Monday, 8 February 2016


Veteran readers will know of my love affair with the European Values Study of 2008 - the last snapshot of Greek society's beliefs and norms before the crisis. It's heartening to know that Greece will also take part in the 2017 iteration of the survey, which will allow us to assess the impact of the crisis on the Greek psyche in more detail than has hitherto been possible.

Today I was hoping to use the 2008 study for a simple stock-take of Greece's libertarian population. The EVS does not discuss libertarianism as such but does ask a range of questions to help establish people's attitudes towards social mores, institutions and the role of the state. Many of these questions are scored conveniently on a scale of 1 to 10 and look into behaviours that respondents approve or disapprove of and their level of agreement with a range of statements about society and the economy.

Factor analysis can be applied to these variables to reveal some of the key attitudes underlying people's responses and score individual respondents based on the extent to which they share each of them (scores will follow a normal distribution with a mean of 0 and a standard deviation of 1).

In the case of Greece, seven core attitudes emerge:
  1. Social Liberalism (tolerance for homosexuality, abortion, divorce, assisted suicide, casual sex, prostitution, adultery, soft drugs, suicide)
  2. Dishonest Self-Interest (tolerance for lying in own interest, accepting a bribe, benefits fraud, paying bribes, tax evasion)
  3. Preference for State Control of the Economy (tolerance for state intervention, competition seen as harmful, belief in state responsibility for the worse-off, preference for public ownership of enterprises, tolerance for less conditionality in unemployment benefits).
  4. Distinction between 'victimless crimes' and behaviours involving obvious detriment (tolerance for joyriding, avoiding fares in public transport, tax evasion)
  5. Appetite for controversial science (GM food, experiments involving human embryos)
  6. Approval of the Death Penalty
  7. Aversion to redistribution of incomes (Preference for rewarding effort over equal outcomes, self-identification as right-wing).

This analysis is based on 999 Greek responses (out of a full sample of 1,500 - each case had to have responses for all of the relevant questions to make it into the factor analysis).

If you run the exact same analysis on the pan-European version of the EVS2008, (using 38,218 out of 67,786 responses) the same factors emerge, with the exception of the 'victimless crimes' factor. This does not mean this way of thinking is unique to Greece, of course; only that it is not universally present throughout Europe. Note that 'Europe' for the purposes of the EVS includes all of the Nordics, as well as Russia, Turkey and the Caucasus. I'll try to use 'Europe+' where possible to remind you of the fact.

By design, the factors are orthogonal and standardised - ie they are not correlated. This means one must be careful and imaginative in interpreting them. For example, most people who want a redistribution of income also have at least some tolerance for state control of the economy. However, for the two factors to be orthogonal the concepts need some rewording so that they can be truly independent of each other. So the State Control factor is more about state ownership of the means of production and "to each according to his needs..." - classical Marxism in one sense - while the Redistribution factor deals more specifically with fiscal policy and the willingness to redistribute income through tax and benefits. A person can believe in public ownership and also not be convinced by redistribution of income - after all what's the point of taxation and a benefits system in the full-employment paradise of centrally planned production (and consumption)?

With this caveat in mind, it's fair to say that factors 1 and 7 are probably enough to identify potential libertarians, who should score high in factor 1 (socially liberal) and factor 7 (averse to redistribution of income), and low on factor 3 (against state control of the economy and redistribution of income). I realise I'm oversimplifying of course. It's possible to be a socially conservative libertarian; you may strongly disapprove of, say, adultery, or abortion, and even lecture people against them privately. However, if you always stop short of demanding that government legislate against such behaviours, or that they be penalised in some other mandatory way, you're a libertarian in my book. Unfortunately we don't have the kind of data that would allow this.

It's possible, using the pan-European version of the dataset, to position 2008 Greece in the spectrum of libertarianism: at .1 standard deviation above the European+ mean we were not doing quite so badly for social liberalism. We were marginally less averse to the redistribution of income than the average European+ country (at .16 of a standard deviation below the mean). But we were also one of the highest-ranked countries (at .32 of a standard deviation above the mean) in terms of wanting the government to have control of the economy. And that was in 2008. Commentators who dub Greece the last Soviet republic in Europe (and are routinely vilified for this) kind of have a point.

Now if we assume that libertarians need to be above the European+ average in terms of social liberalism and below the average in their appetite for state control and redistribution, then the number of Greek libertarians was small in 2008 - a mere 6.4% of the adult population and proportionately less than half the European+ average (13.6%). The leaderboard of countries with big libertarian populations is packed with notorious hell-holes like Denmark, Sweden, Norway, Iceland and the Netherlands. Nearly half of all Danes are libertarian by this grouping.

This is of course only a crude way of grouping people (I'll refer to it as the quadrant method for ease of reference). One could, alternatively, use cluster analysis to arrive at less arbitrary, more cohesive groups within which individuals have more in common. No method I've tried using the variables above creates a libertarian cluster naturally, whether in Greece or EU-wide. More skilled people might be able to produce better results (I'm looking at you @dimmu). The best I could do was a k-means clustering in which I forced SPSS to produce 8 clusters (the 'Octo-grouping'). The Octo-clusters are best interpreted as political tribes - the kind of people likely to take similar sides of a public issue. Often members of the same political tribe have significant ideological differences, but can put these aside for a common cause.

The results change a fair amount using this more cohesive grouping, and I find the way in which the results change particularly disappointing.
Under the 'Octo-grouping', the number of Greek libertarians rises slightly (to 7.2% against a Europe+ average of 10.9%). But the character or the group changes substantially - becoming less socially liberal and more extreme in their views on state control and income redistribution. In fact, Greek 'libertarians' under the Octo-grouping are not, on average, socially liberal; they straddle the axis instead. Only about 40% of the Greek octo-libertarians score against the Europe+ averages the way I described earlier; the bulk of the rest of the group is made up of individuals that I can only call 'free market paternalists' - people who hate state control and income redistribution but are socially conservative. Meanwhile, many 'quadrant' libertarians have more tribal ties with a very different political tradition - social liberals who believe that a certain level of state ownership can obviate the need for mass income redistribution by taking care of citizens' basic needs (think Singapore's policy on housing for instance).

From 2008 to 2015: who are we now?

Will the headline figures have changed much since 2008? On the one hand, some of the broad
attitudes described above surely correspond to stable personal values. On the other, it's been eight disastrous years for Greece since the EVS, and 'neoliberalism' has since been established for many Greeks as the source of all our misfortunes. This term (see here for one of the tortured, rambling definitions held up as definitive) has been applied indiscriminately to libertarians as well as others. Although not all of us identify with it we certainly know we're likely being talked about when it is used.

Well, two surveys of 1,000 Greeks each by DiaNEOSis found that around 10% self-identified as 'neoliberals' between April (10.6%) and November 2015 (9.3%). DiaNEOSis' 'neoliberals' are a fairly good proxy, I think, for my EVS (quadrant) libertarians, at least as far as their political affiliation goes. However, the Greek centre-right was twice as likely to be 'neoliberals' in 2015 as they were to be 'libertarians' in 2008, and at first I wondered to what extent that's because the Greek centre-right has largely emptied in the meantime. In reality, its share of the population has remained remarkably stable (18.5% in EVS, 19.3% in April DiaNEOSis and 20% in November DiaNEOSis). One explanation might be that the Greek crisis (or perhaps Syriza/ANEL rule) has shifted much of the Greek centre-right into the libertarian camp by convincing them of the dangers of excessive spending or state intervention. Or perhaps they've been drawn into the fold by the endless flow of divisive rhetoric of the past year. Or maybe, self-identification with the pejorative term 'neoliberal' suggests a confrontational attitude versus parts of the Greek political spectrum but few actual libertarian beliefs.

This section under contstruction

My original tables for this section were wrong. I am recalculating everything. Bear with me.

I need to provide some context here: EVS fieldwork in Greece took place between 12 September and 26 October 2008. This was before the mass protests of December 2008, and almost a year before the national elections of 2009. It followed the narrow passing of a pension reform bill (modest by today's standards but still strongly opposed) in March 2008, and the polls had already swung (narrowly) in favour of a PASOK win, with margins of anything between 0 and 3%. Syriza, or perhaps Ur-Syriza as it was still a radical, movement-based party at the time, was polling at anything between 8% and 10%.

Annex: Code for the factor analysis

The factor analysis is easy to reproduce if you have the raw EVS data, and the code is identical for both the Greek and the full EVS datasets:
  /VARIABLES v193 v194 v195 v196 v197 v198 v199 v233 v234 v235 v236 v237 v238 v239 v240 v241 v242 
    v243 v244 v245 v246 v247 v248 v249 v250 v251 v252
  /ANALYSIS v193 v194 v195 v196 v197 v198 v199 v233 v234 v235 v236 v237 v238 v239 v240 v241 v242 
    v243 v244 v245 v246 v247 v248 v249 v250 v251 v252