Measuring Values and Culture

While working on my last project, I learned about how scientists try to measure culture. Culture, of course, is an incredibly complicated subject, but with a few huge simplifications, we can measure aspects of culture and see how those aspects relate to each other and to people’s behaviors. But how do we make these simplifications? I will describe two different ways below:

National Culture (Hofstede’s cultural dimensions)

One huge simplification we can make to culture is to treat each country as having a single national culture, and make comparisons between countries.

Geert Hofstede made this simplification when he measured culture at IBM starting in the 1960s. He did surveys of IBM employees around the world to find out how people’s attitudes varied between countries, even when those people worked for the same company. He found the following cultural dimensions:

• Power Distance Index (PDI). How hierarchical a society is. For example: can you tell your teacher they are wrong? Can you say a casual “hello” to your boss or their boss?
• Individualism vs. Collectivism (IDV). For example: Do you think of yourself primarily as an individual or as part of a group? Is your retirement security due to your personal savings account or your good standing with your in-group?
• Competitiveness vs. cooperation (which Hofstede unfortunately calls “masculinity vs. femininity (MAS)” due to the gendered nature of this in many cultures). For example: Do you value assertiveness and winning, or cooperation and taking care of all?
• Uncertainty Avoidance Index (UAI). For example: Do you make sure there is a rule for every situation, or do you allow contradictory rules and ambiguities.
• Long-term vs. short term normative orientation (LTO). I have found all the explanations of this one confusing and I don’t understand it.
• Indulgence vs. restraint (IVR). For example: How free are you to choose to have fun and take breaks whenever you want, vs. it is not socially appropriate to indulge yourself.

Hofstede’s cultural dimensions are good at providing broad generalizations of culture that are relatively easy to understand. Additionally, it is relatively easy on the internet to guess what country a visitor to a website is from based on their internet IP address. This means it is fairly easy to associate some approximate cultural values from website visitors without asking the visitors any additional questions.

Hofstede’s cultural dimensions do run into problems. One problem is simply that turning culture into six numerical values is a huge oversimplification. Another problem is that country borders are not very accurate dividing lines between cultures. Many groups within countries or even cities may have different cultures and values, and many people move between countries.

See also: Hofstede’s country comparison tool

Values Trade-offs Made by Individuals (Schwartz’s Theory of Basic Human Values)

In the 1990s, Shalom H. Schwartz decided to look at how individuals make values trade-offs, and how they do that in different cultures. What he found was a (relatively) universal set of values trade-offs that produce the following structure of 10 values split into four groups:

The universal values structure. Values on near each other are generally held together, while values on opposite sides of the circle are opposed to each other.

As you can see above, the values group Conservation consists of the values tradition, conformity, and security. These are opposed by the values group Openness to Change which consists of the values self-direction, stimulation, and hedonism. Similarly the value group Self-Transcendence (consisting of universalism and benevolence), is opposed to the value group Self-Enhancement (consisting of hedonism, achievement, and power).

One large benefit to Schwartz’s values as compared with Hofstede’s cultural dimensions, is that people can have their values measured individually. This can be done by giving large surveys of dozens of values questions, or by using what we used: the Short Schwartz’ Values Survey, which just asks directly about the 10 values. We did this as part of our previous study.

With our world-wide data sample, we can see how people made values trade-offs and we find that we get a similar relationship to the theory:

Schwartz’ theoretical relationship between values on the left. Our observed relationship in the right, graphed using a force-directed graph on the correlation matrix (darker lines indicates values that were more correlated in user answers).

We can also look at how these values relate to other things about our participants. For example, people’s self-transcendence increased with age in our data set:

Scatter plot showing trend-line with confidence bands (we removed people who said their age was 99).

Schwartz’s values still sufferer from a similar problem to Hofstede in reducing values and culture to a few small dimensions. Additionally, to find people’s values, they have to answer a number of questions, which many people won’t be interested in doing.

See also: The diagram on this page showing the sub-values that make up the basic 10 Schwartz values.

Relationship between Hofstede and Schwartz

In our previous study, we were able to compare people’s individual values to the Hofstede cultural dimensions. We find, for example, that people in more hierarchical societies (high PDI), also value conservation (tradition, conformity and security), which I think makes intuitive sense:

Scatter plot showing average conservation scores for (participant-reported) countries with error bars (for countries with at least 5 participants). Also has trend line with confidence band.

The variation in Schwartz values was very high within each country, but in aggregate, the trend was apparent. I think this is a nice validation that Hofstede’s dimensions can work as at least an approximate measure of individuals’ values.

Conclusion

By making some very large simplifications, such as reducing culture to a set of numerical values, we can measure aspects of people’s cultures and see how those aspects of culture relate to each other, and how they relate to people’s behaviors. The easiest of these to measure is Hofstede’s cultural dimensions, which makes the dubious assumption that countries have uniform cultures, though this still works somewhat. Schwartz values are more nuanced and can be measured individually, but they require people to fill out survey questions, which they might not be willing to do.

Designers of websites or other computer programs can use these measures of culture when designing. They can think about more design possibilities by thinking about what people in different cultures might want, and they can make their programs and websites adapt to users’ cultures, either more specifically by making users take a values survey, or more coarsely by just looking at what countries their computers are in.

Previous Post: How Culture Influences Debugging

Next Post: There Is No “Blue” in Korean: Different Languages Have Different Colors