Elements of estimation theory for causal effects in the presence of network interference

Randomized experiments in which the treatment of a unit can affect the outcomes of other units are becoming increasingly common in healthcare, economics, and in the social and information sciences. From a causal inference perspective, the typical assumption of no interference becomes untenable in such experiments. In many problems, however, the patterns of interference may be informed by the observation of network connections among the units of analysis. Here, we develop elements of optimal estimation theory for causal effects leveraging an observed network, by assuming that the potential outcomes of an individual depend only on the individual's treatment and on the treatment of the neighbors. We propose a collection of exclusion restrictions on the potential outcomes, and show how subsets of these restrictions lead to various parameterizations. Considering the class of linear unbiased estimators of the average direct treatment effect, we derive conditions on the design that lead to the existence of unbiased estimators, and offer analytical insights on the weights that lead to minimum integrated variance estimators. We illustrate the improved performance of these estimators when compared to more standard biased and unbiased estimators, using simulations.

[1]  J. Grizzle THE TWO-PERIOD CHANGE-OVER DESIGN AN ITS USE IN CLINICAL TRIALS. , 1965, Biometrics.

[2]  D. Rubin Estimating causal effects of treatments in randomized and nonrandomized studies. , 1974 .

[3]  Donald B. Rubin,et al.  Bayesian Inference for Causal Effects: The Role of Randomization , 1978 .

[4]  R. Kempton,et al.  Inter-plot competition in variety trials of field beans (Vicia faba L.) , 1984, The Journal of Agricultural Science.

[5]  C. Manski Nonparametric Bounds on Treatment Effects , 1989 .

[6]  Nicholas J. Higham,et al.  INVERSE PROBLEMS NEWSLETTER , 1991 .

[7]  Albert,et al.  Emergence of scaling in random networks , 1999, Science.

[8]  Yousef Saad,et al.  Iterative methods for sparse linear systems , 2003 .

[9]  Mark E. J. Newman,et al.  The Structure and Function of Complex Networks , 2003, SIAM Rev..

[10]  P. Rosenbaum Interference Between Units in Randomized Experiments , 2007 .

[11]  Susan A. Murphy,et al.  Monographs on statistics and applied probability , 1990 .

[12]  M. Hudgens,et al.  Toward Causal Inference With Interference , 2008, Journal of the American Statistical Association.

[13]  Peter D. Hoff,et al.  A First Course in Bayesian Statistical Methods , 2009 .

[14]  Charles F. Manski,et al.  Identification of Treatment Response with Social Interactions , 2013 .

[15]  Edward K. Kao,et al.  Estimation of Causal Peer Influence Effects , 2013, ICML.

[16]  Jon M. Kleinberg,et al.  Graph cluster randomization: network exposure to multiple universes , 2013, KDD.

[17]  Dean Eckles,et al.  Design and Analysis of Experiments in Networks: Reducing Bias from Interference , 2014, ArXiv.

[18]  Edoardo M. Airoldi,et al.  Optimal design of experiments in the presence of network-correlated outcomes , 2015, ArXiv.

[19]  G. Imbens,et al.  Exact p-Values for Network Interference , 2015, 1506.02084.

[20]  M. McQueen,et al.  The National Longitudinal Study of Adolescent to Adult Health (Add Health) Sibling Pairs Genome-Wide Data , 2015, Behavior genetics.

[21]  Anmol Bhasin,et al.  Network A/B Testing: From Sampling to Estimation , 2015, WWW.

[22]  N. Christakis,et al.  Social network targeting to maximise population behaviour change: a cluster randomised controlled trial , 2015, The Lancet.

[23]  Sandra Maurer Bayesian Methods For Finite Population Sampling , 2016 .

[24]  Dean Eckles,et al.  Estimating peer effects in networks with peer encouragement designs , 2016, Proceedings of the National Academy of Sciences.

[25]  Peter M. Aronow,et al.  Estimating Average Causal Effects Under Interference Between Units , 2013, 1305.6156.

[26]  J. Aislinn Bohren,et al.  Optimal Design of Experiments in the Presence of Interference , 2017, Review of Economics and Statistics.