scholarly journals Adaptive Treatment Assignment in Experiments for Policy Choice

Econometrica ◽  
2021 ◽  
Vol 89 (1) ◽  
pp. 113-132 ◽  
Author(s):  
Maximilian Kasy ◽  
Anja Sautmann

Standard experimental designs are geared toward point estimation and hypothesis testing, while bandit algorithms are geared toward in‐sample outcomes. Here, we instead consider treatment assignment in an experiment with several waves for choosing the best among a set of possible policies (treatments) at the end of the experiment. We propose a computationally tractable assignment algorithm that we call “exploration sampling,” where assignment probabilities in each wave are an increasing concave function of the posterior probabilities that each treatment is optimal. We prove an asymptotic optimality result for this algorithm and demonstrate improvements in welfare in calibrated simulations over both non‐adaptive designs and bandit algorithms. An application to selecting between six different recruitment strategies for an agricultural extension service in India demonstrates practical feasibility.




Author(s):  
C. Sanga ◽  
V. J. Kalungwizi ◽  
C. P. Msuya

This article was designed to present the assessment of the effectiveness of radio - based, impact driven smallholder farmer extension service system provided by FVR to enhance accessibility of extension services to women and men in the project areas of Tanzania. Specifically, this paper assessed women and men farmers' access to ICT and factors influencing the utilization of ICT to deliver agricultural information and knowledge. The paper used data from impact assessment survey of the project conducted between April 2012 and June 2012. These data were complemented by focus group discussion involving members of gender advisory panel that had been established in the selected project sites. Quantitative data were analyzed to yield frequencies and percentages. Qualitative data were analyzed by content analysis. Even though ownership of mobile phones and radio was higher among women in all study areas both men and women farmers' had almost the same percentage in accessibility to agricultural extension information. The factors that affected women and men farmers to get quality agricultural information via these ICT tools were namely: poor radio signal reception, power outrage and poor timing of radio programs among others. This is important evidence that careful use of ICT can reduce gender imbalance in agricultural extension services and information delivery.



2018 ◽  
Vol 46 (3) ◽  
pp. 458-469
Author(s):  
Yanqing Yi ◽  
Xuan Li


2010 ◽  
Vol 9 (4) ◽  
pp. 473-502 ◽  
Author(s):  
Angela Firkus

Congress founded the Agricultural Extension Service (AES) in the Smith-Lever Act of 1914 to disseminate agricultural research to individual farmers. In some states the AES also worked to encourage Native Americans to adopt sedentary intensive agriculture and all aspects of assimilation connected with that occupation. J. F. Wojta, AES administrator in Wisconsin from 1914 to 1940, took a deep interest in Indian farmers and used the power and resources of his office to instruct Native Americans. Ho-Chunks, Menominees, Ojibwes, and Oneidas in Wisconsin adopted or rejected these social, economic, and political assimilation efforts during the Progressive Era according to their own circumstances and goals. The experience of Wisconsin tribes with the state's agricultural extension programs illustrates different ways that Native peoples tried to benefit from modern government services while maintaining their own culture and kinship ties.



Author(s):  
Sinan Aral

This chapter considers the design and analysis of networked experiments. As a result of digitization, the scale, scope, and complexity of networked experiments have expanded significantly in recent years, creating a need for more robust design and analysis strategies. This chapter first reviews innovations in networked experimental design, assessing the implications of the experimental setting, sampling, randomization procedures, and treatment assignment. Then the analysis of networked experiments is discussed, with particular emphasis on modeling treatment response assumptions, inference, and estimation, and recent approaches to interference and uncertainty in dependent data. The chapter concludes by discussing important challenges facing the future of networked experimentation, focusing on adaptive treatment assignment, novel randomization techniques, linking online treatments to offline responses, and experimental validation of observational methods. I hope this framework can help guide future work toward a cumulative research tradition in networked experimentation.



Sign in / Sign up

Export Citation Format

Share Document