Multivariate Statistical Techniques for Comparing Countries

A growing practice in the social sciences is the comparison of social and political behaviours between social groups. The focus can be on economic, political, cultural, socio-psychological attitudes and behaviours, it can be at an individual or collective level, and the groups of interest can be defined by gender, ethnicity, age, or generation.

One particular type of group is at the centre of this course: nations or countries. Indeed, the trend in cross-national and cross-cultural research has gained greater popularity in recent decades, especially since the developments of comparative cross-national databases such as Eurobarometer, EVS, ISSP and ESS.

Thanks to methodological developments in comparative survey designs such as the one achieved by ESS or equivalent surveys, we have taken a significant step forward. A new big challenge for comparative survey analysis, therefore, is to analyse these comparative datasets with statistical techniques and methods that make it possible to test for country effects.

Unfortunately, many users still simply juxtapose their results country by country, and few are really dealing with the logic of statistical control for country effects. When they introduce country effects in their explanatory models, it is mostly in the form of dummy effects coding in regression analysis, which only account for the heterogeneity of the dependent variable in the different countries.

But how to test for heterogeneity/homogeneity of beta parameters, the one estimating the effects of the explanatory variables on the dependent one?

How can we know that beta parameters estimated by a logistic regression analysis are statistically different across nations?

How can we know that factorial structures diverge between countries or are similar?

These are the kinds of big issues this course will cover, looking at the most-used methods in statistical comparative analysis.

Each day, we will focus on a specific method, with strong emphasis on two blocks of methods: modelling techniques (regression – linear and logistic, loglinear and latent class models) and data reduction techniques (factor analysis, multiple correspondence analysis). I will emphasise complementarities between these types of methods.

Modelling techniques invite the user to apply a model to their data structures: an algebraic expression such as the classical linear regression model. This family of methods attach great importance to hypothesis testing, in particular to the significance of models and their parameters.

But how do you cope with the countries’ effects in that case? Tests may be sensitive to sample sizes; for instance, if a researcher compares two countries with fairly different N. How do you know that when beta parameters have higher values in one country than another, it really means a country difference? On the other hand, data reduction techniques are great at reducing the space of variables and/or individuals or cases: with big databases like Eurobarometer, ESS or EVS, which have so many indicators, reducing the data space to few dimensions is a very interesting objective. But what constitutes a ‘factorial invariance’ across countries? How could we say that the factorial space is homogeneous or not across nations?

Factor analysis and correspondence analysis propose certain solutions to these problems. These multivariate techniques can also be used as preliminary analysis before running regression models, as can cluster analysis, a powerful method to find groups across databases.

Finally, how do you correlate individual patterns of political/social attitudes and behaviours with the national macro-level characteristics of countries? The course will finish with an introduction to multilevel regression models.

By the end of this course

Though not exhaustive, this course offers a broad overview of some very important methods to help comparatists deal with country effects. It crams a lot into only two weeks; if you were working only with a textbook, you could spend months discovering so many different techniques.

Acknowledging that no method is perfect, the course will show, by theoretical presentations and empirical illustrations, the advantages and limits of each method type. By the end of the course, you should know which methods are best suited to your particular research perspectives and types of data.

Further study

If you want to learn more advanced methods such as confirmatory factor analysis and structural equation modelling, the Summer School offers other courses specialising in these.

This is an introductory level course, even if the methods taught are multivariate.

If you are struggling to enter into the world of multivariate techniques, this course will help you overcome these first difficulties. 

The course requires no skills other than basic descriptive statistics and hypotheses testing, but if you need a refresher in these, email the instructor before the course; if needed the Instructor and/or the TA can propose a remedial on this at the very beginning of the Summer School.