Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/167686 
Is replaced by the following version: 
Title: 

Machine learning to improve experimental design

The document was removed on behalf of the author(s)/ the editor(s).

Year of Publication: 
2017
Series/Report no.: 
FAU Discussion Papers in Economics No. 16/2017
Publisher: 
Friedrich-Alexander-Universität Erlangen-Nürnberg, Institute for Economics, Nürnberg
Abstract: 
This paper proposes a way of using observational pretest data for the design of experiments. In particular, this paper suggests to train a random forest on the pretest data and to stratify the allocation of treatments to experimental units on the predicted dependent variables. This approach reduces much of the arbitrariness involved in defining strata directly on the basis of covariates. A simulation on 300 random samples drawn from six data sets shows that this algorithm is extremely effective in increasing power compared to random allocation and to traditional ways of stratification. In more than 80% of all samples the estimated variance of the treatment estimator is lower and the estimated power is higher than for standard designs such as complete randomization, conventional stratification or Mahalanobis matching.
Subjects: 
experiment design
treatment allocation
Document Type: 
Working Paper

Files in This Item:
The document was removed on behalf of the author(s)/ the editor(s) on: September 28, 2017


Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.