The Challenge Network

   back   menu   next   

National distinctiveness

National distinctiveness

This text presents work that sets out to map nations with respect to their overall relatedness or distinctness. It draws on two sources of data in order to do this. One set is World Bank-derived figures, with the constraints which self-reporting implies for accuracy. The second are data from an Austrian foundation, which set itself to measure national attitudes, and found four significant dimensions along which differences can be arrayed. More detail about this work is given in the Appendixes.

It is necessary to understand something of the methodology to make sense of what follows. An initial sample of over seventy nations were pruned to a sample of over fifty in which the data were satisfactory. Sub-Saharan Africa was consolidated into the Republic, South-East and West Africa, and the 'Arab world' was lumped together, not through our choice but because this was how the attitude data were reported. Eighteen 'hopeful' variables (out of a far larger range of candidates) were reduced to twelve, shown in blue. Those which were excluded are coloured red, below. The choice was made following tests of statistical significance.

No useful measure of institutional viability was found (measuring corruption, perhaps, or the stability of governments) and this seems likely to be a gap that needs to be filled. The selected variables together generate a range of statistical artefacts which we shall describe, in which typically around 70% of the inter-country variability is described in two organising 'factors'. The first factor captures around 51% of the variability. The countries most associated with these dominant factors are listed in Appendix II.

The seems the appropriate place to introduce the idea of a dendrogram. This is a tree-like structure, as shown below. Nearest neighbours, in the statistical sense, are drawn as adjacent leaves on a 'tree'. Branches thus define families, and the distance down the chart at which these branches meet expresses the difference innate to these families and individual variables. The figure shows the relatedness or otherwise of the significant variables which were described above. GNP per capita is, therefore, very close to the measure of individualism that has been used, as well as to state expenditure: rich countries tend to spend more of the GNP on public goods. Related issues cluster on a linked branch, in the centre of the chart.

These clusters are both very distinct from the family of issues on the right of the chart: reverence of authority, high levels of income inequality, a sense that emotionality ('affect') is appropriate in decision-taking and the degree to which the nation in question scores as being a risky place in which to invest.

The variables have been processed in a number of ways. The 'values' data were compared to income per capita. As we get richer, a variety of things happen to the ways in which we see ourselves and the world around us. These are reported in the first section.

The data, once purged, we then subject to cluster and factor analysis. Factor analysis looks across the data with the aim of finding ways of summing it up, and of partitioning the countries across the form of description that this produces. The outcome is rather like the consultants' two-by-two chart, with the analysis suggesting what should be on the axes.

Cluster analysis examines the nations to see which are most related and which the most distant, creating what is called a dendrogram. This work has been done for nations on values data, on 'economic' data and on a composite of the two. This and the factor analysis are presented in the second section. A short third section suggests some thoughts as to what all of this might mean.



Appendix I offers more detail of the values dimensions. These have been extracted from the responses of a large number of people from around the world, covering many countries. The statistical processing created four universal factors ('dimensions') which explained the differences in the data. A fifth is occasionally significant.

These values were combined with the income per capita of the nations in question. The resulting table was tested for its simple and more complex properties.

At a simple level, nations which became richer were, on the whole, less prone to defer to authority and more likely to be individualistic, in the sense of questioning how things ought to be done and setting personal, rather than communal goals. This can be captured in the two charts which follow. Both show the same data, but the second is 'zoomed in' onto the wealthy nations. The convention is that the area of the circle shows the strength of individualism in the nation, the vertical location the proneness to deference to authority and the horizontal position, the GNP per capita.

The trends that exist with economic development are self-evident. In the cadre of the old developed countries, however, only France and Belgium show themselves as markedly distinct. Denmark is strikingly un-deferential, whilst Singapore, Hong Kong and Japan are the opposite. The Asian nations are also less individualistic than their economic status would suggest, perhaps true to cliché.

Below, a second chart shows the detail of the cluster in the lower right of the chart, of which the UK is a member.

More sophisticated analysis allows us to build dendrograms, of the sort demonstrated above, and also allows us to define statistically-meaningful clusters of countries. These are taken to a higher level of sophistication in the next section, where we combine values data with the broader economic and behavioural factors, so we shall pass over these results. However, creating such clusters generates four groups from amongst these countries. These are shown as the numbers 1 to 4 in the key.

This chart uses exactly the same axes as were employed on the charts which were shown above. It excludes, of course, the variable represented by the area of the circle. Instead, the four symbols show membership of the four distinct ways in which nations tend to cluster, as derived from the data series that describe them. The identity of many of the points can be read off from the preceding charts, in which the circles were labelled.

Bringing together the variables.

Bringing together the variables.

The next stage of analysis supplements the 'values' data with a much wider range of factual information. The identity of these variables is detailed in the opening section. They were, however, chosen to reflect political or other choice which was made within the nation or perceptions - such as risk - that were assessed from outside. These data series were significantly winnowed. Only one of the 'values' dimensions dropped out of the analysis - the tolerance of ambiguity. The resulting analysis resolved around 70% of the variance within the data. The results are presented as a dendrogram, below, and as a factor analysis which is shown in Appendix II.

The dendrogram which is shown above represents the best family tree for the nations which survived this winnowing. The results are both intuitive and, on occasions, surprising.

On the left of the figure, there is a family of developed nations. The US is part of a sub-family which contains Australia and New Zealand, but is also closely related to Switzerland and Germany.

Britain, by contrast, is most like Ireland, then Holland and finally similar to Canada, these two clusters merge. Norway and Finland are very similar, but distinct from any of the foregoing.

There is then a significant jump to a family in which Belgium, France, Italy and - surprisingly, Japan - are clustered. It is worth noting that these two major families are more different from each other than - say - Singapore and Hong Kong are distinct from Chile, Salvador and Guatamala.

Israel, Uruguay, Jamaica and South Africa make up this major group. There is then a major leap to the cluster of nations on the right of the chart. Hong Kong and Singapore are very like each other, but hugely different from the industrial countries despite their similar levels of per capita income. This section is, however, filled with the recently industrialised, developing or poor nations. There are some initially surprising - but often intuitive - fellows, such as Colombia and Turkey.

One can quarrel with some of the minor details on this figure. The fine tuning of it is relatively sensitive to the conditions which are imposed on the processing. An excluded variable will have mild effects on the arrangement that results. In general, however, the structure is robust and represents an objective measure (with all the fallibility of the data understood) of how the principle nations relate to each other.

One of the most striking features of this analysis may be the sharp differences amongst the industrial nations. Near neighbours - such as the EU countries - are far more distinct than are, for example, the North Americans, the Germans and the British. Indeed, the closeness between the Germans and the British may be one of the surprises of this figure. Belgium, France, Italy, Spain Greece and Portugal are closer to Japan than they are to their European partners. This speaks ill for future integration.

The previous section showed a scatterplot in which the nations were segmented into groups. The complete analysis allows us to carry out this segmentation in many ways. What follows are a few pertinent snapshots.

The segmentation essentially runs from left to right across the dendrogram. Group 1 includes those on the extreme left. The Mediterranean Europe - Japan branch constitute group two. The third group comprises everything on the right of the chart, including all of the poor nations. Singapore is clearly an anomaly.

The second figure compares external perceptions of risk, again using the same conventions. It is worth noting that Group One countries are viewed as much safer on average than Group Two members.

Contrasting two values-related dimensions, we find tight clustering. The group one nations relish individualism and shun authoritarian approaches. The groups three nations defer to authority and have no great striving for self-direction, preferring the strength of the community. Group two - Southern Europe, Japan - cannot choose between these poles. An anomaly in this group is Israel, which has no taste for authority, but is not strongly individualistic.

In conclusion: the data do not allow us to say whether Britain is 'more like Europe than the US'. Europe is itself very heterogeneous, and whilst Britain is indeed very like some parts of it, and like North America, there is a huge gulf between some parts of the EU and others. Europe is not much 'like itself'.

As was indicated in the introduction, this work has been much extended, based on more replicable data and stronger statistical techniques. The results are presented in The Engines of Change .

Appendix I

Geert Hofstede "Promoting a European Dimension of Intercultural Learning - Developing School Materials", EFIL Seminars, Vienna 1998.

Four dimensions are reported as orthogonal and significant. A fifth dimension is patchily reported and is omitted from subsequent analysis. What follows are the most significant indicator questions, which also give a flavour of the derived dimensions.


Attitude to authority. (High values = authoritarian)

I feel most comfortable in a country where:


Affect and social distance. (High values = individual, material)

I feel most comfortable in a country where .


Individual and communal values. (High values = individual)

I feel most comfortable in a country where .


Tolerance of multivalent ambiguity. (High values = anxious)

I feel most comfortable in a country where .


Confidence & prospective orientation (High values=prospective)

I feel most comfortable in a country where:

Appendix II

The factors which were extracted class countries into up to four groups. The two most significant factors capture most of the difference between countries. A list is given below of those countries which exemplified the individual factors most strongly.

Factor One has, at one extreme, nations which are poor, which do not spend on public health or, indeed, on other public goods. Their population grows rapidly. They accept authoritarian rule and relatively high levels of inequality.

Factor two nations tend to be wealthy, individualistic, to prefer an unemotional and impersonal public life and to use the internet heavily. They use proportionately high amounts of energy to generate a unit of value. They may have a large amount of heavy industry. They tend to be neutral to negative with respect to the dichotomies of Factor One.

Two other dimensions can be extracted, expressing only a few percent of the residual variance amongst the countries. Factor three countries were relatively wealthy, but were more accepting of authoritarianism, and tended to be emotional in their public life.

The nations used were clustered by their affiliation to these factors. Nations can score heavily on more than one , and thus appear in more than one column. The results are shown below. The order in which they are listed shows the relative strength with which they expressed the factor in question.

Factor One countries Factor Two countries Factor Three countries
Ecuador Sweden Italy
Panama Canada Spain
Venezuela Norway Belgium
Guatemala Denmark Japan
Mexico Holland Germany
Thailand USA Greece
West Africa Australia Israel
Salvador New Zealand Argentina
Chile Switzerland Uruguay
Indonesia Finland
Columbia Singapore
East Africa Britain
Arabic world Germany
Turkey Israel
Philippines Ireland
Uruguay Hong Kong
South Korea Belgium
Portugal Japan
Argentina France
Costa Rica
South Africa
Hong Kong
  To top  
 to the top 

The Challenge Network supports the Trek Peru charity.