We attempt to clarify, and suggest how to avoid, several serious misunderstandings about and fal-lacies of causal inference in experimental and observational research. These issues concern some of the most basic advantages and disadvantages of each basic research design. Problems include improper use of hypothesis tests for covariate balance between the treated and control groups, and the consequences of using randomization, blocking before randomization, and matching after treatment assignment to achieve covariate balance. Applied researchers in a wide range of scien-tific disciplines seem to fall prey to one or more of these fallacies, and as a result make suboptimal design or analysis choices. To clarify these points, we derive a new four-part decomposition of the

This thesis presents a measurement-parameterized performance study of deploy-ment factors in wireless mesh networks using four performance metrics: client cov-erage area, backhaul tier connectivity, protocol-dependent throughput, and per-user fair rates. For each metric, I identify and study deployment factors which strongly influence mesh performance via an extensive set of Monte Carlo simulations capturing realistic physical layer behavior. My findings include: (i) A random topology is un-suitable for a large-scale mesh deployment due to doubled node density requirements, yet a moderate level of perturbations from ideal grid placement has minor impact. (ii) Multiple backhaul radios per mesh node is a cost-effective deployment strategy as it leads to mesh deployments costing 50 % less than with a single-radio architecture. This work adds to the understanding of mesh deployment factors and their general impact on performance, providing further insight into practical mesh deployments. Acknowledgments First and foremost, I would like to thank my advisor, Dr. Edward Knightly, for the guidance, support, and opportunities he has provided me. He has been a

Abstract. A basic feature of many field experiments is that investigators are only able to randomize clusters of individuals—such as households, communities, firms, medical practices, schools or classrooms—even when the individual is the unit of interest. To recoup the resulting efficiency loss, some studies pair similar clusters and randomize treatment within pairs. However, many other studies avoid pairing, in part because of claims in the literature, echoed by clinical trials standards organizations, that this matched-pair, cluster-randomization design has serious problems. We argue that all such claims are unfounded. We also prove that the estimator recommended for this design in the literature is unbiased only in situations when matching is unnecessary; its standard error is also invalid. To overcome this problem without modeling assumptions, we develop a simple design-based estimator with much improved statistical properties. We also propose a model-based approach that includes some of the benefits of our design-based estimator as well as the estimator in the literature. Our methods also address individuallevel

for detailed comments which significantly improved the presentation of the analytical results given in this paper.