In many social sciences, and political science in particular, establishing causal relations with the data that is typically available is an effort plagued by a number of empirical concerns. Confronting these challenges has stimulated the use of techniques to establish a credible research design by developing an identification strategy. A critical part of this process, especially in quasi-experimental settings, is the analyst's theoretical account of exogenous variation in the mechanism that assigns units across different treatments. Such accounts often exploit an analogy between the quasi-experimental setting under scrutiny and a laboratory environment. How comprehensive, or how restrictive, are such analogies? In this article, we formulate the potential outcome framework in a way that isolates the role of assignment mechanisms---the key difference between experiments and quasi-experiments---to study the all-else-equal influence of an assignment mechanism on the estimands identified by a research design. We then use our framework to study how strategic factors influencing assignment affect the substantive interpretation of these estimands.