Smith and Remington use the choices made by Russian legislators as they crafted their new parliament to learn about the motives of members of the Duma in particular and about legislators in general. The authors are interested in measuring the relative weight of what they call “policy incentives,” “electoral incentives,” and “party incentives.” To do this they examine three attributes of legislators: policy preferences, measured on a one-dimensional continuum from Left to Right; mode of election, that is, some Duma members were elected from single-member districts (SMDs), and others were chosen through a system of proportional representation (PR); and faction membership. These three characteristics create a taxonomy of Duma members.

The authors link the three characteristics to “incentives,” but this is problematic. For example, knowing that PR and SMD deputies behave differently does not directly measure the strength of reelection incentives (often refered to in other literature as “office motivation”). Differences in what it takes to be elected from SMD and PR constituencies may lead office-motivated deputies from SMD constituencies to behave differently from their PR counterparts, but this is only one of many possible explanations. Another is that ideologically motivated independent deputies with no interest in reelection may view competing on an artificial party list as an infringement on their independence, whereas ideologically motivated members of major parties find competing on PR lists more congenial. The link between incentives and a characteristic such as a deputy's mode of election is not airtight.

Smith and Remington use the multitude of institutional choices made during the mid-1990s by Duma members to gauge the effect of their three measured characteristics. They also look for, and find, evidence that preferences toward institutional arrangements evolved over time as deputies learned. The taxonomy is both a strength and a weakness. It enables the authors to ask concrete questions that can be answered, but this comes at the cost of not being able to address fundamental questions about legislatures directly. For example, how important is the “information processing” roll of committees relative to their role as gatekeepers for political parties or special interests? To what extent do legislators join parties because of their policy preferences, and to what extent are their preferences the result of their party affiliation?

The bulk of the book uses several important decisions reached during the Duma's first session as “natural experiments.” The authors look for statistically significant effects of mode of election, policy leanings, and membership in a legislative faction. They skillfully draw on a range of evidence, including roll-call votes, popular election returns, organizational data, and the results of two detailed attitude surveys of Duma members. The result is a useful array of what might be called “stylized facts.” For example, party discipline was highest among the communists and was generally better explained by faction membership than by location along a Left-Right continuum.

The book is organized around a core of substantive chapters, each of which begins by setting forth a set of propositions. Various relevant roll-call votes and survey responses are then used to test the propositions. Those that survive become the book's stylized facts, such as proposition 2.2, which states that SMD deputies were more opposed to allowing parties to expel PR deputies who bolted their party. The authors could do a better job of tying the hypotheses to the literature on party control and mode of election. Some of this is done, but many propositions are introduced on the basis of their intuitive appeal, and the opportunity to draw wider connections seems not to be fully exploited.

Perhaps Smith and Remington should have noted more often the relationship between legislative politics and the executive in the Russian Federation. For example, they present convincing evidence that the parties on the Left tend to prefer more centralized parliamentary control that would allow the majority to legislate more easily, whereas the “reform” parties of the Right oppose such measures. Readers unaware of the proreform president's ability to bypass the legislature using his decree powers will find this affinity of proreform legislators for the status quo puzzling, even in an opposition controlled legislature. A modicum of extra background material would have made this important book more accessible to scholars who do not specialize on Russia.

The authors are conscientious about describing their data, and they use logit models that treat their explanatory variables, such as faction membership and mode of election, as “risk factors” for voting a certain way or expressing a certain opinion. This offers a useful window on the behavioral links in their data. There is a technical problem, however. We can think of the dependent variable in a logit model as taking on two values, “success” and “failure,” and the model then estimates the probability of success conditional on the values for the explanatory variables. For example, Smith and Remington analyze a vote taken in the Duma in January 1994 on whether to require any group of deputies who want to organize into a faction to have at least 50 members.

In this case, “success” corresponds to voting in favor of the amendment, “failure” to voting against it, and the explanatory variables include indicators of the faction to which a member belongs. In Table 2.2 (p. 41) the authors report that every member of the Liberal Democratic Party of Russia (LDPR) who participated in the roll-call vote was in favor, but no member of the Party of Russian Unity and Accord (PRES) voted for the amendment. An explanatory variable such as membership in LDPR that perfectly predicts success (in this case, a “yes” vote) corresponds to a logit coefficient of “positive infinity.” Likewise, a “no” vote corresponds to a logit coefficient of “negative infinity.” This is tantamount to removing all the LDPR and PRES members from the dataset before estimating the remaining coefficients.

A classic mistake is to include a variable such as LDPR or PRES, estimate the logit model via maximum likelihood, and stop the computer algorithm when the likelihood function ceases to register large improvements. The results will typically be an implausibly large but statistically insignificant positive coefficient estimate for the variables that perfectly predict “success” and an implausibly large negative coefficient estimate for variables that perfectly predict “failure.” This is just what we see in Table 2.3 (p. 43), and it is not an isolated event. The same problem arises at least seven other times in the same table and at various other points in the book (e.g., the “DPR” coefficient in all three columns of Table 2.5, the “NRP” coefficient in both columns of Table 5.6). The authors need to be sure that removing these hidden “perfect predictors” from their data and reestimating their model does not affect the remaining coefficient estimates.

Assuming that the substantive implications of this methodological problem are minor, the book hinges on the authors' energetic pursuit of stylized facts that link behavior with their taxonomy of member characteristics. Their approach has the advantage of posing concrete questions and finding answers: Do SMD deputies exhibit lower party discipline than their PR colleagues? “Yes,” barely. But the connection with members' motives is unresolved: Do the high rates of party discipline on roll-call votes mean that members join factions that reveal their policy preferences (Smith and Remington think so), or do they reflect party whipping (the authors observed no welts during their interviews with legislators)? Yet, this is a generic problem faced by any study that is primarily empirical. The theoretical controversies in legislative politics remain controversial because they tie in ambiguously to objective measures.

The authors make a serious effort to address these wider questions. Although they do not set any of the major controversies to rest, their results bring the workings of the Duma into clearer focus.