Aug 122022
finnish flag, blue cross flag, finland

For people of a certain disposition, there are only two kinds of respondents: those who vote for the far right (recode to 1) and everyone else (recode to 0). Which is why, looking at the fresh (round 10, version 1.2) European Social Survey data from Finnland, I tried to recode those with a value of 4 (Perussuomalaiset (yes, I looked that one up, then copy/pasted it to make sure) or Finns Party, formerly known as True Finns) to 1.

Now picture my shocked face when this gave me precisely zero far-right Finns. Was this political correctness gone mad? The Perussuomalaiset won 17.5 per cent of the vote in 2019, making them the second strongest party in parliament. Even if there was serious underreporting, it is impossible that they disappear completely from a sample the size of the Finnish ESS. However, there is a group of people in the data, roughly equivalent to the size of the Finns Party’s electorate, who claim to have voted for the Citizens’ Party – an organisation that failed to win a single seat in 2019.

You see where this is heading. The ESS is a marvel, and its data quality remains something to behold. Nonetheless, mistakes may happen even here. The good people at the ESS data archive have been notified, and they agree that this looks somewhat odd. Together with the team in Finnland, they will find out what happened. Unless there really was a freakish, supermassive sampling error, the coding of the variables will be corrected eventually. All will be well again.

And this, boys and girls, is why we have versioning and hashes.

Apr 272021

Working with repeated comparative survey data – almost a howto

There is now a bonanza of studies that rely on surveys which are replicated across countries and time, often with fairly short intervals, with the ESS arguably one of the most prominent examples (but also see the “barometer” studies in various regions). Multi-level analysis is now the weapon of choice to tackle these data, but the appropriate structure of such models is not immediately obvious: are we looking at waves nested in countries? Countries nested in waves? Or rather at surveys cross-classified by year and country? What’s the role of the small-n problem when we are talking about countries? And does the notion of sampling even make sense when we are talking about what is effectively the whole population of countries that could be studied?

  • Schmidt-Catran, A. W., & Fairbrother, M. (2016). The random effects in multilevel models: getting them wrong and getting them right. European Sociological Review, 32(1), 23–38.
  • Schmidt-Catran, A. W., Fairbrother, M., & Andreß, H. (2019). Multilevel models for the analysis of comparative survey data: common problems and some solutions. , 71(1), 99–128.

What we liked

It’s difficult to have a discussion about a text that provides a lot of factual information about methodological bits and bobs, especially when you have little prior knowledge. Having said that, students found both texts (which are related but complementary) remarkably accessible and helpful.

Schmidt-Catran, Fairbrother, Andreß 2019: 112

Sad but true: comparative analysis is hard, and multi-level models are no panacea. Nothing ever is. Bugger.

What we did not like so much

Nothing. Students liked these two. So did I. Period.

Oct 292012
Like social networks, multilevel data structures are everywhere once you start thinking about it. People live in neighbourhoods, neighbourhoods are nested in municipalities, which make up provinces – well, you get the picture. Even if we have no substantive interest in their effects, it often makes sense to control for structures in our data to get more realistic standard errors.

Now the good folks over at the European Social Survey have reacted and spent the Descartes Prize money on compiling multilevel information and merging them with their own data. So far, the selection is a little bit disappointing in some respects. Homicide rates, for instance, are reported on the national level only. But there are some pleasant surprises (I guess due to Eurostat, who collect such things): We get unemployment, GDP growth and even student numbers at the NUTS-3 level. Since you asked, NUTS is the Nomenclature of (subnational) Territory, and level 3 is the lowest level for which comparative data are normally published.

Regrettably, the size and number of level 3 units is not necessarily comparable across countries: For Germany, level 3 corresponds to about 400 local government districts, while France is divided into 96 European Departments. But if you need to combine top-notch survey data with small(ish) regional data, it’s a start, and not a bad one.

Apr 122012
The European Social Survey’s Core Scientific Team (formerly known as the Central Coordination Team) has just announced in the User Bulletin (distributed via email, not yet on the website) that they will remove a cool 27 items from the core questionnaire, and three more from the supplementary questionnaire. The items in question are A3-A7, B21-B22, B32-B33, C7-C14, F6a, F34, F43-F47, F51-F52, F57-F58, F71-F73 (“referring to their round 4 question numbers”).

Now I’m sure you all know your round 4 question numbers by heart, but I don’t, so I looked them up. From round 6 on, we will miss information on use of radio, newspapers, and the internet (both global and politics specific), party membership, support for bans of extremist party, believe in scientific solutions to ecological problems,  worries about crime (six items), support for anti-terror measures, field of highest qualification, ability to borrow money from friends or family, detailed information on partner’s, mother’s and father’s work, and phone ownership/access.


Ye olden days (photo credit: Wikipedia)

From a political science vantage point, use of media and party-political questions are obviously absolutely essential, while respondents’ views on torture and terrorism are interesting at the very least. Sociologists, on the other hand, will worry about the loss of information required for Goldthorpe coding and the fact that they cannot measure fear of crime any longer. For me, the ESS is one of the most important collective resources for social research, and my instinct is to object to any cuts to the questionnaire.

On the other hand, this resource has a  price tag attached to it. Some ten years ago, it was estimated that the fieldwork in the original 16 countries would cost 4.2 million euros per round. In the meantime, both the number of countries and the fees charged by the pollsters have risen considerably. But are the savings from sacrificing these items relevant given that they make up only a fraction of the total questionnaire, that there are considerable fixed costs, and that the total costs of the ESS are still small beer compared to what Europe spends on rocket science, its subsidised industries, or agriculture?

The Core Scientific Team has promised to publish a full report on the cuts by autumn 2013. In the meantime, what are your views on the matter?

European Social Survey to cut 30 items from its core questionnaire 1
Aug 292008
Everyone just seems to know that the voters of the Extreme Right hate foreigners in general and immigrants in particular, but robust comparative evidence for the alleged xenophobia – Radical Right vote link is scarce. Moreover, many of the published analyses are based on somewhat outdated (i.e. 1990s) data, and alternative accounts of the extreme right vote (the “unpolitical” protest hypothesis and the hypothesis that the Far Right in Western Europe attracts people with “neo-liberal” economic preferences, championed by Betz and Kitschelt in the 1990s) do exist. Just a few days ago, a journal has accepted a paper by me in which I test these three competing hypotheses using (relatively) recent data from the European Social Survey and a little Structural Equation Modelling. As it turns out, protest and neo-liberalism have no statistically significant impact on the Extreme Right vote whatsoever. Anti-immigrant sentiment, however, plays a crucial role for the Extreme Right in all countries but Italy. Its effects are moderated by party identification and general ideological preferences. Moreover, the effect of immigrant sentiment is moderate by general ideological preferences and party identification. I conclude that comparative electoral research should focus on the circumstances under which immigration is politicised. Wasn’t it blindingly obvious?

Technorati-Tags: extreme right, radical right, populist right, far right, sem, structural equation modelling, western europe, italy, immigration, comparative politics, european social survey, voting, voters