“The world has divided into liberal and illiberal spheres …”
The Centre for the Future of Democracy at Cambridge University, October 2022
My paper is a commentary on the report, A World Divided*, but its aim is to reflect on the abstract structure of world opinion. As such it is a companion piece to a paper two years ago about public opinion within a single country, The Abstract Structure of Public Opinion.
* A World Divided: Russia, China and the West.
https://www.bennettinstitute.cam.ac.uk/wp-content/uploads/2022/10/A_World_Divided.pdf
Gordon Burt 29 November 2022 (Fourth Draft)
Contents
1 Introduction
2 The report “A World Divided”
3 Conceptualising ‘liberal’ and ‘divide’
4 Methodology
5 Conceptualising opinion space
Results
6 UN voting about Russia
7 Opinion about Russia
8 Opinion trajectories
9 Opinion distributions
10 Changing opinion
11 Opinions and attributes; ‘liberal’ and ‘divide’
12 Conclusion: A world distributed in multidimensional space
Appendix: An overview of the report “A World Divided”
Contents in detail
1 Introduction
2 The report “A World Divided”
A list of Figures and variables in the report
3 Conceptualising ‘liberal’ and ‘divide’
Specifications of ‘liberal’ countries; and the measurement of ‘liberal’
‘Divide’: the logic of categories and continua
A liberal divide by definition?
Liberal democracies or European Empires?
4 Methodology
Transforming the variables
My underlying distribution approach to the analysis of data
5 Conceptualising opinion space
Opinion scores for low-intensity relationships
Positive and negative opinion scores
Two-dimensional opinion space
Opinion scores in cooperative and conflict relationships
Three-dimensional opinion space
n-dimensional opinion space
The Odd Number Theorem for a social system
Types of distribution … two-dimensional Q2 and Q3
Two rotated factors in two-dimensional space
Results
6 UN voting about Russia
UN voting on Russia, 2014 to 2022
UN voting on Russia, four occasions, 2014 and 2022
Divided in two? - a symmetry argument
Scores on the underlying distribution
UN votes with population voting
7 Opinion about Russia
A variety of distributions for votes and opinions about Russia
Two maps: UN votes against Russia 2014-2022; opinion of Russia in 2022
World public opinion about Russia, 2012 and 2022
8 Opinion trajectories
Superpower opinion, 1952-2022
Opinion in developed and developing countries, 2012-2022
Opinion in geographical groups, 2012-2022
9 Opinion distributions
Maps: the geographical distribution of opinion, 2022
Two-dimensional USA and Russia-China opinion space: NP, PN and PP
Two-dimensional Russia and China opinion space: PP, NN and NP
Three-dimensional USA, Russia and China opinion space
10 Changing opinion
The change in opinion about Russia, 2021-2022
The changing correlation between USA and Russia-China opinions, 2008-2022
The distribution of opinion, 2012 and 2022; Russia, China and America
11 Opinions and attributes; ‘liberal’ and ‘divide’
Opinion about China and China investment commitments
Liberal democracy and the opinion about Russia and China combined
Satisfaction with democracy; and the opinion about Russia and China combined
Social Liberalism and the opinion about Russia and China combined
Extreme ‘social illiberalism’ – as opposed to Russia, China and the West
Change in Social Liberalism, 1990-2022
Value depends on social distance … single-peaked value functions on multidimensional attribute space
12 Conclusion: A world distributed in multidimensional space
Appendix: An overview of the report “A World Divided”
5. The New Structure of Global Public Allegiances 8/11
6. Visualising a Decade of Rising Geopolitical Polarisation 12/15
7. Regional Divergence Over Time 15/18
8. Why is the World Dividing in Two? 22/25
1 Introduction
“A World Divided: Russia, China and the West” is the latest report from The Centre for the Future of Democracy at Cambridge University. The authors have merged data from 30 global survey projects covering 137 countries representing 97% of the world’s population, the respondents giving their opinions about Russia, China and the USA. The report analyses the data and discusses it in depth. The first of six key findings is that:
“the world has divided into liberal and illiberal spheres …”.
An overview of this valuable report is given in an extended appendix below.
[“Non-western world still looks up to Putin.” The Times, October 21, 2022: 18.]
There is a great deal in this rich report that I have yet to engage with. What I want to do now though is the limited task of looking at the numbers and the Figures and asking:
How do the numbers in the report relate to the notion that the world is divided?
2 The report: A World Divided: Russia, China and the West
Foa, R.S., Mollat, M., Isha, H., Romero-Vidal, X., Evans, D., & Klassen, A.J. 2022. “A World Divided: Russia, China and the West.”
Cambridge, United Kingdom: Centre for the Future of Democracy
The report, direct access: https://www.bennettinstitute.cam.ac.uk/wp-content/uploads/2022/10/A_World_Divided.pdf
Link to report: https://www.bennettinstitute.cam.ac.uk/publications/a-world-divided/
Blog: https://www.bennettinstitute.cam.ac.uk/blog/a-world-divided/
Story: https://www.cam.ac.uk/stories/worlddivided
See the appendix for an extended overview of the report.
A list of Figures and variables
Note: In the pdf via the link, each page of the report has two page numbers. The first number is the page number in the paper report. The second page number is the slide number in the pdf.
https://www.bennettinstitute.cam.ac.uk/wp-content/uploads/2022/10/A_World_Divided.pdf
F1 3/6 Figure 1; paper page 3; pdf file page 6
F1 3/6 percent UN votes against Russia; world map of countries
F2 4/7 positive view of Russia; world map of countries
F3 5/8 change in positive view of Russia 2021-2022
F4 6/9 countries in the project; world map of countries
T1 7/10 sources, surveys, countries and years
F5 8/11 positive view of US; positive view of Russia and China combined
F6 9/12 difference between US and China positive views;
world map of countries
F7 10/13 positive view of US, Russia and China, 2012-2022; all, developing and developed
F8 11/14 positive view of Russia; positive view of China
F9 11/14 correlation 2008-2022, positive view of US; positive view of Russia and China combined
F10 12/15 distribution of positive view of Russia, 2012 and 2022
F11 13/16 distribution of positive view of USA, 2012 and 2022
F12 14/17 distribution of positive view of China, 2012 and 2022
F13 15/18 positive view of Russia, 2012-2022; by region
F14 16/19 positive view of Russia; by region, particularly Africa
F15 17/20 positive view of China, 2012-2022; by region
F16 18/21 positive view of China; world map of countries
F17 19/22 positive view of US; 2012-2022; by region
F18 20/23 positive view of US; world map of countries
F19 21/24 12 difference between US and Russia/China positive views; 2002-2022; West, Latin America, rest of Global South
F20 22/25 Liberal Democracy Index; negative view of Russia/Chain
F21 23/26 Dissatisfied with the functioning of democracy; positive view of Russia/China
F22 24/27 US opinion of Russia and China, 1952-2022
F23 25/28 Chinese investment commitments; positive view of China
F24 26/29 Social Liberalism Index, 1990-2020; high-income democracies, rest of world
F25 27/30 Social Liberalism Index, negative view of Russia/China
F26 34/37 Validation of survey sources
3 Conceptualising ‘liberal’ and ‘divide’
Specifications of ‘liberal’ countries; and the measurement of ‘liberal’
“Key Finding 1. The world has divided into liberal and illiberal spheres …”
The notion of ‘liberal’ is central to the report. I shall refer to it as notion L. Notion L is referred to by a number of different words. Each word refers to a set of countries. Different words refer to different sets of countries. However there is considerable overlap between these different sets. I shall refer to L-words and L-sets.
A key issue is whether statements based on one L-word can also be made when based on a different L-word. For a given L-word, how many such countries are there? … and what are they?
Specifications of ‘liberal’ countries
The following L-words have been used:
West, western democracies, maritime alliance democracies, liberal, liberal democracy (index), high-income democracies, developed, social liberalism (index),
And there are words for subgroups:
USA, European Union, Pacific North Asia, Anglo-Saxon democracies, Latin America, fragile democracies.
Note particularly the distinction between developed and democratic. For example Saudi Arabia is developed but not democratic, and India is democratic but not developed.
“Developed Countries (PPP GDP per capita > $35,000). They are 37 in number.
… by order of population: United States (only for Russia and China figures), Japan, Turkey, Germany, France, United Kingdom, Italy, South Korea, Spain, Canada, Poland, Saudi Arabia, Australia, Republic of China (Taiwan), Romania, Netherlands, Belgium, Czechia, Greece, Sweden, Portugal, Hungary, Israel, United Arab Emirates, Austria, Switzerland, Denmark, Finland, Slovakia, Norway, New Zealand, Ireland, Kuwait, Lithuania, Latvia, Estonia, Brunei.” [Figure 7, 10/13]
High-income democracies (World Bank 2022 high income threshold): They are 40 in number.
Australia, Austria, Belgium, Canada, Chile, Croatia, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, South Korea, Latvia, Lithuania, Luxembourg, Malta, Netherlands, New Zealand, Norway, Poland, Portugal, Puerto Rico, Romania, Slovakia, Slovenia, Spain, Sweden, Switzerland, Taiwan (ROC), Trinidad and Tobago, United Kingdom, United States, Uruguay.
Three groups defined by intervals on the Liberal Democracy Index scale are: Illiberal Democracies / Authoritarian Regimes, [0-30]; Fragile Democracies [30,60]; and Liberal Democracies [70,100].
[Figure 20, 22/25].
[A thought: are there liberal authoritarian regimes? … fragile authoritarian regimes?]
Subgroups are:
Pacific North Asia refers to Japan, South Korea, and the Republic of China (Taiwan).
Anglo-Saxon democracies refers to the United States, United Kingdom, Canada, Australia and New Zealand.
[Figure 24, 26/29]
Likewise there are a variety of ways of referring to the rest of the world.
The measurement of ‘liberal’
Three variables relevant to ‘liberal’ are discussed towards the end of the report: the Liberal Democracy Index; Dissatisfaction with the functioning of democracy; and the Social Liberalism Index. They appear in the following four figures.
F20 23/25 Liberal Democracy Index; negative view of Russia/Chain
F21 24/26 Dissatisfied with the functioning of democracy; positive view of Russia/China
F24 27/29 Social Liberalism Index, 1990-2020; high-income democracies, rest of world
F25 28/30 Social Liberalism Index, negative view of Russia/China
‘Divide’: the logic of categories and continua
“Why is the World Dividing in Two?”
Set theory is a foundational topic in mathematics. The definition of a category involves a specification of a set of objects. There is a complementary set containing all the objects which are not in the first set. A set has subsets. A set can be partitioned into a set of mutually exhaustive and exclusive subsets.
One notion of a divide might be that there are two sets which are mutually exclusive: if an object is in one set, then it is not in the other.
A stronger notion of a divide is when there are two sets which are different from one another. For some attribute, if the objects in one set have that attribute then the objects in the other set do not have the attribute.
Consider the case where the attribute is a continuum of numbers. A divide might be where there is no overlap in the range for one set and the range for the other set. We might think of the overlap as a matter of degree. So divide is a matter of degree. The amount of divide is inverse to the amount of overlap. This relates to the notion of significant difference and shared variance in statistics.
A quite different notion of divide is when two actors have mutually negative attitudes to one another: antagonistic divide.
Two questions are:
Is there a divide?
How many different groups are there in the divide?
Consider the case of a distribution over a continuum. One might think of a continuum as an infinite divide. On the other hand, if the distribution is unimodal one might want to say there is no divide. If the distribution is bimodal one might want to say that the whole is divided into two. In general, the nature of the divide might be said to be depend on the number of modes.
In the report, some of the variables are unimodal and some are bimodal – and one or two are trimodal … or more.
A liberal divide by definition?
Thinking about the relationship between a continuous distribution and a categorical divide, I came up with the following thought.
Liberal countries are more liberal than non-liberal countries.
The more liberal a country is the more likely it is to be a liberal country.
An analogy would be the continuous variable height and dividing people into two groups, tall people and short people.
Liberal democracies or European Empires?
Looking at countries that are listed as liberal democracies is that their defining characteristic? Are they basically European countries, along with their European-ruled empires? For example. the report defines “Anglo-Saxon democracies” refers to the United States, United Kingdom, Canada, Australia and New Zealand.
4 Methodology
The number and the weighted sum of countries
Aggregating the opinions of countries, we can simply take the number of countries or we can take the weighted sum of the countries. For example, the population can be taken as a weight. The Figures in the report show a country not as a point but as a circle, the area of the circle relating to the population of the country.
Transforming the variables
Transforming the variables
.(1) In some cases the scale is simply reversed with the bottom end becoming the top end.
.(2) In one case the logarithm of the percentage is taken rather than the percentages itself.
.(3) A different transformation of percentages is to use z-scores – but this is not used in the report.
One-sided aggregate percentages … underlying mean
The most common variable is sum of the percentages for those indicating any of the positive responses. The sum of negative responses is also used on occasion.
In the section on UN voting, I illustrate my own method of calculating the underlying mean.
My underlying distribution approach to the analysis of data
Survey results are sometimes presented in the form of one-sided aggregate percentages - for example “the percentage having a favourable opinion of X”. However this percentage loses some of the information in the data. It loses the percentage of ‘don’t know’; it loses the percentage ‘unfavourable’; and it loses the distinction between ‘fairly favourable’ and ‘very unfavourable’.
In order to use all the information, I prefer to assign a score to each of the options in the raw data and to calculate the mean.
Also, underlying the options offered in the survey, I hypothesise a continuum of scores – a finite continuum. I then standardise the continuum into [0,1] or [-1,1], depending as the scale is unipolar or bipolar.
I consider the options to correspond to sub-intervals. These sub-intervals are taken to be equal (accepting that further empirical research may indicate sub-intervals of varying sizes). For example, four options may correspond to the four sub-intervals
[-1,-0.5], [-0.5,0], [0.0.5] and [0.5,1].
The options are assigned scores equal to the mid-point of their sub-interval. In the example, the scores are -0.75, -0.25, +0.25 and +0.75. (Each score can be considered as an approximations to the mean score for the individual selecting that option).
Using the option percentages and the option scores, the mean score can be calculated. Other statistics can be calculated such as standard deviation and ‘polarisation’.
From a practical point of view, my mean may give much the same results as the one-sided aggregate percentage. (See pages 21-22 in The abstract structure of public opinion: the racism protests 2020).
From a different practical point of view, I feel my mean gives a more balanced, less dramatized account.
I also feel that an emphasis on the distribution of scores provides a more balanced, less dramatized account.
I feel that more dramatized accounts strengthen conflictual and polarised discourse.
5 Conceptualising opinion space
The focus of the report is different countries’ opinions about Russia, China and the USA. It is one aspect of the relationship between the countries and the three major powers. As with any relationship, key aspects are the intensity of the relationship and its positivity or negativity.
The focus here is on positivity/negativity. Each country has a percentage positive score for each of the major powers. Taking a 50% criterion the percentage is taken as positive ‘P’ or negative ‘N’. Opinion about two major powers gives two percentages giving rise to a pair of labels such as ‘PN’ – and four quadrants in two dimensional space. Opinion about three major powers gives three percentages and a triplet of labels such as ‘PNP’ – and eight octants in three dimensional space. Similar thinking can be applied to spaces of higher dimensions. PP reflects Cooperation while PN and NP reflect conflict. The Odd Number Theorem – see below - considers how conflict relates to the number of groups in a system.
Positive and negative opinion scores
The report refers to countries as being positive or negative about Russia or China or USA. This is based on the favourability score x. It is a percentage somewhere between 0% and 100%. We say it is positive P if it is more than 50% and negative N if it is less than 50%. The scale is split into two halves N and P with 50% in the middle.
. N : P
0 50 100
[Note that the score is the percentage of positive replies. I am unclear about whether all the other replies are negative - as opposed to having some don’t knows. I suspect the latter.]
Two-dimensional opinion space
Consider the favourability score x for Russia given by Ukraine. It is a percentage somewhere between 0% and 100%. We say it is positive P if it is more than 50% and negative N if it is less than 50%. The scale is split into two halves N and P with 50% in the middle.
. N : P
0 50 100
Similarly with the favourability score y for China given by Ukraine.
. N : P
0 50 100
We can now put the two scores x and y in a graph. It has four quadrants NN and PP; and NP and PN.
| y :
| NP : PP
| :
| :
| NN : PN
|. : x
0 50 100
Opinion scores in cooperative and conflict relationships
Suppose there is cooperation between countries A and B. Then A is positive about itself and also positive about B. A’s two opinion scores will give a point in quadrant PP. Similarly B is positive about itself and also positive about A. B’s two opinion scores will give a point in quadrant PP.
In a world where there is cooperation everywhere, all the countries will have points in the top right quadrant PP.
In a world where there is mostly cooperation everywhere, most of the countries will have points in the top right quadrant PP.
Suppose there is conflict between countries A and B. Then A is positive about itself and negative about B. A’s two opinion scores will give a point in quadrant PN. B is positive about itself and negative about A. B’s two opinion scores will give a point in quadrant NP.
Now consider a world where there is mostly cooperation but also a conflict between A and B. There will be points in three quadrants: most of the points in the cooperative PP quadrant; a few points in the PN quadrant for A and its allies; and a few points in the NP quadrant for B and its allies; and none in the NN quadrant.
Now consider the trajectory of the relationship between countries A and B. Suppose it is an alternation between cooperation and conflict … perhaps with varying degrees of cooperation and conflict. The trajectory of the world will be an alternation between cooperation everywhere and mostly cooperation but also a conflict between A and B.
Let us now use the quadrants to look at the whole scatter of points. If the points are in the three quadrants NP, PP and PN, then there may be a curvilinear relationship between x and y. In particular it may have the points (100%,0), (0,100%), (74%,74%) on it – in decimals (1,0), (0,1) and (0.74, 0.74). In other words the curve may be the top right quarter of the unit circle, x2+y2=1. There are segment of the quarter circle in the NP, PP and ON quadrants. Possibly!
Now let us return to the trajectory of the relationship between countries A and B.
An alternation between cooperation and conflict, perhaps with varying degrees of cooperation and conflict, will correspond to an oscillation between the full quarter circle and a lesser arc of that quarter circle – and a corresponding oscillation in the length of the arc.
Three-dimensional opinion space
We now consider three scores x, y and z – for Russia, China and USA – given by Ukraine. The three scores give a point in three-dimensional opinion space. Just as two dimensional space had four quadrants so three-dimensional space has eight octants:
z<50
| y :
| NPN : PPN
| :
| :
| NNN : PNN
|. : x
0 50 100
z>50
| y :
| NPP : PPP
| :
| :
| NNP : PNP
|. : x
0 50 100
n-dimensional opinion space
In some surveys there are many countries giving opinions about each other, That gives a space of many dimensions.
The Odd Number Theorem for a social system
It is being suggested that the world is dividing in two. But might it not instead be dividing in three? Rather fancifully I have dreamt up the Odd Number Theorem! The theorem states that the number of groups is always an odd number.
Theorem In any social system, if there are n two-sided independent conflicts, then the system is divided into (2n+1) groups.
Proof Each conflict involves two opposing groups. The n conflicts are independent and so there are 2n groups in conflict. There is also the non-aligned group consisting of all those who are not partisan in any of the conflicts. So (2n+1), an odd number.
Corollary If there is only one conflict then n=1. So there are three groups, an odd number …
… as in ‘third party’, Third World, etc. Later I shall be looking at one of the graphs and suggesting that it shows three groups: the West, Russia and China combined, and the Rest.
Moreover there is more than one conflict in the world and so I would argue that the number of groups is an odd number, larger than three.
And then I shall emphasise that countries are not so much divided into groups but rather distributed in multi-dimensional space …
Types of distribution … two-dimensional Q2 and Q3
One kind of one-dimensional distribution seems quite common – let us refer to it as the C distribution. The C distribution is fairly evenly spread across the range, skewed somewhat towards one end of the range and mildly bimodal, with a strong mode and a weak mode. The strong mode is associated with more points and the weak mode is associated with fewer points.
[The weak mode may involve countries of the West.]
One kind of two-dimensional distribution seems quite common – let us refer to it as the Q2 distribution. The Q2 distribution appears as two blocks of points, a stronger block and a weaker block. The two blocks are diagonally opposite. So there is a correlation between the x and y variables. The strong block is associated with more points and the weak block is associated with fewer points.
Variation within the block may be as great as variation between the blocks. The regression line may be a piecewise regression rather than the usual single straight line.
[The weak block may involve countries of the West.]
Also common is the Q3 distribution which also exhibits blocks. There are blocks of points in three quadrants this time, but almost no points in a fourth “sparse quadrant”. The quadrant opposite the sparse quadrant is a “non-conforming quadrant” – non-conforming for what would be a correlation based on the other two opposite quadrants. (The quadrants are suitably defined for there to be a sparse quadrant.) The appropriate regression line may be curved rather than straight (or there may be a suitable transformation of the variables).
Two rotated factors in two-dimensional space
Many of the figures in the report give two-dimensional graphs with a scattering of points. Let us call the two dimensions x and y. Of interest is the distribution of x and the distribution of y. Parameters of each distribution include the mean and the standard deviation. Also of interest is the bivariate distribution. An important parameter here is the correlation r and ‘the proportion of variance explained’, r2 (sometimes expressed as a percentage of variance explained).
The scatter of points spread across the graph and sometimes it looks a bit like an ellipse with the major axis along a diagonal line and the minor axis along a second diagonal line at right angles to it …
… This provides the basis for principal component analysis (or factor analysis). The major axis is the first principal component or factor; and the minor axis is the second principal component or factor. We refer to ‘rotated factors’. Each rotated factor is some combination of x and y.
Statistical theory attaches special importance to the normal distribution and to the bivariate normal distribution. The normal distribution is a special type of unimodal distribution. Knowing the type of distribution allows a variety of inferences to be made. (Note however that we shall not be doing this here.)
Results
We now turn to the empirical results. Section 6 is about the opinion of states - about
voting at the UN on the Russian invasion of Ukraine. The remaining sections are about the opinion of people – about public opinion in various countries. Sections 7 to 9 look at opinion about three major powers: Russia, China and the USA. They look in turn at opinion trajectories, opinion distributions and changing opinion. Section 10 looks at the relationship between opinions and a number of country attributes, including some attributes related to the notion of ‘liberal’.
In a variety of situations we shall be continually considering the number of variables, the number of dimensions, whether the variables are discrete or continuous, whether the distribution is unimodal, bimodal or multimodal, and the nature of the modes – such as whether the modes are in the middle or at the extremes.
We shall keep in mind our focus on the thesis that the world is dividing into two spheres, liberal and illiberal. Strong, extreme and distinct bimodality will be taken as evidence supporting the thesis of the two spheres. A mode is associated with a central cluster of countries and the nature of these countries will indicate whether the mode/cluster is ‘liberal’ or otherwise.
Briefly the conclusion is that UN voting about Russia is unimodal and not distinctively ‘liberal’; the opinion distributions are variously unimodal, bimodal and multimodal and in some cases the mode/cluster is ‘liberal’, but not in all.
6 UN voting about Russia
The Report looks at an index of aggregated UN votes on Russia over the period 2014-2022 whereas I have restricted my attention to just four voting occasions, one in 2014 and three in 2022.
It looks as if the distribution of the index of aggregated votes may be unimodal with the mode in the middle and with two tails, one tail consistently against Russia and the other tail consistently for Russia. The distribution of votes on the four occasions I have selected are unimodal also - but this time the mode is the ‘extreme’ option of voting against Russia (and it is a strong mode).
One figure below illustrates ‘one middle mode’ and another figure illustrates ‘one extreme mode’.
None of the distributions present a unique non-arbitrary natural divide into two groups. A symmetry argument is used to suggest an alternative to the two-group view.
One specific definition of ‘liberal’ specifies 40 ‘liberal’ countries (see earlier section). There appears to be a positive correlation between membership of the liberal group and being against Russia. However, the correlation is not perfect and non-liberal countries are also against Russia.
UN voting on Russia, 2014 to 2022
[Figure 1, 3/6]
The report aggregates UN votes on Russia over time, giving scores on a continuous scale. Figure 1 uses these to produce an interesting world map in which colour shading shows the percentage of support for Russia. What the map shows is not a divide but a distribution. It looks like this distribution is not polarised at two extremes but rather mostly “in-between”, possibly even a unimodal distribution.
“Figure 1. Index of country votes to condemn Russia in the United Nations, from 2014–2022 inclusive. The pattern of diplomatic activity maps closely to the distribution of global public sentiment towards Russia. Western countries have maintained a consistent demand to sanction Russian aggression, while continental Asian countries have been opposed, and the rest of the world in-between. Key votes include the 2014 UN resolution on the territorial integrity of Ukraine, the 2022 UN resolution on aggression against Ukraine, and 2016-21 votes on human rights in Crimea and militarisation in Crimea and the Black Sea.”
[I would add that some “continental Asian countries” are also in-between.]
The statement “the pattern of diplomatic activity maps closely to the distribution of global public sentiment towards Russia” is considered below.
Figure 1 One middle mode
. x
. xxxxxx
. xxxxxxxxxx
UN voting on Russia, four occasions, 2014 and 2022
The vote against Russia has increased: there were more UN votes against Russia in 2022 than there were in 2014.
Back in 2014 the UN General Assembly voted 100 (52%) against Russia.
On February 25th 2022, UN Security Council voted 11 (73%) against Russia.
On March 2nd 2022, the UN General Assembly voted 141 (73%) against Russia.
On October 12th 2022, the UN General Assembly voted 143 (74%) against Russia.
2014, UN General Assembly
Back in 2014 the UN General Assembly voted 100 (52%) against Russia over Crimea, 58 abstaining (30%), 11 for Russia (6%) and 24 absent (12%). (N=193).
There are three discrete options for the voting (a fourth option is being absent). The distribution is unimodal and the mode is at one extreme (voting against Russia). If we define ‘liberal’ to be the 40 countries listed in an earlier section*, then at least 60 non-‘liberal’ countries voted against Russia.
*[High-income democracies (World Bank 2022 high income threshold): They are 40 in number.]
February 25th 2022, UN Security Council
Of the five permanent members of the UN Security Council, three voted against Russia – USA, UK, France - (60%); one abstained – China (20%); and one used their veto – Russia (20%).
Of the ten current non-permanent members of the UN Security Council, eight voted against Russia – Brazil, Mexico, Gabon, Ghana, Kenya, Albania, Norway and Ireland - (80%); and two abstained – India, UAE (20%).
A combined 11 out of 15 voted against Russia, 73%; three abstained, 20%; and one voted against, 7%.
There are three discrete options for the voting. The distribution is unimodal and the mode is at one extreme (voting against Russia). It is a very strong mode: 60%, 90% and 73% combined. Of the fifteen, five were ‘liberal’ on that they were in the n countries listed in an earlier section. Looking now only at the ten non-‘liberal’ countries 6 voted against Russia, 3 abstained and 1 voted against.
Crudely, in terms of continents, Europe, North America, Latin America and Africa voted against Russia - but Asia and the Middle East abstained.
March 2nd 2022, UN General Assembly
The UN General Assembly overwhelmingly adopted a resolution on Wednesday demanding that Russian Federation Immediately End Illegal Use of Force in Ukraine, Withdraw All Troops. It voted 141 (73%) against Russia (i.e. for the motion), 35 abstaining (18%), 5 for Russia (3%) and 12 absent (6%). (N=193).
The distribution is unimodal and the mode is at one extreme (voting against Russia). It is a very strong mode. If we define ‘liberal’ to be the 40 countries listed in an earlier section*, then at least 101 non-‘liberal’ countries voted against Russia.
Almost all of Europe, America and non-mainland Asia voted against Russia. Africa was half against Russia with half abstaining. Almost all of mainland Asia abstained. Russia, Belarus, N Korea, Syria and Eritrea supported Russia. The Times, 3.3.22, 9.
https://www.un.org/press/en/2022/ga12407.doc.htm
…
October 12th 2022, UN General Assembly
The UN General Assembly voted 143 (74%) against Russia over Crimea, 35 abstaining (18%), 5 for Russia (3%) and 10 absent (5%). (N=193).
“The UN General Assembly passed a resolution by a large majority on Wednesday, calling on countries not to recognise the four regions of Ukraine which Russia has claimed, following so-called referendums held late last month, and demanding that Moscow reverse course on its "attempted illegal annexation".
The distribution is unimodal and the mode is at one extreme (voting against Russia). It is a very strong mode. If we define ‘liberal’ to be the 40 countries listed in an earlier section, then at least 103 non-‘liberal’ countries voted against Russia.
The countries who voted against were Belarus, the Democratic People's Republic of Korea, Nicaragua, Russia and Syria.”
https://news.un.org/en/story/2022/10/1129492
Figure 2 One extreme mode
. x
. x
. x x
. x x x
Divided in two? - a symmetry argument
It might be argued as follows. The distribution of the index of votes against Russia has a tail of countries that are against Russia. The presence of this tail shows that the world is divided into two groups, A and B.
A problem with this argument is how to specify the tail. The distribution is continuous. There is no unique non-arbitrary natural dividing point to specify the tail and the rest of the distribution.
However suppose we were allowed to choose an arbitrary dividing point X. Then we could say the world is divided into two groups, A and B – taking arbitrary dividing point X … [1].
A problem with this step is that we can appeal to symmetry and make a corresponding statement about the other tail: the world is divided into two groups, C and D – taking arbitrary dividing point Y … [2].
It follows that the world is not divided uniquely into two groups.
Rather than running with both statements [1] and [2], an alternative formulation would be the world is divided into three groups, A, M and C – taking arbitrary dividing points X and Y.
Note that M is the middle and A and C are the two tails (arbitrary, as is M).
In this way we have moved from the view that the world is divided into two groups to the view that the world is divided into three groups.
The notion of a world divided into an odd number of groups is discussed later.
Scores on the underlying distribution
In an earlier section I discussed “my underlying distribution approach to the analysis of data”. We now apply this approach to the UN voting.
The net percentage has moved from -46% in 2014 to -71% in October 2022, thus becoming increasingly negative towards Russia.
Let us score the scale: +1 for being for Russia, 0 for abstaining and -1 for being against Russia (and 0 for absent). Then the percentages become mean scores: -0.46 in 2014 and -0.71 in October 2022.
In my underlying distribution approach I assume there is an underlying continuous scale [-1,+1] representing different degrees of being for or against Russia. I assume countries are distributed along this continuous scale. Lacking more accurate information I assume that the mean score for ‘for Russia’ countries is +0.5 (halfway on the positive side) and that the mean score for ‘against Russia’ countries is -0.5 (halfway on the negative side). With these assumptions the underlying mean scores are -0.23 in 2014 and -0.36 in October 2022. [In other words half the amount of the ‘raw’ mean scores given earlier.]
UN votes with population weighting
UN voting is on the basis of one country, one vote.
(An interesting aside is that the Soviet Union arranged that Ukraine and Belarus were separate members of the UN with a separate UN vote in 1945).
https://en.wikipedia.org/wiki/Ukraine_and_the_United_Nations ;
How would the voting be affected if it was weighted by the population of the country? For example the Security Council vote for abstention would increase greatly because China and India both abstained.
The issue of population weighting is relevant in the next section.
7 Opinion about Russia
Much of the report involves a comparison of Russia, China and America. However, there are sections in the report which look at Russia only. These sections contain the following figures:
F1 3/6 percent UN votes against Russia, 2014-2022; world map of countries
F2 4/7 positive view of Russia; world map of countries
F3 5/8 change in positive view of Russia 2021-2022
F10 12/15 distribution of positive view of Russia, 2012 and 2022
F13 15/18 positive view of Russia, 2012-2022; by region
F14 16/19 positive view of Russia; by region, particularly Africa
F22 24/27 US opinion of Russia and China, 1952-2022
A variety of distributions for votes and opinions about Russia
The previous section on the UN has described four occasions at the UN when the voting distribution had one extreme mode, with the mode being against Russia. Moreover, the vote against Russia increased: there were more countries voting against Russia in the 2022 vote than there were in the 2014 vote. How does that relate to the information in the report? …
… How does the UN vote distribution relate to the two maps in Figures 1 and 2 discussed below? One can look at each map in two ways: one can think about the area for each colour (noting that area is not the same as population); or one can think about the number of countries for each colour. Thinking about area, Russia and China form a large area. Thinking about the number of countries, what one sees is a full spectrum of colour. It looks to me like possibly a uniform distribution (?) – a fairly even spread - over the full range of the spectrum. This is somewhat different from the reference to “polarisation” in the report’s comment about Figure 2, noted below. None of these distributions are similar to a UN vote distribution with a single strong ‘extreme’ mode against Russia.
… How does the UN vote distribution relate to public opinion about Russia in different countries as displayed in Figure 10? Figure 10 in the report shows the distribution of global public opinion towards Russia, in 2012 and in 2022. Note that the bottom part of Figure 10 presents the same data as presented in the map in Figure 2 – namely the 2022 data. Just as Figure 2 looked like a fairly even spread so Figure 10 also looks like a fairly even spread. This distribution is different from a UN vote distribution which has a single strong ‘extreme’ mode against Russia.
But that is me looking at Figure 10 in terms of number of countries. (To see the points for individual countries better, I got the computer to magnify the Figure.)
Instead one can look at Figure 10 in terms of the populations of countries. Looking at it in that way, it is bimodal with one large mode dominated by China and India and a second smaller mode dominated by USA.
One might feel that here at last one has found an empirical basis for the notion that the world is divided in two. However there is a band of countries between the two modes, so one needs to define a dividing point in order to define the two groups of countries. Also we shall find a similar bimodal distribution when we look at opinions about the USA but this bimodal distribution gives rise to two groups that are different from the two groups that area defined in terms of the Russian bimodal distribution. This prompts consideration of a two dimensional distribution based on Figure which we shall discuss later and which suggests three groups. And a three-dimensional perspective suggests more groups. Finally it has to be said that none of these suggestions lead us back to a UN vote distribution which has a single strong ‘extreme’ mode against Russia.
Two maps: UN votes against Russia 2014-2022; opinion of Russia in 2022
The following are the captions in the report for Figures 1 and 2.
“Figure 1: Index of country votes to condemn Russia in the United Nations, from 2014–2022 inclusive. The pattern of diplomatic activity maps closely to the distribution of global public sentiment towards Russia. Western countries have maintained a consistent demand to sanction Russian aggression, while continental Asian countries have been opposed, and the rest of the world in-between. Key votes include the 2014 UN resolution on the territorial integrity of Ukraine, the 2022 UN resolution on aggression against Ukraine, and 2016-21 votes on human rights in Crimea and militarisation in Crimea and the Black Sea.”
“Figure 2: Positive view of Russia, 2022 (or most recent survey, %). Over the course of the last decade, global public opinion towards Russia has polarised, with large majorities of the public in high-income democracies holding negative views, while pro-Russian sentiment persists across continental Asia, the Middle East and Africa.”
World public opinion about Russia, 2012 and 2022
[Figure 10, 12/15]
What follows is further comment on Figure 10.
“Figure 10: The spectrum of global public opinion towards Russia, from 2012 (top) to 2022 (bottom). Over the course of the past decade, Russia’s reputation improved across countries of the Global South, notably in populous nations such as India, China, and Indonesia. At the same time, almost all high-income democracies now view Russia negatively – and to an exceptional degree. Data points for Russia itself are excluded from display.”
The distribution was pretty much unimodal in 2012 – one middle mode. The distribution then stretched and had become bimodal in 2022 - there was a large positive mode and a smaller negative mode. Liberal countries moved negative and India and China moved more positive.
Looking back at Figure 10, I wonder how it should be interpreted. There are individual countries marked along the scale; and there is a continuous line showing the distribution. It is the continuous line that indicates the two modes, large and small. Looking at the positions of the individual countries, I see a fairly even spread across the scale with several clumps of countries (four or five clumps) and perhaps the extreme clump against Russia is larger than the extreme clump for Russia.
What is happening here? I wonder if the continuous line is the population-weighted percentages? I wonder if it is a smoothing of the data and whether a less smoothed line would have more modes? And what does the unweighted distribution look like?
8 Opinion trajectories
Having looked at the opinion of states at the UN we now turn to look at public opinion. It is useful to start our investigation of opinions by looking at how overall opinion has changed over time.
We start by looking at change over the long term, namely how people in the USA have viewed Russia and China over the past seventy years. The past seven decades have seen the initial easing of negativity, the long middling position between positive and negative opinion and the more recent descent into strong negativity.
We then look at the more recent period in detail (2012-22), looking not just at what the people in the USA think but at what all the countries think, and looking not just at opinion about Russia and about China but also about the USA. We also split the countries into two groups, developed countries and developing countries.
We then take a more detailed look at different groups of countries, still in the period 2012-2022. Developed countries can be split into three groups: European Union, Anglo-Saxon Democracies, and Pacific North Asia. Developing countries have five groups: South-east Asia, South Asia, Sub-Saharan Africa, Middle East and Latin America.
An overview of the trajectories can be obtained by looking only at the start year 2012 and at the end year 2022 (ignoring intermediate fluctuations), and by designating the points as either positive, P, or negative N. We then look at the triple of points for America, Russia and China (in that order). PPP means positive for all three countries.
PPP was the most commonly observed pattern. Pattern PNN was next most common. Also observed were PNP and NNP. Thus there were four different patterns in all.
Superpower opinion, 1952-2022
[Figure 22, 24/27]
In a conflict relation each party is often positive about the self and negative about the other. This is so in the conflict between the liberal USA (L) and the non-liberal Russia (non-L) and China (non-L).
In 1952, USA opinion of Russia was at rock bottom. It rose to near neutral in the early 1970s but then fell back into negative territory. It rose into slightly positive territory in 1988 and stayed at that level until 2012 but since then has steadily declined back to rock bottom now in 2022.
Since 1984, USA opinion of China was unchangingly neutral up to 2018 but in the last four years has declined very sharply. (There was a solitary high positive in 1989.)
Since 1984 the two opinion trajectories (of Russia and of China) have been fairly similar.
In 2022 the antagonism of seventy years ago has returned.
Opinion in developed and developing countries, 2012-2022
[Figure 7, 10/13]
The word ‘developed’ is an L-word and refers to the set of developed countries listed in an earlier section. They are 37 in number.
Figure 7 has three separate graphs. An overview of the three graphs can be obtained by looking only at the start year and at the end year of the period, and by designating the points as either positive, P, or negative N. We then look at the triple of points for America, Russia and China. PPP means positive for all three countries.
The situation at the start year is the same as the situation at the end year. The situation for all countries combined is the positive triple PPP and this is the situation also for all developing countries combined. The situation for all developed countries combined is the triple PNN, positive towards the USA but negative towards Russia and China:
The opinions of all countries about all three countries (A, R and C) were PPP
in 2012 and PPP in 2022.
The opinions of developed countries about all three countries (A, R and C) were PNN
in 2012 and PNN in 2022.
The opinions of developing countries about all three countries (A, R and C) were PPP
in 2012 and PPP in 2022.
Let us look now at what happened between the start year and the end year. China was seen as negative in the years 2020 and 2021 – by all countries combined – giving PPN in those two years. All developing countries combined were at PPP throughout.
All developed countries combined were at PNN throughout.
Looking beyond the simple classification as P and N, the opinions of developing countries about R, C and A were higher than were the opinions of developed countries about R, C and A – and did not change all that much. In contrast the opinions of developed countries about R and C declined dramatically over the period while their opinion about A declined and then recovered.
Opinion in geographical groups
[Figure 13, 15/18] [Figure 14, 16/19] [Figure 15, 17/20; world map Figure 16, 18/21]
[Figure 17, 19/22; world map Figure 18, 20/23] [Figure 19, 21/24]
We now take a more detailed look at different groups of countries. Developed countries have three groups: European Union, Anglo-Saxon Democracies, and Pacific North Asia. Developing countries have five groups: South-east Asia, South Asia, Sub-Saharan Africa, Middle East and Latin America.
South-east Asia, South Asia and Sub-Saharan Africa stay at the PPP pattern.
The Middle East moves from NNP in 2012 to PPP in 2022.
Latin America stays at PNP (negative towards Russia).
Anglo-Saxon Democracies move from PNP in 2012 to PNN in 2022.
European Union and Pacific North Asia stay at PNN.
Table
.
. 2012 2022
.
South-east Asia PPP PPP
Sub-Saharan Africa PPP PPP
South Asia PPP PPP
Middle East NNP PPP
Latin America PNP PNP
Anglo-Saxon Democracies PNP PNN
Pacific North Asia PNN PNN
European Union PNN PNN
.
What follows consists only of brief notes about each figure.
Russia
[Figure 13, 15/18]
“Russia offers perhaps the clearest case where opinions in the developed and developing worlds have separated (Figure 13).” 2012-2022
South Asia and Southeast Asia: improved markedly
Sub-Saharan Africa: positive and unchanging
USA? (“Anglo-Saxon Democracies”): neutral then deteriorating markedly
Middle East: from negative to neutral
Latin America: from negative to neutral
European Union: fluctuating but no trend over the decade
Pacific North Asia: fluctuating but no trend over the decade
In summary the divergence is marked between South Asia and the USA: both starting slightly positive then ending at the opposite extremes
Russia continued
[Figure 14, 16/19]
Russia still remains popular across much of Africa today, especially in countries around the Sahel region – such as Francophone West Africa and the Arabic-speaking countries to its north.
China
[Figure 15, 17/20; world map Figure 16, 18/21]
“When we look at regional opinion trends regarding China, we find a similar global divergence.”
Former Soviet Union: high positive and unchanging
Sub-Saharan Africa: high positive and unchanging
Latin America: moderate positive and unchanging
Southeast Asia: moderate positive turning neutral
Middle East: neutral turning moderate positive
South Asia: neutral then moderate negative then neutral
Anglo-Saxon Democracies: neutral (2012-2017) then marked fall (2017-2022)
European Union: slightly negative (2012-2017) then fall (2017-2022)
Pacific North Asia: negative (2012-2017) then smaller fall (2017-2022)
USA
[Figure 17, 19/22; world map Figure 18, 20/23]
“Finally, the United States enjoys high levels of support across most regions of the world.”
Pacific North Asia: most positive at start and remains high throughout, while declining through the period until a slight increase in the recent post-Trump years
Several - Sub-Saharan Africa, Anglo-Saxon Democracies, South Asia, South-East Asia: remaining in the high positive band although showing some of the trends noticed elsewhere, albeit in a mild form.
Latin America: positive then declining to neutral during 2016-2020 then increasing
EU: positive then declining to neutral during 2016-2020 then increasing
Former Soviet Union: positive then declining and neutral from 2016
China: neutral (2012-2017) then decreasing markedly (2017-2022)
Middle East: low then rising to neutral
USA continued
[Figure 19, 21/24]
While much of the Global South remains neutral between the United States and China or Russia, Latin America stands out as the major exception – with relative positive feelings towards the region’s northern neighbour reaching record highs. Population-weighted trendlines for all regional groups.
9 Opinion distributions
The previous section looked at the trajectories for a number of percentages. In some cases, the percentage was the aggregate of the percentage scores for different countries. In this section we look at the unaggregated percentages scores for different countries and their distribution. The focus is on distributions in 2022.
Maps give a useful initial general impression of the geographical distribution of opinion. There are maps for countries’ opinions about Russia, China and the USA.
Three basic variables are involved, namely countries’ opinions about Russia, about China and about the USA. Each country has three percentages which can be represented by a point in three-dimensional space. We are interested in the distribution of points in this three-dimensional space.
The Figures help us see this three-dimensional space but can do so only in a limited way because they are inherently two-dimensional. So what we do is piece together a variety of two dimensional figures to gain an understanding of the three-dimensional space.
In a number of places the report makes use of a combined measure, combining the opinion about Russia and the opinion about China. The rationale for doing so is that these two variables are strongly positively correlated and this is shown in Figure 8, 11/14. To the extant that there is a perfect correlation this use of a combined variable reduces the three-dimensional distribution to a two-dimensional distribution.
Figure 5, 8/11 presents this distribution. It is the graph which to my mind most directly addresses the notion that the world is divided into two spheres. It is a graph of opinion about USA and opinion about Russia and China combined. I see the figure as having three clusters not two.
We then present a reminder. ‘P’ denotes positive and ‘N’ denotes negative. There are four quadrants in two dimensions: NN, PN, NP and PP. There are eight octants in three dimensions: NNN, PNN, NPN and PPN; and NNP, PNP, NPP and PPP. The first letter refers to USA (U); the second letter refers to Russia (R); and the third letter refers to China (C).
From the preceding work we note some of the countries which are in the eight octants. Most are in PPP. ‘Liberal’ countries are in PNN. Russia and China are in NPP. Some countries are specifically negative about Russia: Russia: PNP. Some countries are specifically negative about China: PPN. This seems to go beyond the notion that the world is divided into just two spheres.
Maps: the geographical distribution of opinion, 2022
[Figure 1, 3/6, UN votes against Russia; Figure 2, 4/7, Russia; Figure 16, 18/21, China; Figure18, 20/23, USA; and Figure 6, 9/12 difference between US and China]
Corresponding to the map for the UN voting on Russia, there are four maps about worldwide opinion, based on public opinion data. They concern opinion about Russia, opinion about China and opinion about USA. There is also a map about the difference between the opinion about China and the opinion about USA.
The broad impression of the maps is similar – but not identical - to the impression given by the map for UN voting on Russia.
As an aside I feel it is not easy to compare maps. So I have reservations with the report’s comment that “… if we look at a map of how different peoples around the globe feel towards Russia, we discover an almost identical reflection of how their governments have handled the country diplomatically since 2014 (Figure 2).”
My conclusion above for the UN voting map was:
“What the map shows is not a divide but a distribution. It looks like this distribution is not polarised at two extremes but rather mostly “in-between”, possibly even a unimodal distribution.”
My conclusion is similar for these other maps They also seem to me to show a distribution not a divide. Again it looks like the distribution is not polarised at two extremes. Whether the distribution is strictly unimodal I find difficult to judge from the maps.
The range for the single country maps is 10% to 90% having a positive favourable opinion. The map which shows the difference between the USA and China positive percentages has a range somewhat wider than the range -25% to +25%. The USA-China map looks like having a unimodal distribution, possibly with a mean zero percentage.
The maps give a useful initial general impression. This is now supplemented by a look at the Figures.
Two-dimensional Russia and China opinion space: PP, NN and NP
[Figure 8, 11/14]
“… by and large, countries with a positive view of China, tend to feel positively about Russia, and vice-versa.”
Figure 8: Across the world, there is now a high correlation between how societies view Russia and how they view China. This correlation has increased over time, and is now higher than ever before.”
A cooperative relation between Russia and China is mutually positive PP. Joint allies will also be PP. Joint enemies will be NN. A China preference over Russia gives NP. Hence PP, NN and NP.
My earlier account of this graph:
Figure 8 gives a graph of scores for China versus scores for Russia (positive opinion percentages). Looking at the scatter of points it is hard to say but the Russia scores may be unimodal the China scores may be unimodal. Or there may be a second mode for lower values – giving a bimodal distribution.
The scatter of the points gives something of the appearance of an ellipse. The major axis might be roughly y=x (the line in the graph is possibly the regression line); and the minor axis might be roughly y=c-x.
The major axis runs from USA NN (x and y low) up to Russia PP (x and y high). The minor axis runs from Moldova NP (x low and y high) down to PN Vietnam (x high and y lowish). Ukraine is quite near Moldova on the graph; NP. Looking at populous countries Nigeria is middling x and high y, NP, whereas India is highish x and middling y, PN.
There seems to be a quarter circle USA, Mexico and Russia; (0,0), (0.3,0.7) and (1,1). Alternatively curves might be straightened out using z-scores.
My guess is that the two rotated factors of a factor analysis would give a first factor of Russia-China sum and a second factor of Russia-China difference.
Two-dimensional USA and Russia-China opinion space: NP, PN and PP
[Figure 5, 8/11]
Third parties to a conflict may be positive to both the parties that are in conflict. So PP.
“Now consider a world where there is mostly cooperation but also a conflict between A and B. There will be points in three quadrants: most of the points will be in the cooperative PP quadrant; a few points will be in the PN quadrant for A and its allies; and a few points will be in the NP quadrant for B and its allies; and none will be in the NN quadrant.”
Each party in a conflict relation between USA and Russia/China is positive about the self and negative about the other. So NP and PN. Third parties to the conflict may be positive to both. So PP.
This description applies to Figure 5. Most points are in PP, positive about both USA and Russia-China. Some points– NATO/EU and others – are in PN, positive about USA and negative about Russia-China. A few points – Russia, China, Laos and Serbia – are in NP, negative about USA and positive about Russia-China. Almost no points (just a few just across the border from NP or PN) – are in NN, negative about both USA and Russia-China.
Divided? Most, those in PP, are not divided. NATO/EU does have a secondary mode on the Russia-China opinion distribution. Note though that NATO/EU has a wide range of opinions about USA, and indeed a wide range of opinions about Russia-China.
Within PP, there does look like a modest correlation between positive USA and positive Russia-China …
A reminder: the four quadrants and the eight octants
There are four quadrants in two dimensions …
NP PP
NN PN
… and eight octants in three dimensions
NPP PPP
NNP PNP
NPN PPN
NNN PNN
Three-dimensional USA, Russia and China opinion space
When we looked at two-dimensional USA and Russia-China space we identified the three quadrants NP, PN and PP as being significantly occupied. The only occupants of NN were Palestine and Iran. [Figure 5, 8/11]
What happens to these four quadrants when we separate out the Russia and China scores? Each quadrant becomes two, giving eight octants. We can infer what is in these octants by looking at the graph for Russia and China. [Figure 8, 11/14]
Ukraine, Georgia and Moldova are positive about China but negative about Russia. The three are also positive about USA. So they are in the octant PNP (USA, Russia, China).
Vietnam, Indonesia and India are positive about Russia but (somewhat) negative about China. Vietnam and India are positive about USA, so PPN; but Indonesia is also negative (somewhat) about USA, so NPN (USA, Russia, China).
Noting the remarks about Palestine and Iran earlier, the report does not
provide the information about them to decide whether they are NNN, NNP or NPN.
We now return to an earlier section where we considered three scores x, y and z – about USA, Russia and China (U, R and C) – given by different countries. The three scores for U, R and C give a point in three-dimensional opinion space. Just as two-dimensional space has four quadrants so three-dimensional space has eight octants. The figure below displays the eight octants and some of the countries which are located in them. In summary.
URC. Most of the countries are in PPP.
RC. Russia and China are in NPP – as is Serbia, and also possibly Laos?, Brunei?, and
Iraq? – so only one or a few countries share the same octant as Russia and China.
UC. Ukraine, Georgia and Moldova are in PNP.
UR. Vietnam and India are in PPN.
C. any? NNP.
R. Indonesia is in NPN.
U. Many liberal countries are with USA in PNN.
? Iran, Palestine are in NNN?
[The report does not give the full set of tables - and not all points on the graphs are labelled.]
Figure The eight octants of the three-dimensional URC space
China z>50
| y :
| NPP : PPP
| RC : URC
| China :
| Russia : MOST
| :
| :
| NNP : PNP
| C : UC
| ? :
| : Ukraine et al
|. : x
0 50 100
China z<50
| y :
| NPN : PPN
| R : UR
| :
| Indons : Vietnam et al
| :
| :
| NNN : PNN
| - : U
| Palestn : many
| Iran : USA
|. : x
0 50 100
The one-dimensional distributions
The one-dimensional distributions are projections of the three-dimensional distribution onto each of the three dimensions. The populous PPP octant will give a populous positive mode on each dimension. The PNN octant will give a smaller negative mode on the Russia and on the China dimension. The NPP octant will give an even smaller negative mode on the USA dimension (albeit a more populated one because of China). This is indeed what is found in the distributions for 2022 given in [Figures 10, 12/15; 11, 13/16; 12, 14/17].
10 Changing opinion
An earlier section looked at opinion trajectories and here we continue looking at opinion change.
First we look at the changing opinion about Russia between 2021 and 2022 in the aftermath of its invasion of Ukraine.
Next, if opinion (such as USA’s) about Russia goes negative then we might expect Russia’s opinion about others (such as USA) to reciprocate and also go negative. This in turn suggests that the correlation between the opinion about Russia and the opinion about others (such as USA) might go negative. This is explored by looking at the correlation between opinions about Russia-China combined and opinions about the USA over the period 2008-2022. We also imagine what the correlations might look like when Russia and China are treated separately.
Finally, we turn to look at the distributions. We compare the distributions in 2012 and the distributions in 2022, for USA, Russia and China. In 2012 each of the three one-dimensional distributions was pretty much unimodal. In contrast each of the three one-dimensional distributions in 2022 was bimodal or nearly so. The one-dimensional distributions are not sufficient to deduce the three-dimensional distribution. It may be a trimodal distribution such as that discussed above in relation to Figure 5, 8/11.
The change in opinion about Russia, 2021-2022
[Figure 3, 5/8 … Figure 10, 12/15]
Countries’ opinion about Russia became more negative in 2022 but the size of a country’s change does not appear to depend on the country’s initial opinion in 2021.
Note that the vertical axis in Figure 3 puts countries in order of the 2022 opinion and the horizontal axis gives the amounts of the 2021 and 2022 opinions and the associated change. I found this presentation format quite awkward to think about.
“Figure 3: Change in public favourability towards Russia, 2021–22. Across all countries, a clear majority of the public views Russia negatively (?*), and this proportion has increased in 2022. Nonetheless, attitudes remain positive across continental Asia, including China, South Asia, Southeast Asia, and the Middle East.”
The underlined statement is puzzling since Figure 10 shows the taller mode at a positive 75% in 2022.
The changing correlation between USA and Russia-China opinions, 2008-2022
[Figure 9, 11/14]
Are countries’ opinions about the USA correlated with their opinions about Russia-China combined? How has the correlation changed over the period 2008-2022?
For most of the period the correlation was almost zero - the correlations ranged between -0.1 and +0.1. However, in the last five years or so the correlation has become steadily more negative and now stands at roughly -0.4 (in other words modestly negative).
What the correlation of almost zero in 2012 involved is suggested by the one-dimensional distributions discussed in the following section.
What the correlation of -0.4 in 2022 looks like on a graph can be seen in Figure 5, 8/11, which was discussed earlier. There we noted that most points were in the PP quadrant. There were also some points in the NP and PN quadrants and it was these points that gave rise to the negative correlation. The reason the correlation in 2022 is only modest is because most points are in the PP quadrant.
Let us remind ourselves that the correlation which we are discussing is that between the USA and China-America combined. What of the correlation matrix for the three separate countries? The work of earlier sections suggests the correlation matrix for 2022 might be …
. USA Russia China
USA 1 -0.4? -0.4?
Russia 1 high, 0.9?
China 1
… and for 2012, ten years previously, the correlation matrix might have been:
. USA Russia China
USA 1 0? 0?
Russia 1 middling, 0.5???
China 1
The distribution of opinion, 2012 and 2022; Russia, China and America
[Figures 10, 12/15; 11, 13/16; 12, 14/17]
How have the distributions of opinion changed over the past ten years? A comparison is made between the distribution in 2012 and the distribution in 2022. There are three separate comparisons, one for each of Russia, China and America.
Russia. Pretty much unimodal in 2012 … the distribution then stretched and had become bimodal in 2022. In 2022 there was a large positive mode and a smaller negative mode. Liberal countries moved negative and India and China moved more positive.
USA. Pretty much unimodal in 2012 … the distribution then stretched and had become bimodal in 2022. In 2022 there was a large positive mode and a smaller negative mode. India and China moved more negative and Liberal countries moved more positive.
China. Pretty much unimodal in 2012 … the distribution then stretched and was marginally bimodal in 2022. In 2022 there was a large positive mode and a marginal smaller negative mode. Liberal countries moved negative and India moved more positive.
Thus each of the three one-dimensional distributions in 2012 was pretty much unimodal. This is not sufficient to deduce what the three-dimensional distribution in 2012 looked like. One possibility however is that the three-dimensional distribution was also unimodal.
Each of the three one-dimensional distributions in 2022 was bimodal or nearly so. This is not sufficient to deduce what the three-dimensional distribution in 2022 looked like. In particular it might not be bimodal. The two smaller negative modes contain different sets of countries. This suggests a trimodal distribution as is indeed exhibited in Figure 5, 8/11, which was discussed in Two-dimensional USA and Russia-China opinion space: NP, PN and PP above.
Finally, the stretching and bimodality refer to the continuous distribution. It does not constitute a categorical divide.
11 Other countries’ opinions; and country attributes … ‘liberal’ and ‘divide’
Most of the figures in the report are concerned with the opinions which countries have about the USA, Russia and China. However a few of the figures concern attributes of countries and relate the attributes of the countries to the countries’ opinions about the USA, Russia and China.
The first of these attributes is economic, namely China’s investment commitments. The other three attributes concern social and political values and relate to the notion of ‘liberal’, the central notion in the report. The three attributes are: liberal democracy, satisfactory democracy and social liberalism.
As in earlier sections we shall consider positive P; and negative N. Positive attribute and positive opinion PP.
We shall also consider the nature of the distribution in each figure. The Q2 distribution has points in two diagonally opposite quadrants; and none in the other two quadrants. The Q3 distribution has points in three quadrants; but none in the fourth quadrant. … or approximately so.
Opinion about China and China investment commitments
[Figure 23, 25/28]
Consider the “Average Annual Chinese (Belt and Road) Commitments, as Share of GDP (%)”. This is a positive action by China towards another country. One might expect that the country would reciprocate by having a favourable opinion of China. Or, with causation in the opposite direction, if a country is positive towards China then it might be expected to obtain a commitment from China. A positive attribute and a positive opinion – and so PP.
Also, the greater the commitment the more positive the response.
Countries not receiving this positive action might be expected to have a negative opinion of China. Or, with causation in the opposite direction, if a country is negative towards China then it might be expected not to obtain a commitment from China. A negative attribute and a negative opinion – and so NN.
The figure suggests a modification to this line of reasoning. The key distinction is between zero Z and non-zero investment X. With this amendment it is indeed the case that most countries are in the either the XP or ZN quadrants (more in XP) and few in either ZP or XN. In other words the graph exhibits a two-quadrant Q2 distribution. Within XP, the correlation looks moderate at best. Many liberal L countries are in ZN.
Liberal democracy and the opinion about Russia and China combined
[Figure 20, 22/25]
Consider the relationship between liberal democracy and the opinion about Russia and China combined. This explicitly relates to the notion that there is a divide between liberal and illiberal countries. Russia and China are considered illiberal countries, under most of the definitions discussed earlier. Liberal countries will have high scores on the Liberal Democracy Index and will be negative about Russia/China. So PN.
Illiberal countries will have low scores on the Liberal Democracy Index and will be positive about Russia/China. So NP. There will be few countries if any in NN or PP. In other words the graph will exhibit a two-quadrant Q2 distribution.
The graph does indeed exhibit a Q2 distribution. Taking into account the note in the following paragraph, the two quadrants are PN and NP as described in the previous paragraph. Few countries are in NN or PP.
Note that in the report the Russia-China opinion scale is given the other way round, but this does not affect the overall argument. This makes it tricky to relate the figure to my argument – the second label is the wrong way round - but my argument nevertheless corresponds to what is in the figure.
However there is variation within each quadrant; and in the case of the NP quadrant occupies the entire quadrant, with what I imagine might be a zero correlation between the liberal democracy index and the opinion about Russia/China.
An additional problem arises because of different bases for categorising liberal democracies. The figure splits the x-axis into three intervals labelled “Illiberal Democracies / Authoritarian regimes”; “Fragile Democracies”; and “Liberal Democracies”. In the y-axis, “Fragile Democracies” are in the same block as
“Illiberal Democracies / Authoritarian regimes”.
“… a closer look at Figure 20 reveals a number of electoral democracies, such as Indonesia, India or Nigeria, in which the public remains sympathetic to Russian or Chinese influence, in spite of a difference in political regime.” Report 23/26.
The specification of three groups on the liberal democracy scale in Figure 20 would seem to sit uneasily with the observation that the world is divided into two groups, liberal and illiberal.
Satisfaction with democracy; and the opinion about Russia and China combined
[Figure 21, 23/26]
Consider the relationship between the opinion, “satisfaction with the functioning of democracy” and the opinion about Russia and China combined. Here the report considers just a set of democratic countries). Russia and China are considered illiberal countries, under most of the definitions discussed earlier – not only illiberal but also non-democratic. Liberal countries will have high scores on satisfaction with democracy and will have negative opinions about Russia/China. So PN. Illiberal countries will have low scores on satisfaction with democracy and will have positive opinions about Russia/China. So NP. There will be few countries if any in NN or PP. In other words the graph will exhibit a two-quadrant Q2 distribution.
Note that here too in the report a scale is given the other way round – the x-axis is dissatisfaction, but this does not affect the overall argument. This makes it tricky to relate the figure to my argument – the second label is the wrong way round - but my argument nevertheless corresponds to what is in the figure.
In fact the graph exhibits a Q3 distribution. Certainly the two quadrants PN and NP as described in the previous paragraph appear. And there are almost no points in the NN quadrant, an isolated case being Kosova. However there are points in the PP quadrant, satisfaction with democracy and positive about Russia/China – such as India and Tanzania.
Note that here too in the report a scale is given the other way round – the x-axis is dissatisfaction, but this does not affect the overall argument. This makes it tricky to relate the figure to my argument – the second label is the wrong way round - but my argument nevertheless corresponds to what is in the figure.
An additional point is that there is variation within each quadrant; and the points occupy much of each quadrant.
Social liberalism and the opinion about Russia and China combined
[Figure 25, 27/30]
Consider the relationship between social liberalism and the opinion about Russia and China combined. Russia and China are considered illiberal countries, under most of the definitions discussed earlier. Liberal countries will have high scores on social liberalism and will have negative opinions about Russia/China. So PN. Illiberal countries will have low scores on social liberalism and will have positive opinions about Russia/China. So NP. There will be few countries if any in NN or PP. In other words the graph will exhibit a two-quadrant Q2 distribution.
The graph does indeed exhibit a Q2 distribution. The two quadrants are PN and NP as described in the previous paragraph. Few countries are in NN or PP.
Note that in the report the Russia-China opinion scale is given the other way round, but this does not affect the overall argument. This makes it tricky to relate the figure to my argument – the second label is the wrong way round - but my argument nevertheless corresponds to what is in the figure.
However there is variation within each quadrant. In the case of the NP quadrant (low social liberalism and having a positive view of Russia-China) points occupy the entire quadrant. This means that a country with almost middling social liberalism can have a very high positive view of Russia-China. Indeed Russia and China have almost middling levels of social liberalism – and a very high positive opinion of themselves.
“At the same time, Chinese social attitudes are relatively progressive with respect to secularism, women’s rights, or sexual diversity, and China has never presented itself as a “bulwark” against western liberalism in the same way as Putin’s Russia.” Report 27/30.
“Figure 25: Across the world, social liberalism is now one of the strongest predictors of whether a society holds a positive – or negative – view of Russia and China. This association has developed only during the past decade, and was quite weak ten years ago (R = 0.64 today; R = 0.35 in the past). Index of social liberalism selects from items that reflect values of individualism, freedom of choice, support for democracy, and personal autonomy (see Roberto S. Foa, Yascha Mounk & Andrew J. Klassen (2022). “Why the Future Cannot be Predicted". Journal of Democracy, 33(1)). The figures for Russia and China here only report favourability towards the other country.” Report 27/30.
The quoted “one of the strongest predictors” correlations refer to the entire graph. However within the NP quadrant it looks like the correlation is zero. So overall piecewise regression is possibly the best.
An SLI of 42 is a key point. Above 42, y takes a high level; and below 42, y takes a low level (y is the variable, negative about Russia-China). More detailed information about the Social Liberalism Index does not yield better predictions. A very tentative thought comes to mind: it may be that there is a U-shaped relationship between x and y – the lowest values for y may come from middling values of x (x is the variable SLI):
Positive opinion about Russia-China has a maximum for middling levels of social liberalism.
Extreme ‘social illiberalism’ – as opposed to Russia, China and the West
[Figure 25, 27/30]
Russia, China and the West – none of them have a low score on the Social Liberalism Index – see above. So which countries do? Which have extreme ‘social illiberalism’? They are:
Libya, Tunisia, Jordan, Iraq, Azerbaijan, Egypt;
Turkey, Georgia, Iran, Saudi Arabia, Morocco, Armenia, Kyrgyzstan, Algeria,
Pakistan
Predominantly they are Muslim countries of the Middle East, a crescent cross running from Morocco in the west to Pakistan in the east; and from Saudi Arabia in the south north to Georgia and Kyrgyzstan in the north.
The report looks at world opinion about USA, Russia and China. What is world opinion about Pakistan, Saudi Arabia and Iran?
Change in social liberalism, 1990-2022
[Figure 24, 26/29]
The Social Liberalism Index in high-income democracies has increased throughout the period 1990-2022 from 45 to 53. In the Rest of the World it has stayed flat at just below 40.
Value depends on social distance … single-peaked value functions on multidimensional attribute space
[Figure 25, 27/30]
Figure 25 suggests a general theory of single-peaked value functions on multidimensional attribute space.
The value of an object to the self depends on the distance between the object and the self on some social dimension in some multi-dimensional attribute space.. In particular the value of Russia to the self depends on the distance between Russia and the self on the dimension of social liberalism.
This dependence on distance gives rise to single peaked functions of two kinds. Firstly the opinions which a country has about other countries are single-peaked. Secondly the opinions which all the countries have about one particular other country are single-peaked. In particular the opinions which all the countries have about Russia-China are single-peaked on the social liberalism dimension.
To see this we need to turn Figure 25 upside down so that high positive values for Russia-China are at the top. The peak occurs in the middle of the scale just below 40. Russia is at the peak and China close at hand. The value of Russia-China falls off as we move away from the peak on either side. On one side of the peak the value falls off as social liberalism increases (and this is consistent with the line in the figure).
On the other side of the peak the value falls off as social liberalism decreases. The countries with the lowest social liberalism have neutral views about Russia-China. (Note that this is not consistent with the line in the figure. The line I think is based on linear regression whereas a test for single-peaked functions needs something like curvilinear or piecewise regression).
12 Conclusion: A world distributed in multidimensional space
My latest thinking
In general one imagines a world with N interacting countries where each country has a positive/negative opinion score about itself and about every of the (N-1) other countries. On the basis of these scores, each country is located at a point in N-dimensional space. There is a distribution of countries in the multidimensional space. The scores for each country give a single-peaked value function for each function (I think?).
The Odd Number Theorem for a social system states that in any social system, if there are n two-sided independent conflicts, then the system is divided into (2n+1) groups. This gives a particular shape to the N-dimensional distribution. Possibly most countries have mildly positive or neutral opinions about all other countries. A pair of countries in conflict have mutually negative opinions. The opinions about these two opposing countries exhibit three groups as in:
F5 8/11 positive view of US; positive view of Russia and China combined
… likewise for other conflicts – thus giving rise to the (2n+1) groups. A “Figure 5” will exist for each of the other conflicts. For each conflict there will be a statement: “focusing solely on this conflict there is a “Figure 5” showing three groups in the corresponding two-dimensional space.”
Countries are also located in M-dimensional attribute space. Each opinion value function constitutes a single-peaked value function on attribute space for the associated country.
An earlier observation
If the world is divided, how many divisions are there in the divide? Looking at some of the Figures in the report what one sees is not a binary divide but a divide into more than two divisions. Some figures are time graphs showing different trajectories
Moreover looking at some of the Figures in the report what one sees is not a divide but a distribution on a continuum. Some of the figures are in the form of maps of the world with countries coloured in. The colours on a colour continuum relating to a continuum for the statistical indicator relating to that indicator.
The maps indicate a -dimensional space and countries form a distribution in multi-dimensional space. Looking at the maps by eye there does seem to be a correlation between the different dimensions. It would be interesting to know what these correlations are.
One of the maps provides a plot of countries in two-dimensional space. By eye it looks as if the correlation between the two dimensions is quite low.
Having noted that the data has a distribution, we ask what is the nature of the distribution. Is the distribution uniform? … or unimodal or bimodal – or multimodal? What is the amount of unimodality as opposed to multimodality?
Returning to the grouping of countries, what is the variation between countries within the groupings – and what is the variation between the groupings. An initial visual impression of this can be obtained by looking at the world maps and noting the variation in colour within each continent.
Appendix: An overview of the report “A World Divided”
The latest report from The Centre for the Future of Democracy is entitled A World Divided: Russia, China and the West.
The report, direct access: https://www.bennettinstitute.cam.ac.uk/wp-content/uploads/2022/10/A_World_Divided.pdf
Note: Each page of the report has two page numbers. The first number is the page number in the paper report. The second page number is the slide number in the pdf.
Contents of the report
1. Executive Summary 1/4
2. Key Findings 2/5
3. Introduction – A World Divided 3/6
4. The Data 6/9
5. The New Structure of Global Public Allegiances 8/11
6. Visualising a Decade of Rising Geopolitical Polarisation 12/15
7. Regional Divergence Over Time 15/18
8. Why is the World Dividing in Two? 22/25
9. Conclusion 28/31
Methodology I: Survey Source Items 32/35
Methodology II: Variable Selection and Validity 33/36
…
1. Executive Summary 1/4 [quotation from the report]
“In this report, we examine how worldwide attitudes towards the major international powers – China, Russia, and the United States – are shifting in the wake of the Ukraine war, China’s rising assertiveness, and recent challenges to American democracy.
We do so by harmonising and merging data from 30 global survey projects that collectively span 137 countries which represent 97% of world population. This includes 75 countries surveyed since the Russian invasion of Ukraine, giving us updated insights into the current views of 83% of all people across the globe …”
…
2. Key Findings 2/5 [quotation from the report]
“The world has divided into liberal and illiberal spheres. …
Perceived democratic shortcomings are associated with greater public receptivity towards authoritarian powers. …
China is now ahead in the developing world. …
However this boost in approval across the Global South has come at the cost of a dramatic collapse in support [for China] in developed nations. …
Russia too has lost its “fringe” support within western democracies. …
However, the real terrain of Russia’s international influence lies outside of the West. …”
3. Introduction – A World Divided 3/6 [quotation from the report]
“On February 22nd, 1946, the American chargé d’affaires in Moscow, George Kennan, sent an 8,000 word message to his superiors in Washington DC. This “long telegram,” as it was later known, warned of a fundamental difference in worldview between Soviet and American leaders. Rather than settle for “peaceful coexistence,” the Soviets would seek to expand their global influence, leading the United States to respond and countries to divide into rival competing blocs.” …
“A New Global Divide? In this report, we ask a simple question: in the wake of the war in Ukraine, is the world now experiencing a similar moment of great power division – and if so, where are different societies situated respective to these countries, and why have they divided as they have? We provide an answer by looking at public opinion data from across the world, from surveys asking respondents about their feelings towards geopolitical rivals, and use this to identify when, where and ultimately why this new global divide has emerged.”
Figure 1 World map. UN votes against Russia 2014-2022. Scale 10% to 90%.
Variation between and within continents. Russia and China for; NATO against. P 3/6.
Figure 2 World map. Positive view of Russia 2022. Scale 10% to 90%
Variation between (large) and within continents (small). P 4/7.
Figure 3 Change in positive view of Russia 2021-2022. Substantial falls. P5/8.
4. The Data 6/9
Figure 4: The data. Countries with harmonised and pooled time-series data on attitudes to Russia, China and the United States that were included in this project. For the overwhelming majority of countries, we include data that was collected during the year of the Ukraine conflict (2022), representing the current views of 83% of global population. For most other countries, data is available from 2021. P 6/9.
Sources, surveys, countries and years:
Table 1 presents the thirty sources and their surveys, countries and years. Four sources covered 50 or more countries (Pew 63, IRI 59, FIP 55, Latana 52); one source covered 39 countries (Afro); five sources covered 10 to 19 countries (10, 10, 12, 13, 19); and twenty sources covered 5 or fewer countries – eighteen covering just one country. P 7/10.
5. The New Structure of Global Public Allegiances 8/11
Figure 5: The structure of global allegiances in 2022. Countries with a more than 15 percentage-point lead towards either i) Russia/China or ii) the United States, are indicated by connecting lines. By comparison, the United States enjoys a much larger number of ties to societies that favour America over authoritarian revisionist powers, though, this may in part be due to suppressed favourability towards Russia in the wake of the Ukraine invasion. P 8/11.
Figure 6: Public opinion lead (lag) of the United States versus China, using the latest available surveys for each country. While America has a clear lead in public favourability across the western hemisphere, China is viewed more positively in Asia, and also enjoys strong support across much of Africa. P 9/12.
Figure 7: Trends in global public opinion towards Russia, China and the United States over the course of the past decade. Globally, the United States has retained its popularity lead over China and Russia. However, this masks a major divergence between developed and developing countries. In fellow western countries, America’s relative favourability has soared to newfound highs, though Russia and China have overtaken in the Global South. Each series aggregated using population weights; “self-responses” excluded (e.g. China excluded from measure of global attitudes to China). P 10/13.
Figure 8: Across the world, there is now a high correlation between how societies view Russia and how they view China. This correlation has increased over time, and is now higher than ever before. P 11/14.
Figure 9: Meanwhile, this is accompanied by a growing inverse relationship between whether countries are positive towards Russia and China, and whether they are positive towards the United States. P 11/14.
6. Visualising a Decade of Rising Geopolitical Polarisation 12/15
Figure 10: The spectrum of global public opinion towards Russia, from 2012 (top) to 2022 (bottom). Over the course of the past decade, Russia’s reputation improved across countries of the Global South, notably in populous nations such as India, China, and Indonesia. At the same time, almost all high-income democracies now view Russia negatively – and to an exceptional degree. Data points for Russia itself are excluded from display. 12/15
Figure 11: The spectrum of global public opinion towards the United States, from 2012 (top) to 2022 (bottom). A decade ago, the weight of global public opinion was overwhelmingly positive towards the United States, barring a notable exception among the countries of the Middle East. Today, by contrast, attitudes have polarised – with western nations more than ever behind the U.S., yet a much longer tail of countries where public opinion is ambivalent or even hostile to the United States, led by Iran, China and Russia. In spite of this, however, the average perception of the U.S. in the world as a whole has become more positive over time. Data points for the United States itself are excluded from display. 13/16
Figure 12: The spectrum of global public opinion towards China, from 2012 (top) to 2022 (bottom). Quartiles shown by vertical lines; data points for China itself excluded from display. Over the past decade, there has been little change in the overall spectrum of global public opinion towards China. However, there has been a very clear sorting by political system - with the peoples of high-income democracies now all holding negative views of China, while countries in the Global South – in particular, those under autocratic rule – became more positive. 14/17
7. Regional Divergence Over Time 15/18
Figure 13: Over the past decade, positive sentiment towards Russia has turned downward among high income democracies across the world – including in Europe, the Americas, and the Asia-Pacific region. Yet at the same time, Russia’s reputation has improved across many parts of the Global South – in particular in South Asia, Southeast Asia and the Middle East. 15/18
Figure 14: Russia still remains popular across much of Africa today, especially in countries around the Sahel region – such as Francophone West Africa and the Arabic-speaking countries to its north. 16/19
Figure 15: Over the past decade, China’s image has remained broadly favourable across the Global South, with the exception of a brief dip during the global coronavirus pandemic. However, attitudes towards mainland China have deteriorated significantly across the world’s high-income democracies, whether in Europe, North America, or the Asia-Pacific region, and this has resulted in a growing global divide. Thickness of regional trendlines are relative to total population. “Anglo-Saxon democracies” refers to the United States, United Kingdom, Canada, Australia and New Zealand. 17/20
Figure 16: Positive perception of China, 2022 (%). Favourable views of China have fallen dramatically in western countries, though China has retained and grown in popularity across the Global South, in particular among members of the Belt and Road Initiative. 18/21
Figure 17: In spite of a moderate decline during the Presidency of Donald J. Trump, global perceptions of the United States have become more favourable over the course of the past decade. America also retains a high degree of popularity across the Global South, and even in the Middle East perceptions of the United States have begun to recover since they reached a low-point during the War on Terror. Thickness of regional trendlines are relative to total population. 19/22.
Figure 18: Global map showing percentage of the public with a positive opinion of the United States, 2022 (or most recent observation). Not only have global attitudes become more polarised with respect to Russia and China, but also with respect to America. The U.S. is perceived positively across most of the world, especially among the peoples of Latin America, sub-Saharan Africa and Central-Eastern Europe, yet hostility to the United States is notable among the “Eurasian triad” of Iran, Russia and China. In spite of improvement since the peak years of the global “war on terror,” it still remains elevated in the Islamic world, including in Indonesia, the Arab Middle East and parts of North Africa. 20/23.
Figure 19: While much of the Global South remains neutral between the United States and China or Russia, Latin America stands out as the major exception – with relative positive feelings towards the region’s northern neighbour reaching record highs. Population-weighted trendlines for all regional groups. 21/24
Figure 20: A new Cold War? Index of liberal democracy (V-Dem), rescaled from 0-100, and average public views towards Russia and China (percentages “unfavourable”). Across the world, there is a clear association between the degree of liberal democracy in a country, and how its public feels towards Russia and China. 22/25.
Figure 21: In societies in which democratic institutions are perceived to be underperforming, citizens are more likely to hold receptive attitudes regarding Russia and China. Data on satisfaction with democracy from the Cambridge Global Satisfaction with Democracy Dataset/HUMAN Surveys Project (latest update, January 2022). 23/26.
8. Why is the World Dividing in Two? 22/25
Figure 22: Long-term public opinion in the United States regarding Russia and China. Today’s negative assessments of the two countries are similar to the levels of hostility recorded by public opinion surveys in the 1950s, during the very early years of the Cold War. Cold War figures are based on historical surveys by Gallup, rescaled for equivalence with contemporary sources; both survey sources referred to “Russia” (rather than the Soviet Union) in question formulation. 24/27.
Figure 23: Across the world, countries that have received more support from China’s Belt and Road (BRI) initiative express more positive views towards China today. Moreover, this correlation has strengthened over time – suggesting that Chinese assistance has been effective in improving China’s image in the broader world. Data on Chinese Belt and Road commitments over the period from 2013-2017 are from the AidData Global Chinese Development Finance Dataset, Version 2.0. GDP data are from the World Bank’s World Development Indicators Data API. 25/28
Figure 24: Index of social liberalism, consisting of bootstrap selections from the complete range of items within the World Values Survey that reflect belief in individualism, freedom of choice, support for democracy, and personal autonomy (Foa et al. 2022). While high-income democracies have shifted dramatically towards more socially liberal attitudes, this has led to a wide “values-gap” between the west and the rest of the world. 26/29
Figure 25: Across the world, social liberalism is now one of the strongest predictors of whether a society holds a positive – or negative – view of Russia and China. This association has developed only during the past decade, and was quite weak ten years ago (R = 0.64 today; R = 0.35 in the past). Index of social liberalism selects from items that reflect values of individualism, freedom of choice, support for democracy, and personal autonomy (see Roberto S. Foa, Yascha Mounk & Andrew J. Klassen (2022). “Why the Future Cannot be Predicted". Journal of Democracy, 33(1)). The figures for Russia and China here only report favourability towards the other country. 27/30
9. Conclusion 28/31
Methodology I: Survey Source Items 32/35
Public opinion data on attitudes towards Russia, China and the United States come from 30 different survey sources that were combined and standardized for this report (see Table 1).
For the majority of sources, questionnaires requested a general attitudinal response on a four-point response scale: “very favourable”; “favourable”; “unfavourable”; “very unfavourable”.
For example: Pew Global Attitudes and Trends: “Please tell me if you have a very favorable, somewhat favorable, somewhat unfavorable, or very unfavorable opinion of [country].”
Methodology II: Variable Selection and Validity 33/36
The report involves the combination of data from multiple survey projects. The two main constituent sources are Latana and Pew Global Attitudes and Trends. Latana percentages are roughly equal to combined percentages excluding the Latana percentages. Similarly for the Pew percentages. It may be that this is because Latana percentages are roughly equal to Pew percentages.
(Figure 26, p. 34/37).
Two of the sources
Latana/Alliance of Democracies (2022). The Democracy Perception Index 2022.
https://latana.com/democracy-perception-index-report-2022/
https://www.allianceofdemocracies.org/initiatives/the-copenhagen-democracy-summit/dpi-2022/
Pew:
https://www.pewresearch.org/global/database/
https://www.pewresearch.org/search/global+attitudes?_time_since=past-2-years
A World Divided: Russia, China and the West.
https://www.bennettinstitute.cam.ac.uk/wp-content/uploads/2022/10/A_World_Divided.pdf
THE END
.