism (IC), masculinity vs. femininity (MAS) and uncertainty avoidance (UA) and long term vs. short term orientation (LTO) may apply to global web sites.

As an exercise, I looked at several corporate and consumer websites that might illustrate - or perhaps contradict - the patterns Marcus and Gould described.The first pattern I discovered was a distinct lack of companies who even had widely varying sites for each region. Most companies had standardized around a central site. In the case of Amazon, sites were structurally equivalent whilst products sold may vary. Nike separated their sites by language but they maintained the same aesthetics for each (Figure 1).

Nike's Homepage Launches to Various Sports-based Sites
Figure 1: Nike’s Homepage Launches to Various Sports-based Sites

This pattern could infer several ideas. For consumer sites such as Amazon, consistency in the interaction may be important to the consumer. Also, Amazon’s tabs and overall structure have become a part of their corporate identity and thus, visual and interaction consistency also strengthens their branding. In the case of Nike, their culture was in the sports themselves. and, for example, both led to the same site despite the obvious difference in vocabulary.

Maintenance costs may also be a consideration. Instead of creating and maintaining multiple designs, companies may favour a more centralized approach.

Eventually, I found Kodak’s collection of worldwide sites, each containing its own flavour. The sites I observed were Kodak Malaysia, India, USA, Sweden, Costa Rica and Japan. Marcus and Gould stipulated many web site design patterns based on Hofstede’s ranking. In some cases, Kodak’s sites adhered to these ideas. However, in at least as many or more cases, examples could be found which directly contradicted Marcus and Gould’s claims.

Marcus and Gould suggested that high PD countries such as Malaysia (ranked first) would focus on authority figures and crests and use a deep hierarchical structure. In reviewing the home pages of each site, I noticed Malaysia was indeed, the only site which focussed on the country itself rather than featuring photographs of individuals (Figure 2a). This observation seems more relevant to Malaysia’s relatively high IC ranking than its PD ranking. In terms of its PD ranking, there was little evidence to suggest any correlation between the PD ranking and the site’s design. For example, finding information on the executives or other authority figures through the site was difficult.

Conversely, Kodak India’s homepage (Figure 2b) seemed heavily focussed on individuals with their picture of two children. India is ranked fairly high (10) in the PD index and this can be illustrated in its emphasis on the management of the company; a listing which could not be found on other sites.

(a) Kodak Malaysia's Homepage Features the Country (b) India Features Management
Figure 2: (a) Kodak Malaysia’s Homepage Features the Country (b) India Features Management

Kodak Malaysia is also noteworthy as it is ranked one of the lowest (50) in the UA index. Contrasted with Japan, which has the highest UA ranking (7), the Malaysian site would be expected to be much more complex, containing more options and selections. Figure 2a and Figure 3a illustrate the two have similar levels of complexity, or perhaps even more complexity in the Japanese site.

According to Hofstede’s rankings, Japan also is ranked highly (4) in his LTO index. Marcus and Gould stipulate that such websites will “require more navigational patience to achieve functional and navigational goals”. Although this statement is supplemented by examples, I find it rather vague and interpret it as a description of the hierarchy. That is, a high LTO ranked site would use a deeper hierarchical structure and focus less on instant gratification of information through searching and redundant navigational elements.

An overview of Japan and USA (ranked 17 LTO) Kodak sites show very little disparity between their designs. Both sites (Figure 3) feature a rich number of options to instantly access information on consumer products, professional equipment, film development, press releases, etc. In addition, both sites feature site-wide searching to quickly gain access to information.

(a) Japan has an American Feel, With High Complexity (b) USA Site
Figure 3: (a) Japan has an American Feel, With High Complexity (b) USA Site

One area that illustrates multiple supporting cases is the IC index. As mentioned earlier, Malaysia ranks quite low in the IC index, and their site emphasizes the country’s beauty and history over the individuals who live there. Equally, Costa Rica, the lowest ranked, has a similar layout to most other Kodak sites including USA but features a picture of Costa Rica instead of individuals (Figure 4a). A link from the homepage takes the user to a page describing Costa Rica, Guatemala, Ecuador, etc.; all of which have low IC rankings. Kodak’s American website supports this contrast with a nearly identical layout, but a focus on an individual instead.

(a) Costa Rica Focuses on Their Country (b) Sweden's Site Shows Little Evidence of Gender Differentiation but Neither Do the Other Kodak Sites
Figure 4: (a) Costa Rica Focuses on Their Country (b) Sweden’s Site Shows Little Evidence of Gender Differentiation but Neither Do the Other Kodak Sites

Marcus and Gould also discuss gender differentiation in websites. They suggest that Japan, with the highest ranking in the MAF index, differentiate gender more explicitly than Sweden, ranked lowest in the same index. However, Kodak Sweden’s website (Figure 4b) does not overtly show any major differences from other Kodak websites. In particular, Japan and Sweden’s sites both have similar navigational setups, and both features mixed gender photographs on the homepage. If anything, the camera and printer featured in the Swedish site appeal more to a masculine culture than Japan’s features of multicultural children.

Gauging whether the design of a site is driven by culture is difficult. Many other factors, such as corporate budget, branding, culture separated along other axis (such as sports) or simply poor design can contribute to the decisions of a design. In addition, many cultures and web sites attempt to emulate other cultures. Marcus and Gould’s application of Hofstede’s rankings seem applicable in some cases, perhaps often enough that they seem interesting to consider but not frequently enough to apply as rules or guidelines. Their task was not a trivial one. Creating a set of web site design patterns based on nationally bounded cultures is similar to reading a horoscope; they only apply if you believe in them.

[Author’s Note: This article was originally written in December 2003 and screenshots are taken from that writing. Some sites may now differ.]

14 Responses to “Culture and Websites: Not Believing in Aaron Marcus’s Dimensions of Culture and Global Web Design”
David Heller wrote:

Interesting article. It was a bit tough for me to follow without the appropriate background in understanding the distinctions made here, but I do have a question. Are these ratings supposed to be used to help people outside of those countries design for them, or are they just a measure of that country and could be tested against the existing data of site designed within those countries?

Basically, in the kodak examples are we to assume these sites were designed in Rochester, NY for all those countries, or are we to assume that they were designed at each locale’s ad agency? If the latter what value does any of this have for us as practitioners? Or is it just an exercise in deconstructing cross-cultural advertising and information management techniques with an attempt to quantify the results?

If we are to assume that it is the former, I would be very interested in how you would take the matrix of a country and change an existing design if the issues you mentioned above (about why we don’t do this) go away.

— dave

Kevin Cheng wrote:

I should mention that Aaron Marcus’s article I linked, he does explain Hofstedes dimensions in more depth. However, a quick Google search identifies this site which covers it sufficiently.

The two sentence version is that the dimensions I mention were created by Hofstede when he did a study on 40 countries. The study was done about thirty years ago with some updates later but it was obviously not directed towards web sites.

Marcus and Gould write about how the dimensions could be applied to global design (I suppose in the manner you describe where one design firm is doing multiple countries).

On one hand, in the case you specify of “how do I design for other countries when I live in NYC” I think using Hofstede is a very solid starting point and Marcus/Gould’s application could help.

On the other hand, they present examples of websites with this characteristic as if to say, “see, these are characteristics of websites from this country.” This part was where I thought, “really? I’m not so sure.”

I had enough trouble finding any site that had numerous localizations. Unfortunately, I don’t know where each one is developed. Do I think sites should be designed with the culture in mind? Definitely. Do I think sites from that country display evidence of those characteristics at the moment? Maybe … it’s there if you’re looking for it.

Ron Zeno wrote:

First, while I think there is great value to Hofstede’s theories, I’ve become more and more skeptical at how generalizeable and valid his theories actually are considering how little confirming evidence has appeared since the theories were originally published. Also, I’m very disappointed by Hofstede’s responses to his critics.

Second, I’m highly skeptical of anyone’s assertions that certain interfaces demonstrate certain qualities, in this case qualities that appeal to or result from Hofstede’s cultural dimensions, without some sort of confirmation of the assertions.

Third, while I strongly advise that products be designed and/or customized for different audiences, there’s no evidence that designing around cultural dimensions (if it is possible to do so) is more important than any other attribute.

Fourth, since cultural dimensions apply to both to the methodology used in design as well as what is important for a design, I’m highly skeptical of the claims from those who ignore the harder part, the methodology. Putting it another way, why should I believe an assessment of a design, where the assessment is done from US-centered values, of designs for other cultures?

Daniel Forer wrote:

I have researched the application of Hofstede’s Cultural Dimensions (HCD) to UI Design for the past 3 years and use them in UI evaluations & designs.

I do have a number of comments on what has been said, so make sure you have 10 minutes spare to read what I have to say.

Practitioners need to take Zeno’s fourth point seriously: if you don’t know what you’re doing when it comes to culture in terms of design & methodology then “don’t try this at home”. I have encountered numerous published articles from high profile Usability experts attempting to prescribe cultural approaches (including HCD) in UI design, who have made some of the most elementary mistakes. Methodology is one key ingredient (& not the only one) to design culturally sensitive UI accurately.

Zeno’s first point however is not accurate in my opinion(no offence intended). Hofstede conducted his study twice over, with a high degree of correlation between data from both studies. Additionally, his study has been duplicated with different samples by independent researchers whose data also indicated a high degree of correlation to Hofstede’s findings (the slight differences that exist are due to differences in methodology). Further, HCD’s also match a number of key national indicators like population size, type of government, GNP, etc, again verifying the construct validity of HCD. I agree a construct like culture has its strength in definition, but from what is available I prefer to use Hofstede.

Zeno’s second point is the concernment with the assertion that UI’s display certain qualities. This assertion is definitely true. Consider operating systems for a moment: a prompt (C:\) based UI is high uncertainty, while a GUI is low uncertainty. Now consider computer games: a strategy game requires analytical & tactical skills to succeed, whereas action games require fast reflex and coordination (different qualities). In terms of Hofstede, strategy games are generally longer in duration and require greater persistence than action games: hence, strategy games may be long term time oriented while actions games are short term time oriented. These differences influence people’s perception, recognition and decision-making differently. If you design a UI in a way to contradict the expectations of users in terms of these qualities and others, chances are they won’t like or use your product.

Zeno’s third point has merit. Designing for HCD is difficult, but I would say that it is generally better than ‘other attributes’. ‘Other attributes’ tend to focus on localisation rather than culturalisation of UIs. ‘Other attributes’ also focus on explicit cultural artefacts that are a manifestation of inherent / learned human values (the true make-up of human culture). Hence, while ‘other attributes’ suggested a character set and language that should be used, do they suggested how or what UI components should be incorporated to support the way of thinking of a group of users?

In comment to Cheng’s article (who I must commend on generating a good debate), Marcus & Gould (M&G) were attempting to find examples, where possible, of websites that displayed similar dimensions to the country where the website originated. I can state though, they looked at the Costa Rica tourist site, which I believe more tourists would be interested in than Costa Ricans. Cheng’s scepticism of “websites from a certain country having the same cultural dimensions of that country” is something I agree with. I wouldn’t design all websites in the same country in the same way on the basis of a dominant national cultural profile? You need to use different parameters for cultural design. Some of the websites that Cheng considered in the original articles probably aren’t suitable for evaluating M&G’s work. Kodak and Nike are multinationals that adopt a standardised approach in terms of products and marketing so won’t fairly reflect cultural design. Picking these sites is understandable though, especially when there are virtually no sites on the web that were purposefully designed to be culturally sensitive.

In reference to M&G article, it provides a starting point to consider deeper and more meaningful attributes for UCD across diverse cultures rather than just general localisation according to language and character sets. The assessment criteria/ guidelines in M&G’s articles are suggestions. They have not been tested and aren’t claimed to have been tested – so should not be criticised on any other basis. Practitioners need to be discerning when reading new approaches to UI design, and should explore the area further before implementing or discarded an approach.

To Zeno & Cheng, I appreciate your interest on this topic. It is with this in mind I would like to comment that too few practitioners are aware of cultural approaches to UI design. The fact that you consider new ideas illustrates a proactive approach to UI design on your part.


Ron Zeno wrote:

Referring to my first point, the validity of Hofstede’s “cultural dimensions” and Forer’s comments on it: Could you provide references (email them to me if you prefer)? As for what’s led me to my conclusions, the following provide an introduction:
+ McSweeney’s criticism of Hofsted (abridged)
+ Criticism of Sweeney, response to Sweeney by Hofstede, Sweeney’s rebuttal
+ Fang’s critique of Hofstede’s fifth cultural dimension

Note that I do think the cultural dimensions are valuable.

Also, Daniel, could you email me your SACLA’03 presentation, “User Performance & User Interface Design: Usability Heuristics vs Cultural Dimensions”?

Anonymous wrote:

Becoming emotionally embroiled in the arguments between Hofstede and McSweeney does not serve practitioners in their interest to pursue culturally sensitive UI designs.

I see points I agree with and disagree with in the articles posted by Zeno. Having read Hofstede’s (2001) book and applied his methods, I have made up my mind on his theory’s worth.

Having read McSweeney’s article after reading Hofstede’s book, I can appreciate McSweeney’s arguments but can also see the flaws in them. His arguments do assist, however, in understanding some of the intricacies of implementing Hofstede’s cultural dimensions.

Additionally, I see McSweeney’s article and Hofstede’s response as another instance in the validation ( & social acceptance) of the cultural dimensions.

Articles that validate the cultural dimensions are listed by Hofstede, in his articles and books. While people may think that Hofstede would only provide supporting literature, it must be remembered that this cited literature is independent (Hofstede didn’t tell them to write it). I suggest that his book is addressed for these references.

To use Hofstede’s cultural dimensions effectively in UI Design requires research and work. The fruits of my effort have been reviewed and received enthusiastically by a number of usability experts whose opinions I respect.

I am pleased that Zeno would like to read my work and am honoured he searched the Web for info on me (of which I have returned the honour), but I don’t believe (rightly or wrongly)that I will receive objective criticism. But your interest is noted.

I will leave this debate to those interested parties to follow up these publications, and especially Hofstede’s book (2001). You can’t claim to know Hofstede’s theory by reading articles based on Hofstede’s work.

Ron Zeno wrote:

I guess the Anonymous post was from Daniel. Sorry that my offering information and requesting an exchange caused such a response. Would you consider sending me your presentation under the terms that I withhold all public comment on them?

Just to clarify, I’m quite familiar with Hofstede’s research: I studied it in depth some 8 years ago, have reviewed related researh for SIGCHI, and still keep up on the research related to applying it to hci.

Summarizing my viewpoint yet again: I find cultural dimensions valuable, but I question their validity.

On to my second point: be skeptical of an individual’s (and to a lesser extent group’s) unqualified assessments of what interfaces appeal to or result from cultural dimensions. Is that more clear? I’m not questioning that interfaces can have these qualities, only to be skeptical of how they are identified.

Kevin Cheng wrote:

This thread has long since gotten far out of my own depth. I only have comparatively, a tertiary knowledge of Hofstede’s Cultural Dimensions and am quite happy to learn more from both sides of the story through this discussion.

Some comments:

Daniel (presumably) said:

but I don’t believe (rightly or wrongly)that I will receive objective criticism. But your interest is noted.

Obviously, Ron Zeno has his views and biases and has stated them quite clearly. The fact that he is willing to consider other perspectives and read about them should be sufficient. No criticism is objective in any case. I’d just like to add that I am also interested in reading this paper if you would like to e-mail it.

Daniel also said:

Having read McSweeney’s article after reading Hofstede’s book, I can appreciate McSweeney’s arguments but can also see the flaws in them.

I’m curious about this. I can appreciate that McSweeney’s article validates the acceptance of cultural dimensions but his points against Hofstede’s methodologies seem quite valid to me.

Ron said:

Summarizing my viewpoint yet again: I find cultural dimensions valuable, but I question their validity.

Just to clarify, are you questioning the validity of Hofstede’s in particular then? i.e., cultural dimensions exist and can be identified but thus far, nobody has done a good enough and reliable enough job of the information gathering and analysis?

Ron Zeno wrote:

On validity: I think Hofstede has identified something real and valuable, I’m just not convinced that enough research has been done to know what has been identified. What’s not well-understood is who these cultural values apply to and how they are acquired.

My third point, that designing for cultural dimensions may or may not be more valuable that designing for other attributes. Basically, I’m agreeing with many of Kevin’s comments, expecially his concluding paragraph.

My fourth point is that these cultural dimensions apply to how we design. The design methodologies that work well in the USA are unlikely to work as well in other countries that have very different cultural dimensions, unless the differences are understood and accounted for.

Daniel wrote:

My statement about the flaws in McSweeney’s article is misinterpretted. He is highlighting flaws in what Hofstede wrote in his book, but not flaws in his method. I agree, Hofstede’s indices for his dimensions are not reflective of the countries they represent - but reflective of the sample Hofstede used. But does that mean his method is wrong?

The dimensions Hofstede identified are believed by some to apply to only certain cultures and are wrong/incomplete for others. This is most possibly true, but will any culture of people truely understand another? I don’t think so. So I return to some of my first comments: Hofstede encourages more research to identify dimensions for other groups (if they exist) and Marcus&Gould built on the starting point offered by Hofstede in understanding culture to approach culturally sensitive UI design.

So are Hofstede’s dimensions invalid? Maybe, maybe not - that is for each person to decide based on their culture & research. But then I also ask - are generic usability engineering approaches (i.e. Nielsen’s usability heuristics and like) appropriate for users outside the USA? No, because they are culturally subjective and lacked representation for other cultures when they were developed. A scary trend if you take it further. We could start questioning all the people we once considered to provide good usability offerings. Maybe we should learn to “Satisfice”, and accept all work as a starting point - chances are, that in 5 to 10 years at the most they will be bettered. So I will use Hofstede and learn more in doing so.

Daniel wrote:

Oh yes. I failed to mention, I am not from the US and haven’t lived there for any extended period. But I have lived in numerous countries for a moderate period of time and have experienced rich cultural influences. So maybe I can appreciate cultural diversity & similarity more? I won’t suggest for one instance that this validates Hofstede but it validates my interest in his work. After all, my country of birth has 11 official languages.

Daniel wrote:

Oh yes. To see fruits of my research & my colleagues, they should be presented in Las Vegas at an HCI conference in July. I never take paper acceptance for granted, so I say should be there.

Kevin Cheng wrote:

Both Daniel and Ron have talked about the need for methodoligical considerations and modifications when considering other cultures. I watched a presentation by Apala Chavan who spoke of the Bollywood methodology in a talk for the UK UPA a few months ago and I think it is an example of addressing such concerns perfectly.

The influence of cultural and aesthetic factors on trust in relation to local multilingual Internet wrote:

The influence of cultural and aesthetic factors on trust in relation to local Internet Shopping site - report

Leave a Reply

OK/Cancel is a comic strip collaboration co-written and co-illustrated by Kevin Cheng and Tom Chi. Our subject matter focuses on interfaces, good and bad and the people behind the industry of building interfaces - usability specialists, interaction designers, human-computer interaction (HCI) experts, industrial designers, etc. (Who Links Here) ?