TL;DR: A 2025 paper by Akaliyski, Vignoles, Welzel and Minkov in the Journal of Personality and Social Psychology quantified, for the first time, how far off Hofstede’s individualism scores are. The result: English-speaking countries are overestimated by ~27 points, East Asian countries are underestimated by ~22 points on a 100-point scale. The replacement, built on the World Values Survey, is already operational at CultureFlow.
If you have ever attended a cross-cultural training session, read a book on international business, or Googled “cultural differences at work,” you have almost certainly encountered Geert Hofstede’s cultural dimensions. Power distance. Individualism versus collectivism. Masculinity versus femininity. Uncertainty avoidance. Long-term orientation. Indulgence.
These six dimensions have dominated the field of cross-cultural studies for nearly half a century. They are cited in thousands of academic papers, embedded in corporate training programmes worldwide, and treated, in practice, as settled science.
They are not settled science. And recent research is making it increasingly difficult to pretend otherwise.
The Data Problem
Hofstede’s original research was conducted between 1967 and 1973, using survey data from a single multinational corporation: IBM. The respondents were IBM employees in various countries, a self-selecting, demographically narrow sample of people who had all chosen to work for the same American technology company during the same historical period.
This is not a minor methodological footnote. It is the foundation on which the entire framework rests. When we say that Japan scores 46 on individualism or that the United States scores 91, those numbers derive from how IBM employees in those countries responded to workplace attitude surveys more than fifty years ago.
The world has changed considerably since 1970. Japan’s economy has been through a bubble and three decades of stagnation. The United States has undergone profound demographic and political shifts. China has transformed from a closed agrarian economy to the world’s second-largest GDP. Yet the Hofstede scores for these countries remain substantially the same numbers that emerged from that original IBM dataset.
Some updates have been made over the decades, and additional countries have been scored using various methods. But the core architecture, and many of the most widely cited scores, traces back to that single-company, single-era dataset.
The Criticism Is Not New
Academic criticism of Hofstede’s framework has been mounting for decades. Brendan McSweeney’s 2002 paper in Human Relations systematically dismantled several of the framework’s foundational assumptions: that nations are the appropriate unit of cultural analysis, that a single company’s employees can represent an entire national culture, and that culture can be meaningfully reduced to numerical scores on a handful of dimensions.
McSweeney’s critique was sharp but measured: the framework’s popularity in business, he argued, owed more to its simplicity and apparent precision than to its empirical rigour. Managers want a number they can point to. Hofstede provides numbers. Whether those numbers accurately represent what they claim to represent is a different question.
Other scholars have raised additional concerns. The GLOBE study, involving 17,000 managers across 62 societies, found significant divergences from Hofstede’s scores on several dimensions. Researchers working with the World Values Survey, a dataset that is larger, more recent, more representative, and continuously updated, have demonstrated that cultural variation is more complex and more dynamic than a six-dimension model can capture.
But none of this dislodged Hofstede from its dominant position. Until, perhaps, now.
New Research: Systematic Biases Revealed
A 2025 paper by Akaliyski, Vignoles, Welzel, and Minkov, published in the Journal of Personality and Social Psychology, provides the most precise quantification yet of where Hofstede’s scores go wrong. The researchers constructed a new Individualism-Collectivism index for 102 countries using data from the World Values Survey and the European Values Study, then compared it systematically against Hofstede’s original scores.
The findings are striking.
Hofstede’s scores overestimate individualism in English-speaking countries by an average of approximately 27 points on a 100-point scale. The United States, the United Kingdom, Australia, and Canada all appear significantly more individualistic in Hofstede’s data than the values-based evidence suggests.
Conversely, Hofstede’s scores underestimate individualism in East Asian countries by approximately 22 points. Japan, South Korea, and China all appear significantly more collectivist in Hofstede’s framework than the WVS data supports.
These are not small discrepancies. A 27-point bias is the difference between characterising a culture as moderately individualistic and highly individualistic. A 22-point bias in the other direction is the difference between collectivist and moderately balanced. When cross-cultural training programmes tell participants that “Japanese culture is highly collectivist” based on a Hofstede score of 46, they may be perpetuating a characterisation that is measurably inaccurate, and one that carries undertones of Orientalist stereotyping, whether intended or not.
The researchers attribute these biases to Hofstede’s original sampling method. IBM employees in the United States in the 1960s were not representative of American society. They were disproportionately white, male, educated, and corporate. IBM employees in Japan were equally unrepresentative of Japanese society. The biases in the sample became biases in the scores, which became biases in a generation of cross-cultural training.
Why This Matters for Business
If you are using Hofstede’s framework to prepare for a meeting with Japanese counterparts, you may be over-preparing for collectivism and under-preparing for the individual agency that Japanese professionals, particularly younger ones in internationally-facing roles, actually exercise. If you are onboarding a new hire from India and relying on Hofstede’s power distance score, you may be making assumptions about deference to hierarchy that do not match the reality of their specific professional background.
The problem is not that cultural differences do not exist. They do, and they matter enormously. The problem is that the most widely used tool for measuring those differences is producing distorted readings, and most people using it do not know that.
This has real consequences. Training programmes built on inaccurate cultural profiles waste time and money. Worse, they can reinforce stereotypes rather than dispelling them. A manager who has been told that “Chinese culture is collectivist” may unconsciously treat Chinese colleagues as interchangeable members of a group rather than as individuals with distinct perspectives and expertise. The cultural framework that was supposed to build understanding instead becomes a more sophisticated form of bias.
What Better Data Looks Like
The alternative is not to abandon empirical approaches to culture. It is to use better data.
The World Values Survey has been running since 1981 and is currently in its seventh wave. It covers over 100 countries, uses representative national samples (not employees of a single company), and is updated roughly every five years. It measures not six dimensions but hundreds of values, beliefs, and attitudes, from trust and authority to work ethic and tolerance, providing a far richer and more granular picture of cultural variation.
The WVS data confirms that cultural differences are real and systematic. But it also shows that they are more complex, more dynamic, and more nuanced than any single framework can capture. Cultures shift over time. Generational change is significant. Urban and rural populations within the same country can differ markedly. And individual variation within any culture is enormous.
Good cultural intelligence requires engaging with this complexity rather than reducing it to a scorecard. It means using multiple data sources, acknowledging uncertainty, and treating cultural profiles as starting points for inquiry rather than definitive classifications.
Hofstede vs Minkov-Kaasa: A Direct Comparison
It is one thing to say that Hofstede’s data has problems. It is another to see exactly how the alternative compares on every dimension that matters. The Minkov-Kaasa model, developed by Michael Minkov and Alexander Kaasa using World Values Survey data, represents the most rigorous contemporary alternative. Here is how the two frameworks stack up.
Data source. Hofstede’s scores derive from attitude surveys administered to employees of a single American multinational, IBM. The Minkov-Kaasa model draws on the World Values Survey, a nationally representative dataset designed specifically to capture values and beliefs across populations, not just corporate workforces.
Sample size. Hofstede’s original study covered approximately 100,000 IBM employees. The World Values Survey has collected responses from over 500,000 individuals across its seven waves, with each wave adding new respondents and new countries.
Data recency. Hofstede’s core data was collected between 1967 and 1973. Some supplementary data has been added over the decades, but the foundational scores remain anchored to that period. The WVS is continuously updated, with the most recent wave (Wave 7) covering 2017 to 2022.
Methodology. Hofstede used corporate attitude surveys designed for internal IBM purposes and later reinterpreted for cultural analysis. The WVS uses purpose-built survey instruments administered through nationally representative sampling, with rigorous protocols for translation, sampling, and data quality.
Country coverage. Hofstede’s framework covers approximately 76 countries, though many of these scores were extrapolated or estimated rather than directly measured. The Minkov-Kaasa model covers 102 countries, all directly measured through representative surveys.
Peer review. Hofstede’s original study was published as a book, not as a peer-reviewed journal article. The Minkov-Kaasa work and the 2025 Akaliyski et al. comparison have been published in the Journal of Personality and Social Psychology, one of the field’s most rigorous peer-reviewed outlets.
Dimension structure. Hofstede’s framework uses six static dimensions that have remained largely unchanged since they were first proposed. The Minkov-Kaasa approach treats cultural variation as dynamic and multi-dimensional, acknowledging that the number and nature of relevant dimensions may shift as societies evolve and as measurement improves.
The contrast is not subtle. One framework is anchored to a single company in a single decade. The other is built on representative data that is updated as the world changes.
What Hofstede Gets Right
Criticism of Hofstede should not become dismissal of Hofstede. The framework’s contributions to the field are genuine and significant, and understanding them is essential to understanding where the field stands today.
Hofstede pioneered the quantitative study of cultural differences in business at a time when culture was widely considered too vague and too subjective to measure. Before Hofstede’s work, cross-cultural business advice was largely anecdotal: travellers’ impressions, case studies, and received wisdom passed from one expatriate to the next. Hofstede demonstrated that cultural variation could be captured empirically and expressed in terms that business leaders could act on.
He established, more broadly, that culture is measurable. This sounds obvious now, but it was not obvious in the 1970s. The idea that you could survey people across countries, identify systematic patterns in their responses, and map those patterns onto dimensions of cultural variation was a genuine intellectual contribution that opened up an entire field of research.
The vocabulary Hofstede created remains useful. Terms like “power distance” and “uncertainty avoidance” give cross-cultural practitioners a shared language for discussing real phenomena. Even researchers who have moved well beyond Hofstede’s framework still use these terms as shorthand, because they capture something real about how cultures differ in their approaches to hierarchy, ambiguity, and social organisation.
Hofstede’s work also inspired the subsequent research that now supersedes it. The GLOBE study, the Minkov-Kaasa model, and the 2025 Akaliyski paper all exist, in part, because Hofstede demonstrated that this type of research was possible and valuable. The researchers who identified the biases in Hofstede’s scores built on the foundation Hofstede laid.
Finally, the framework’s simplicity made cultural differences accessible to non-academics. A six-dimension model with country scores is easy to teach, easy to remember, and easy to apply in a training room. That simplicity is also the framework’s greatest weakness, but it played a real role in getting cross-cultural awareness into boardrooms and HR departments where it had not previously been present.
The point is not that Hofstede was wrong about everything. The point is that the field has progressed, as all scientific fields should. Using Hofstede in 2026 is like using a 1970s road map in the age of GPS. The map was a remarkable achievement for its time, and you can still see the roads on it. But you would not rely on it for navigation.
How Cultural Dimensions Have Shifted Since the 1970s
One of the most fundamental problems with Hofstede’s framework is its implicit assumption that cultural values are essentially static. The scores were measured once, in a specific historical context, and then applied as though they described permanent features of national character. The WVS data tells a different story. Cultures shift, sometimes dramatically, and the pace of change has accelerated in the decades since Hofstede’s original research.
Japan is perhaps the most striking example. Hofstede assigned Japan an individualism score of 46, placing it firmly on the collectivist side of the spectrum. But WVS data collected over the past four decades shows significant generational movement toward individualism, particularly among professionals in internationally-facing roles. Younger Japanese professionals, especially those in technology, consulting, and finance, operate with a degree of individual agency and self-direction that Hofstede’s score does not capture. This is not to say that Japanese culture has become “Western” or that collectivist values have disappeared. It is to say that reducing an entire society to a single number, measured fifty years ago, obscures the complexity and dynamism of what is actually happening.
South Korea presents a similar pattern. Hofstede scored South Korea at 18 on individualism, one of the lowest scores in the dataset, characterising Korean culture as deeply collectivist. Since those scores were calculated, South Korea has undergone one of the most rapid modernisation processes in human history. Korean professionals working in technology, finance, and creative industries today operate in ways that bear little resemblance to the collectivist profile Hofstede assigned. The rise of Korea’s global cultural influence, from Samsung to K-pop to the country’s thriving startup ecosystem, has been accompanied by measurable shifts in values around individual expression, personal achievement, and authority. The WVS data reflects these changes. Hofstede’s scores do not.
The Nordic countries offer a different kind of lesson. Hofstede characterised Denmark, Sweden, and Norway as low on power distance and high on individualism, a profile that broadly aligns with the egalitarian consensus cultures these societies are known for. On the surface, this seems like a case where Hofstede’s scores have held up. But the picture is more nuanced than the scores suggest. Younger professionals in Oslo and Stockholm, particularly those working in startup and technology environments, show increasing comfort with hierarchical structures and decisive leadership styles that would surprise anyone relying solely on Hofstede’s low power distance characterisation. The egalitarian foundation remains, but it is being negotiated and adapted in contexts that did not exist when Hofstede was collecting data.
These examples illustrate a broader point. Culture is not a fixed attribute of nations. It is a living system that evolves in response to economic development, technological change, generational turnover, globalisation, and countless other forces. Any framework that treats cultural scores as permanent features rather than snapshots of a moment in time will inevitably fall further behind reality with each passing decade. Hofstede’s scores are now more than fifty years old. The gap between those scores and cultural reality is not shrinking. It is growing.
Where CultureFlow Sits
This is the approach we have built CultureFlow around. We draw on the World Values Survey, peer-reviewed research in cultural psychology and intercultural relations, and continuously updated datasets to provide cultural intelligence that reflects how cultures actually operate today, not how they appeared to IBM employees half a century ago.
We do not reduce culture to a handful of numbers. We use multiple validated frameworks to build contextual, nuanced cultural profiles that account for industry, generation, and individual variation. When the data says something is uncertain, we say so.
Hofstede’s framework was genuinely pioneering. It opened a field that barely existed before. But pioneering work is, by definition, the starting point, not the destination. The cultural intelligence field has moved on. The data has moved on. It is time the tools caught up.
Hofstede reduces you to a national average. Your Cultural Signature does the opposite. It reads how you specifically communicate, decide, and build trust, drawing on three decades of peer-reviewed data. See your Cultural Signature →