G. Rosas and Y. Shomer, “Models of Nonresponse in Legislative Politics”, Legislative Studies Quarterly, 33 (4), 2008.
Tools dedicated to inferring the ideological leanings of legislators from observed votes — techniques such as NOMINATE (Poole and Rosenthal 1997) or the item-response-theory model of Clinton, Jackman, and Rivers (2004) — rest on the assumption that the political process that generates abstentions is ignorable, an assumption not always easy to justify. We extended the item-response-theory model to analyze abstention and voting processes simultaneously in situations where abstentions are suspected to be nonrandom. We applied this expanded model to two assemblies where the existing literature gives reason to expect nonrandom abstentions, and we demonstrate how our extensions yield nuanced analyses of legislative politics. We also acknowledge limits to our ability to decide on the adequacy of alternative assumptions about abstentions, since these assumptions are not readily verifiable.