Blog

In Africa, the methods that led to the Big Five tell us something very different

Originally posted December 15, 2020 to Psychology Today

• 

• 

• 

A Model of Personality Can Be Built from the Everyday Words We Use to Describe Each Other

Personality questionnaires that ask us about our patterns of behavior or feelings predict important life outcomes: success in work and marriage, health and longevity, and risk of psychological disorders. But before creating a questionnaire, personality psychologists must decide which traits are most important to measure. In the early part of the 20th century, they did this based on their observations and experiences, and everyone used a different model. It was hard to compare results.

Lexical studies of personality introduced a more objective way to decide which traits are important. Words to differentiate between people exist in every language. Lexical studies are based on a rationale from linguistics: people seek efficiency, and where an idea is expressed often, it will become encoded into a single word. Lexical Studies are conducted in four basic steps: (1) extract all terms used to describe psychological differences from a dictionary; (2) reduce the list to a manageable number by determining which are most commonly used; (3) administer the list as a questionnaire, asking participants how well each term describes a target person; (4) use factor analysis to find out which terms naturally group together into broad traits. For example, people who are described as highly kind are often also described as highly patient, but as not at all rude. The results factor might be labelled something like ‘Agreeableness’

In the 1990s, lexical studies in English, German, and Dutch, converged on the same model, and the Big Five (Extraversion, Agreeableness, Conscientiousness, Emotional Stability, Openness/Intellect) was born. Several questionnaires to measure it were developed and have since been used in thousands of studies.

Is the Big Five Universal?

It’s hard to overstate how significant it was for the field of personality psychology to settle on a standard, consensus model, so enthusiasm is understandable. Claims that it was a universal model were made early and probably went too far, considering that the original studies were in very closely related languages and cultures.

Lexical studies in other languages have not consistently found the ‘Big Five’. Other common-denominator models have been proposed: For example, some suggest that adding a sixth factor with content related to Honesty integrates more results. Others suggest that only two factors may emerge as universal: Social Self-Regulation (content from Agreeableness, Conscientiousness, Honesty) and Dynamism (more or less Extraversion and Openness). Ultimately, however, it is impossible to test what is universal without evidence from all over the world.

Lexical Studies of Personality in African Languages

The initial lexical studies were in European languages, and later a few came from Asia, but Africa was missing. With over a thousand languages spoken in Africa, where to start? Our team decided to begin with one each in the three main language groups in Sub-Saharan Africa, in far-separated regions with different environmental, social and economic conditions, in order to reflect the great diversity in Africa. (And in written languages with a dictionary of translations to English.)

The first were conducted in small samples among the Maasai people, herders in Kenya and Tanzania, and among speakers of Supyire-Senufo, horticulturalists in Mali, and published in 2019. When restricted to two factors, the content in both languages matched the broad Big Two. A five-factor solution in Maa and a 10-factor solution in Supyire were the best descriptions of the local data, but these models did not look like the “Big Five”. There was a lack of any Openness content for one thing. The two languages had similar dimensions up to three-factors (virtue; well-being; power). But they included culture-specific content starting with the fourth.

A lexical study in a Khoesan clicking language

In 2018 in Namibia we embarked on the most ambitious lexical study yet in Africa, in Khoekhoegowab, the most widely spoken of extant Khoesan clicking languages. As in our prior work in Africa, but unlike previous lexical studies, we sought responses from adults throughout the country, rather than university students. In addition to a more systematic approach to data analysis, this was also the first lexical study to incorporate follow-up qualitative interviews after the survey: we sought the perspectives of Khoekhoegowab speakers from the community on how they use words from the study, and their understanding of Big Five concepts that did not show up in the study.  

A list of words that can be used to describe differences between individuals was extracted from the Khoekhoe dictionary, and reduced to the 272 most relevant. To succeed in getting data from throughout the country, we recruited a team of 15 school teachers of Khoekhoegowab. After a training weekend in Windhoek they went home and invited neighbors, colleagues, and acquaintances to participate, sometimes going door to door. Sitting privately, the interviewer asked the participant to think about a specific person they knew well. The interviewer read the list of terms aloud, and the participant rated how well each term described their target person.

We explored the resulting data set of over 500 cases, with ratings on the 272 terms, using factor analysis. This indicated which terms grouped together. For example, people who were described as helpful (sîsenhuixa) were also described as attentive and meticulous (ǃûiǃgâxa), but as not untidy (ǃkhaera(si)b/s).

Follow-up conversations with Khoekhoe-speakers

Some terms that grouped together, however, were harder to interpret, and we were not confident that all definitions in the Khoekhoe-English dictionary were adequate. For example, the word ǁgoaraxa was defined as ‘prone to blackmail’ but it was hard to picture what was meant.

We spoke to 23 Khoekhoegowab-speakers in six towns to find out. The examples they gave for how they used the terms in our study led us to update the definitions of 22 words to better summarize their usage. For example, we clarified the meaning of ǁgoaraxa to be: “someone who wants to be begged by others, creates situations where others are in a position to beg, and/or withholds in order to be begged”.

Local Understanding of Extraversion and Openness

We had other questions. Of the 272 most-used terms to describe people in Khoekhoe, none related to Big Five Extraversion or Openness. This absence has been reported in other African contexts and sometimes in Asia, but what does it mean? Are these distinctions unknown? Or have they just not been encoded into single words?

We asked about aspects of Extraversion (friendliness) and of Openness (creativity), and we found them to understood by Khoekhoegowab speakers very similarly to how they are used in the West. Terms related to these concepts are not frequently used by Khoekhoegowab-speakers to describe others, but when asked to describe someone high in these qualities, the examples they gave matched those Americans might give.

This is a good example of how personality description has both universal and culturally-specific aspects. On one hand, the lack of equivalent frequently-used terms suggests that these concepts have been less useful and relevant in other societies. But when we describe a trait of interest, people easily think of examples of people who are like that. This might be analogous to explaining an imported concept such as the Japanese amae (seeking to be indulged by an authority figure or spouse, acting unnecessarily dependent) to Americans, and asking if they can think of someone who embodies this trait. Many of us could come up with an example once it is explained, even if it isn’t a key concept locally.

Personality Structure in Khoekhoegowab

The patterns of association among the 272 terms suggested 11 dimensions for how Khoekhoe speakers think about and describe others (See Table, below). Eight had only small correlations (.30 or less), if any, with imported Big Two, Big Five, and Big Six scales. These unique dimensions may give us clues about the local culture.

Only three Khoekhoe factors (Prosocial Diligence, Bad Temper, and Fear vs. Courage) had meaningful overlap with imported scales. Of these, interestingly, Prosocial Diligence correlated with both Agreeableness and Conscientious. These factors are sometimes correlated in the West, but not as strongly. Their association might be stronger in collectivistic settings, particularly those under economic stress. In an individualistic context, one person’s Conscientiousness might be considered their own business: your lower grades or earning power, or worse long-term health, may be your own problem. But in contexts where obligations to extended family are the social safety net, your hard work or laziness impacts those around you.

What does it mean if the Big Five Isn’t Found in Africa?

A personality model created in one place can be used in others: people are similar enough that we can understand each other’s distinctions once they are explained. Can you think of someone you would describe as amae? What about ǁgoaraxa? Imported concepts can be a pleasure, giving us new ideas for how to think about human variation. Their ability to be translated and understood speaks to one of the universal aspects of personality description.

But the trait dimensions that arise naturally through a lexical study don’t just tell us about personality differences; they also tell us something about the local society. Extraversion, for example, is particularly important in places like the United States where people move often and interact with many strangers, where there is high ‘relational mobility’. In such contexts, Extraversion brings many advantages, and subtle differences in how extraverted people are become quite apparent. A rich vocabulary has arisen to indicate subtle distinctions: gregarious, friendly, outspoken, and brash, to shy, withdrawn, and laconic. Speakers of Supyire-Senufo, on the other hand, have a different lifestyle, commonly remaining in the same rural area, interacting with well-known others with whom they share the labor of raising food crops. There, a trait dimension of ‘laziness’ appeared; their most-used terms included many that defined subtle nuances in hard-working-ness, but almost none that related to Extraversion. It isn’t as useful to talk about.

In Asian and African studies, where Extraversion terms are found they often sort out differently with those of Agreeableness. The warmth, gregariousness, and positive emotion aspects of Extraversion tend to combine with Agreeableness’ trust and altruistic aspects, whereas assertiveness and excitement-seeking associate with dis-Agreeableness, something like ‘Abrasively Bold’. A highly positive view of Extraversion, as a dimension of ambitiousness and social skill, associated with being attractive and well-liked, may be specific to the United States, where it arguably defines an American personality ideal!

What does it mean for a Personality Model to be Universal?

If the dominant model had been first created in Africa or Asia, under the same pressure to choose a consensus model, it is possible that we might now use something very different than the Big Five. People have enough in common that this foreign-born measure would probably work acceptably in the United States, and its authors might claim it as a universal model. But we know that it would be missing some content, like Openness, that is meaningful to individualistic Americans and captures relevant distinctions between our possible roles and careers. It might also combine dimensions, like Extraversion and Agreeableness, or Agreeableness and Conscientiousness (as in Khoekhoe Prosocial Diligence), in ways that less precisely capture our local experience. It might include domains like Temperance (religiousness versus substance use), that don’t closely match local attitudes. This model could work to some extent, the same way that the Big Five, translated to dozens of languages and imposed around the world, also works to some extent. But it would likewise be an awkward and incomplete fit in many societies.

Hopefully the future will see more interest in the nuances of personality traits across contexts, and less fixation on the Big Five. If we can make room in our science for universal and for culturally-specific aspects of personality traits and models, they can teach us a lot about the societies from which they emerge.

Scientific, moral, and professional reasons to increase inclusion in research

• 

• 

• 

Originally posted October 19, 2020 to Psychology Today

People the world over have many similarities:  We all use metaphors, jokes, and insults, we have marriages and kinship rules, and group affiliations. Anthropologist Donald Brown compiled an intriguing list of hundreds of universal aspects of human societies. But we also know that psychological differences across cultures emerge early. Cultural upbringing determines things as basic as where your eyes will move when you look at a painting, and whether you fall prey to certain optical illusions. And even if all societies have marriages, how marriage partners are chosen and what we expect from the experience vary dramatically.

The vast majority of psychological research, however, is conducted on a small subgroup, the 11 percent of us who live in Western, industrialized countries. A recent report by my colleagues and I shows that over 90 percent of recent articles in six top psychology journals drew on such samples (over 60 percent of American). We know a lot about this narrow slice of the human family, but we are neglecting to learn about everyone else.

Studies with White or male samples are less likely to indicate participant characteristics in the title, as compared to those with more women or ethnic minority samples; such persons may still be treated as more “proto-typical” people. This is also true when it comes to the nationality of participants. Article titles speak of psychological phenomena (e.g. “self-esteem”, “language acquisition”, “personality predictors of life outcomes”), without specifying who we are talking about. There has been an unstated and convenient assumption that the college students near at hand in North America and similar societies can stand in for everyone. But this almost certainly isn’t true, and a mono-cultural research base deprives us of observations of psychological diversity and the new insights that might result. 

International research does exist. Our analysis focused on six particularly influential empirical journals, among the most cited in psychology, which function to some extent as “gatekeepers” to their subdomains. There are also hundreds of peer-reviewed journals specific to regions or with the word “international” in their names, whose missions are global.

This research, however, is less likely to reach a broad audience. During a visiting scholar stay at the national university of Malaysia last winter, I gave a workshop on publishing in international journals. Many participants described rejection letters from important journals in their areas, which suggested that their studies would be more relevant to a local journal and questioned whether a Malaysian study would interest an international audience.

I urged my Malaysian colleagues to point out to editors and reviewers that 60 percent of the world’s population lives in Asia, making Malaysia’s multi-ethnic Asian society more globally representative than the United States. Given the Muslim majority, they could also point to the relevance of their work to the 25 percent of the global population who identify as Muslim. There is no objective reason to consider a Malaysian sample less relevant than a Western one, but old habits (and biases) die hard.

Does this lack of representation in science matter? If you live in the United States and don’t plan to work or live abroad, should you care? I would argue: yes and yes. As leading cultural psychologist Fons van de Vijver wrote, there are moral, professional, and intellectual reasons that psychological science should be more international. The moral reasons seem to me the most self-evident. Ignoring most of the world, while supposing that psychology is a general science, echoes imperialist attitudes that modern scientists should make an extra effort to avoid.

Professionally and practically, more representative science means better tools and treatments for people all over the world. This includes the diverse population in the United States. A lot can go wrong when Americans export their ideas about diagnosis and treatment to other countries. Ethan Watters’ Crazy Like Us is a hair-raising but great read about how this can happen despite good intentions. In Namibia, where I’ve been conducting research, mental health needs are significant. There is trauma due to a long war for independence and the widespread HIV/AIDS epidemic. Training for psychologists and social workers relies on American materials and at clinics diagnoses are made with the Diagnostic and Statistical Manual of the American Psychiatric Association. This can lead to incongruities, which professionals are on their own to resolve. For example, a Namibian clinical psychologist told me that eye contact with the client was emphasized in her training in the United States and South Africa. But among some groups in Namibia, direct eye contact is seen as disrespectful and aggressive. While diagnoses such as depression and anxiety may have some worldwide similarities, the specific ways they will be spoken about and suitable ways to treat them can vary in important ways. 

Scientifically and intellectually, we have a lot to gain from better representation in our science. One benefit is new ideas. For example, Dixon Chibanda and his team in Zimbabwe adapted principles of cognitive-behavioral therapy to local needs and resources, in a way that maximizes accessibility and reach and minimizes costs. This may be found useful in many other settings, including in the United States. Another scientific benefit is increased precision and efficiency in science. As we understand cultural differences better, we can make better-targeted hypotheses about where a phenomenon or treatment is likely to work similarly. Understanding group differences can also account for unexplained variance in outcomes.

Finally, more international research can help us define the foreground of culturally specific aspects of psychology against the background of human universals. Among other things, this teaches us about ourselves. If a theory, model, or instrument created in the West also works in places with different social conditions and values, perhaps we’re on to something universal. In the industrialized West, there is a wealth of psychological knowledge, but often we don’t know which of these research findings define the human experience and which define our particular societies.

Making psychology more globally inclusive, however, is easier said than done. It is complex to translate measures to mean the same thing across contexts and to find methods that are analogous but socially and practically workable. For example, while it might be most efficient to collect survey responses online in the United States and Europe, this could lead to biased samples in Africa, among other places, where fewer people have computer access. For this reason, it might make the most sense for a local interviewer to read survey questions aloud.  No single study can answer our questions. It takes a multiplicity of creative approaches, seeking corroboration across sources, and taking time to get the bottom of conflicting findings. In future blog posts, I look forward to sharing examples and some of the challenges and joys of working towards a more globally inclusive science.