The Use of Instrumental Variables and Determining the Direction of Effects in Cross-Sectional Data

On the 19th of November, the SELS staff and students attended a seminar on the use of instrumental variables. The guest speaker was Viktor Lushin from New York University.

The seminar focused on the use of instrumental variables and on determining the direction of effects in cross-sectional databases. Practical aspects of the use of instrumental variables received a special attention. 

Dr. Lushin demonstrated the basic principles for selecting the appropriate variables, in the context of reciprocal effects. He gave examples of mediation and multiple-regression submodels, and explained the ways of establishing the direction of effects between strongly correlated variables with the use of instrumental variables.

He also focused on practical recommendations for preparations to the cross-sectional studies, and on processing the data already collected by SELS. Finally, he examined the way of correcting models that don’t appear to be valid in order to obtain more accurate and representative results.

The lecturer presented the material in such a way that these complicated ideas were clear even to the researchers with little experience in analyzing the data.