Model Uncertainty and Robustness

Empirical results can be very dependent on model specification. How can analysts show model robustness in their research?

 

 Model Uncertainty and Robustness: A Computational Framework for Multi-Model Analysis.  Sociological Methods and Research. February 2017. Vol. 46(1): 3 – 40. (Lead article, with Katherine Holsteen)

robustness

Download: This do file installs our Stata program, loads in data sets, and replicates all the analyses in the paper.

OR: Paste the following command into Stata:

do http://web.stanford.edu/~cy10/public/mrobust/install_mrobust.do

OR: type “ssc describe mrobust” in Stata

 

 

The multi-model analysis framework allows researchers to

(1) estimate thousands of regression models across combinations of control variables, functional forms, alternative variable definitions, and estimation commands;

(2) display the resulting modeling distribution of estimates; and

(3) discover which aspects of model specification have the most influence on a parameter estimate.

This allows analysts to clarify and demonstrate which modeling assumptions are essential to their empirical findings, and which are not.

 

NEW working paper:

We Ran 9 Billion Regressions: Eliminating False Positives Through Computational Model Robustness.” 2017. With John Muñoz.

 

This is part of an enduring theme in my research, focusing on robust results. My earlier research:

Young, Cristobal. Model Uncertainty in Sociological Research: An Application to Religion and Economic Growth. American Sociological Review. June 2009.