Colchester, United Kingdom

Comparing Countries with Multivariate Statistical Techniques

online course
when 22 July 2024 - 26 July 2024
language English
duration 1 week
credits 4 EC
fee GBP 492

Learn key statistical techniques for cross-national data analysis, including 'country effect' in multivariate methods, comparing regression estimates, and exploring factorial analysis (PCA) with a focus on political science methodology intricacies.

Need to Know

You must have some introductory familiarity with basic descriptive statistics, statistical inference and some basics about regression models. If you are not, take the course Introduction to Inferential Statistics before you sign up for this one.

In Depth - Key topics covered

The course consists of five sessions, organised into four main topics.

Topic 1: What is a ‘country effect’ and how is it translated in multivariate statistical methods?

In this topic, you will review substantial and methodological issues in multivariate statistical analysis for comparing countries, particularly what is a 'country effect' and how this translates in statistical terms. You will also engage in very simple recalls using descriptive and bivariate techniques applied to 'country effects' problems.

Topic 2: Comparing regression estimates (linear, logistic, log-linear regression) between countries

In this topic, you will learn how regression methods respond to a much more complex problem than commonly acknowledged: How can we compare the values of estimated parameters, goodness of fit statistics, and statistical tests between countries? Notably, this varies significantly for linear regression and logistic regression. There’s even a debate over whether we can compare logit coefficients between countries directly. What strategies can we employ to navigate this challenge?

Topic 3: Comparing factorial analysis results (PCA analysis dimensions, factor loadings)between countries

In this topic, you will learn how data reduction techniques such as principal components (and more generally scaling techniques) respond to the main question of the course, a seemingly simple question but in fact much more complex: Can we directly compare the values of the coefficients of a PCA type factor analysis obtained between different countries?You will discover that researchers have proposed interesting technical solutions to compare factor loadings (Procruste rotations, Tucker coefficient for example). But what are the limits of these methods, and can we go further in the comparison between countries? What steps should we take?

Topic 4: Comparing countries' variability with multilevel regression models

In this topic, you will learn that regression analysis of harmonised data from multiple countries (like the ESS, Eurobarometers, Share) in which individual-level responses are modeled as a function of both individual-level and country-level characteristics, the so-called ‘multilevel models’ or ‘hierarchical models’ are among the most popular of quantitative approaches. Is this the best way to approach cross-national comparisons and how do these methods deal with comparing estimated coefficients? What are the limitations? Can we apply these techniques to a small number of countries?

Course leader

Bruno Cautrès is attached to Centre de recherches politiques de Sciences Po (Paris. He is a senior CNRS research fellow with interests in voting behaviour, political attitudes and behaviours, comparative survey research and quantitative techniques.

Target group

Researchers, professional analysts, advanced students

Course aim

The course offers an introduction to the main statistical techniques used to analyse cross-national comparative data analysis, particularly survey data. This course highlights the most important methods of quantitative analysis in the social sciences but revisits them from the perspective of comparing the statistical estimates between countries.

The central enquiry revolves around how different statistical methods treat the ‘country effect’ and identify it: how statistical models (linear regression, logit models, multilevel regression models), scaling techniques of data reduction methods (PCA analysis for instance) test for the ‘invariance’ of the relationship between variables across countries.

The course seeks to address a couple of seemingly straightforward questions , that unearth methodological intricacies far beyond initial expectations: Can we feasibly compare coefficient values derived from primary statistical methods in political science across different countries? What are the primary pitfalls and viable resolutions within this pursuit?

The objective of a statistical analysis between countries is, in fact, to determine whether the variance between countries is greater than the variance within each country. To do this, most researchers compare the values of the coefficients (for example the coefficients of regression models, indicators of goodness of fit). However, they often do so empirically without taking a certain number of methodological precautions (for example, the fact that the sample size is not the same, or that the residuals are not distributed the same). The objective of the course is thus to teach simple but efficient methods to avoid falling into certain traps of comparative analysis using quantitative techniques

The course has also two pedagogical focal points:
1. To teach with a reasonable level of formalisation, as much as required to understand the methods.
2. To highlight the links among the different methods and facilitate understanding of how they complement each other.

Credits info

4 EC
You can earn up to four credits for attending this course.
3 ECTS credits – Attend 100% of live sessions and engage fully with class activities.
4 ECTS credits – Attend 100% of live sessions, engage fully with class activities and complete a post-class assignment.

Fee info

GBP 492: ECPR member - check whether your institution is a member here:
GBP 985: ECPR non-member


Funding applications for the 2024 ECPR Methods School summer programme are now opening for applications. Apply before mid-April 2024. For more details on funding opportunities for ECPR's other events, please visit