Interview with Eduard Ponarin

The LCSR Summer school “Causal models and structural equations” recently finished in Zelenogorsk. Eduard Ponarin, the director of the Laboratory, shares his impressions about the school and tells about further plans of the LCSR.

Last week the 2^nd Summer school of the Laboratory for Comparative Social Research “Causal models and structural equations” came to the end. The program of the school included courses on advanced statistical methods, regular report session of the LCSR’s international research network, and guest lectures by famous sociologists. This year more than 30 young researchers from universities and research centers of Russia, the CIS and Europe were the participants of the school.

Eduard Ponarin, the director of the Laboratory for Comparative Social Research, delivered a course on multilevel regression analysis in the R statistical package. He shared his impressions about the past event and told us about the further plans of the LCSR regarding the “school” program development.

Eduard, what is your opinion about the school in general, what are your impressions?

Personally, I appreciate the school. It seems to me that the researchers were able to master a pretty complicated method of statistical analysis within a short time.  Presentations of models created by the participants using their own data and structural equations were planned for the last day, and I was surprised by how well they managed it.

Maybe the program seemed to be too full of content because not only participants of our research network, but also people aimed at learning structural equations themselves to a greater extent came to the school. We were also discussing their projects which touch upon subjects slightly different from the ones which are of higher priority for the Laboratory. This took some time; on the other hand, it was useful for those who presented those projects. They received feedback, pieces of advice and comments of high-level specialists. I hope this will contribute to their professional development.

Could someone among the new participants become an associated researcher at LCSR in the future?

Potentially yes. Judite Goncalves showed herself as a prospective researcher. Her project about the effects of home care policy on health care use in Switzerland is comparative in general, although it is based on subnational data. I could also mention some participants from Russia and the CIS: Irina Vartanova from Saint-Petersburg State University, Julia Sereda from Ukraine. They are strong researchers with whom it would be interesting to work.

How did you come up with the idea about structural equations? Why this method was chosen to be taught at the school this year?

Indeed it was planned from the beginning of the Laboratory’s work. Our first school was devoted to multilevel models. Due to the fact that the LCSR is engaged into cross-country comparisons, it is impossible to deal without multilevel analysis. It is the main statistical instrument for this research field.

The second stage of the necessary technical skills for research on the LCSR’s topics is structural equation modeling. It is a contemporary method which lets us verify the scales we use. Scholars who apply international databases such as World Values Survey often construct sum indices. A number of assumptions is usually made when such indices are constructed, for example, about equal weights of all variables; but this is not always reasonable. In order to check how correct our assumptions are in each specific case, confirmatory factor analysis has to be used. In order to observe relations between latent variables, structural equations are necessary. We were learning exactly these methods at just finished school.

Next year, if everything is fine, we expect to carry out a summer school on multilevel structural equation modeling so that we could sum up the experience of our first and second schools.

Program of the School included guest lectures delivered by such prominent scholars as Peter Schmidt, Hermann Duelmer, and some others. They told about the most advanced fields of research in social science. Could you please tell us a few words about significance of their studies?

From the methodological point of view the most attractive lecture for me was the lecture by professor Duelmer.  It was devoted to methodological issues of testing the modernization theory developed by Ronald Inglehart and Christian Welzel (2005). He used a very sophisticated approach, namely, multilevel structural equations. The general results of analysis presented by Duelmer corroborated the main postulates of modernization theory. However, some aspects of the latter were not confirmed. In particular, Duelmer showed that societies were aligned on the global cultural map more clearly along a diagonal that reflects economic development, so disturbances by cultural zones appeared to be much less important than it was supposed by Inglehart.

Professor Schmidt’s talk about his study of value diversity in Europe using latent class approach was also very impressive. He criticized both Inglehart’s theory and Duelmer’s empirical results from the Shalom Schwartz’s basic human values theory’s point of view. I think it was very nice that outstanding scholars discussed so influential theories and also showed the participants how their claims should be supported by statistical arguments. As for me, the methodological dispute between Schmidt and Duelmer after the presentation of the latter was the culmination of the school!

If to evaluate the school from the point of view of your expectations before its beginning how do you think it was better or worse than you had expected?

I think it was better. I enjoyed the place very much. Zelenogorsk was wonderful. Concerning the organization of the school the only problem was an excessive number of participants’ presentations in the program. I think we should avoid such an overloaded program during the future schools.

In conclusion, let me ask you some questions about the future. It is interesting whether the LCSR will hold only one school per year or it will be possible to carry out both summer and winter schools?

We will not organize a winter school during next winter because we are not sure about our funding. Nevertheless, I hope that terms of our grant will be prolonged. So it is very likely that the next school on multilevel structural equations will be held in June 2013.

If we have a long-term funding it will also be possible for the Lab to provide two schools per year. Our Laboratory is one of the leading research centers in Russia, especially in terms of methodology. And in my opinion one of the main functions of the LCSR is to provide assistance for young researchers to help them learn advanced statistical methods. If it depended on me, the Lab would conduct more schools for young scholars.

I hope that the Lab will be successful in receiving the funding to provide more statistical training for social scientists. And my final question – to rouse readers’ interest – who are planned to be the lecturers at the next school?

Now we are considering some candidates. Among them are Hermann Duelmer, Bart Mueleman from the University of Loeven and Elhmar Schlüter from the University of Cologne. However, now we have only preliminary arrangements. So I cannot answer your question definitely.

Well, an intrigue is still alive! Thank you very much for your interview, Professor!

 

You can read more about the LCSR Summer School at the Lab’s website. During the next week we will publish reviews of keynote lectures delivered during the event and PDF-presentations of the reports of our researchers. Everyday notes about the most interesting events during the school have already been published.

 

by Boris Sokolov