LIEPP Methodological Seminar

October 12th 2018, 12:30 am - 3:00 pm
  • Actualité Sciences PoActualité Sciences Po

 [language: English]

The Laboratory for interdisciplinary evaluation of public policies (LIEPP) proposes a new initiative: a series of brown bag seminars to discuss methodological issues in an interdisciplinary setting. We seek researchers and doctoral students happy to present and discuss a working paper presenting and discussing substantive methodological issues, i.e. technical problems, coding or measurement issues, link between theory and measurement, etc.

The main aim is to develop a more informed and conscious use of diverse methodological techniques that can foster and enrich our substantive debates around academic work and public policies.

The seminar will take the following format: 

  • A paper is circulated in advance among the members of the seminar group.
  • The discussion will be shortly introduced by the author (no more than 15 min.) and the seminar will take place at lunchtime for one and half hour.

If you are interested in presenting, please contact Emanuele Ferragina ( or Jan Rovny (

For more information, see the detailed presentation of the seminar.


Upcoming sessions:

Session 7: October 12, 2017 (12:30 pm - 3:00 pm) - Salle Lavau (98, rue de l'Université) 

"Survey Data and Multilevel Modeling — Advances on the First and Second Level" by Lucas Leemann

Register for this session

We will start with a presentation, followed by a 1h practical workshop where participants will learn to apply the method. Please bring your computers!

Description of talk: Multilevel regression with post-stratification (MrP) has become quickly a standard approach in public opinion research. Lucas Leemann first presents two distinct motivations for using MrP and also touches on questions of representativity of polling data. He then shows two challenges when employing MrP. The first challenge emerges on the individual level since we hardly ever have the correct data to include all relevant individual level predictors. To overcome this first level problem Lucas Leemann presents MrsP (MrP’s better half) who relies on a synthetic joint distribution. In applying MrP or MrsP we then often also face a second challenge on the context level - which variables do we include and how do we shield our model from over-fitting? Here, he shows what machine learning models can add and compare them to baseline models.


Past sessions:

Session 1: November 15, 2017 (12:30 pm - 2:00 pm) 
"Dealignment Meets Cleavage Theory: Assessing Electoral Stability in Europe" by Jan Rovny. 

Session 2: December 13, 2017 (12:30 pm - 2:00 pm) 
"Between X and Y: how process tracing contributes to opening the black box of causality" by Bruno Palier. 

Session 3: February 14, 2018 (12:30 pm - 2:00 pm) 
"Qualitative analysis with quantitative methods : analyzing teachers' blogs" by Muriel Epstein & Nicolas Bourgeois, SAMM (Statistiques Analyses et Modélisations Mathématiques), Universite Paris 1 Panthéon Sorbonne.

Session 4: March 21, 2018 (12:30 pm - 2:00 pm) 
"Field experiments for policy evaluation: an illustration from a parental reading intervention" by Carlo Barone (OSC/LIEPP).

Session 5: April 12, 2018 (12h30 - 14h00)

"How do inclusionary and exclusionary autocracies affect ordinary people?" by Anja Neundorf (University of Nottingham)

"The cost of Diversity: Tax Progressivity, Immigration, and Preferences for Spending" by Sergi Pardos-Prado  (University of Oxford).

Session 6: May 9, 2018 (12h30 - 14h00)

"Theorising the effect of transition on female labour force in the European semiperiphery: An interdisciplinary methodology" by Sonja Avlijas.