I analyze a model of electoral competition with entry under the run-off rule. I consider both two- and multiple-party systems. The principal result is that two-party systems may prove stable under the run-off rule: I show that a continuum of equilibria exists in which only two parties enter and subsequent entry is deterred. This finding conflicts with the accepted wisdom on the run-off rule encapsulated by Duverger's Hypothesis. The results of the model are then reconciled with Duverger's Hypothesis and a more precise formulation is proposed.

Experiments conducted in the field allay concerns over external validity but are subject to the pitfalls of fieldwork. This article proves that scalable protocols conserve statistical efficiency in the face of problems implementing the treatment regime. Three designs are considered: randomly ordering the application of the treatment; matching subjects into groups prior to assignment; and placebo-controlled experiments. Three examples taken from voter mobilization field experiments demonstrate the utility of the design principles discussed.

In this article, I use geographic information systems to develop a continuous measure of district continuity and change following the 2000–02 congressional redistricting cycle. The new measure provides details of where the new population in a district came from and how the old population was distributed within new districts. This measure is then used to demonstrate the independent and interactive influence of district change on competition for congressional elections.

Gordon and Smith (2004) do a great service by introducing innovative and creative quantitative methods that incorporate information from qualitative sources. It is nevertheless important to examine the conditions under which the proposed estimators will be useful in practice. These conditions prove to be surprisingly restrictive: with the possible exception of extremely low-information settings, virtually all of the cases of discernible causation must be coded as such, those codings must contain virtually no errors, and the process by which qualitative researchers produce evaluations of discernibility must conform to the authors' model of the qualitative data-generating process (QDGP) if the procedures are to retain any comparative advantage.

We discuss the motivation for integrating qualitative information in statistical models of complex, partially observable causal mechanisms and suggest ways to minimize the dangers posed by Braumoeller and Kirpichevsky. We stress the importance of linking our statistical estimators to underlying theories of the data generating process, qualitative or otherwise.

In a recent paper published in this journal, Katz (2001) compares the bias in conditional and unconditional fixed effects logit estimation using Monte Carlo Simulation. This note shows that while Katz's (2001) specification has “wrong” fixed effects (in the sense that the fixed effects are the same for all individuals), his conclusions still hold if I correct his specification (so that the fixed effects do differ over individuals). This note also illustrates the danger, when using logit, of including dummies when no fixed effects are present.

I agree with the author's main point. Although I tried to fit a fixed-effects model to the simulated data, those data were generated from a model without fixed effects. In my experiment, therefore, use of the unconditional estimator was perfectly confounded with misspecification of the model. I thank the author for catching this flaw.