The Influence of the President and Government Coalition on Roll-Call Voting in Brazil, 2003–2006
| DOI | 10.1177/1478929920904588 |
| Published date | 01 May 2021 |
| Author | Tsung-han Tsai |
| Date | 01 May 2021 |
| Subject Matter | Articles |
https://doi.org/10.1177/1478929920904588
Political Studies Review
2021, Vol. 19(2) 193 –208
© The Author(s) 2020
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/1478929920904588
journals.sagepub.com/home/psrev
The Influence of the President
and Government Coalition
on Roll-Call Voting in Brazil,
2003–2006
Tsung-han Tsai
Abstract
In Brazil’s legislative process, political exchanges between the government and legislature is an
essential feature. This article focuses on the role of the president and political parties in Brazil’s
national legislative process. Because nonideological factors influence voting, roll calls do not suffice
for estimation of legislators’ policy preferences. In this article, we derive a spatial model of voting
in which voting behavior is induced by both ideological motivations and coalition dynamics and
develop a multilevel ideal-point model implied by the spatial voting model. After the proposed
model is applied to the analysis of roll-call votes in the Brazilian Chamber of Deputies between
2003 and 2006, coalition dynamics is found to influence the voting behavior of legislators. We
also confirm the finding in previous studies that the ideological alignment of political parties in the
legislature contrasts with the perceived positions.
Keywords
legislative politics, spatial model of voting, roll-call votes, Brazil, Bayesian inference, item
response theory
Accepted: 16 January 2020
Introduction
The framework of spatial models has been widely applied to explain numerous aspects
of the political process. Interpretations of the American Congress in the context of spa-
tial models (e.g. Cox and McCubbins, 1993, 2005; Krehbiel, 1998) inspired theories of
lawmaking, and those theories were then applied to cross-national studies (e.g. Carey,
2009). The empirical implications of these theories can be tested; one essential require-
ment for operationalization is the development of measurements of political actors’ posi-
tions in policy space or ideological space. For instance, in the study of legislative politics,
Department of Political Science, Election Study Center, Taiwan Institute for Governance and
Communication Research, National Chengchi University, Taipei, Taiwan
Corresponding author:
Tsung-han Tsai, Department of Political Science, Election Study Center, Taiwan Institute for Governance and
Communication Research, National Chengchi University, Taipei 11605, Taiwan.
Email: thtsai@nccu.edu.tw
904588PSW0010.1177/1478929920904588Political Studies ReviewTsai
research-article2020
Article
194 Political Studies Review 19(2)
many analysts statistically analyze roll-call votes—the recorded votes in legislatures—
to measure the policy preferences of legislators, which are called ideal points (e.g.
Clinton et al., 2004; Poole and Rosenthal, 1997). Despite its merits, however, the princi-
pal problem of using roll-call data is that voting records are usually a consequence of
political bargaining and, thus, they may not accurately reveal legislators’ preferences
(Clinton, 2012; Krehbiel, 2000).
In Brazil’s Chamber of Deputies, typical legislative behavior can be as much a product of
political negotiation as it is of ideological or policy concerns (Ames, 2002; Desposato and
Scheiner, 2008). The legislative powers of the executive authorized by the 1988 constitution
allow the president considerable control over the political resources on which legislators
depend for their political survival. With these powers, the president can exchange political
favors for legislative support by strategically forming coalitions (Alston and Mueller, 2006;
Amorim Neto, 2006; Pereira and Mueller, 2004a). Although they hold different points of
view, leading scholars of Brazilian politics agree that legislative behavior is structured by
coalition dynamics (Ames, 2002; Amorim Neto, 2002a; Figueiredo and Limongi, 2000).
When ideology is not the only factor, government–opposition dynamics drive voting behav-
ior in the legislature; standard ideal point estimates do not distinguish the impact of coalition
dynamics from individual policy preferences (Zucco, 2009; Zucco and Lauderdale, 2011).
In this article, we show how item response theory (IRT) can be used to set item-diffi-
culty parameters in the framework of multilevel modeling; these calculations can recover
estimates of the policy preferences of legislators and to test the government–opposition
effect on legislative behavior. The logic underlying this modeling strategy is as follows:
Assuming that the underlying political problem has one-dimensional ideological content,
two legislators can make different voting decisions even if they hold the same policy
position. The difference between voting decisions is systematically explained by the gov-
ernment–opposition status through changing the “difficulty” of roll calls confronted by
legislators. The proposed model was applied to the analysis of roll-call votes in the
Brazilian Chamber of Deputies between 2003 and 2006 to estimate policy positions of
parties as well as to investigate the government–opposition effect. According to the
results of the analysis, the dynamics of coalition formation indeed were influential for the
voting behavior of legislators. Moreover, the results indicated that the ideological align-
ment of political parties in the legislature contrasts with the perceived positions.
Uncovering whether Brazilian presidents construct coalitions in the Congress and rule
through them is important not only for the enhancement of our knowledge of executive–
legislative relations but also from the perspective of democratic representation. This arti-
cle illustrates party-based voting behavior and the role of the president in the legislative
process and the representation of voters’ preferences. Therefore, by explaining the role of
parties in executive–legislative relations, this article contributes to several themes in this
field, including party influence in legislatures (Aldrich, 1995; Cox and McCubbins, 1993;
Krehbiel, 1999; McCarty et al., 2001), patterns of governance in presidential systems
(Linz and Valenzuela, 1994; Shugart and Carey, 1992), and the role of parties as the pri-
mary vehicle of representation (Kitschelt, 2000; Sartori, 1976).
The remainder of this article proceeds as follows. Section “Presidential Powers and
Legislative Behavior in Brazil” reviews the extant literature on legislative politics in
Brazil. Section “Ideal-Point Model” presents a spatial model of voting in which voting is
induced by both ideological motivations and coalition dynamics and develops a Bayesian
random item-difficulty ideal-point model. Section “Roll-Call Data Analysis of Brazil”
illustrates the data, model specification, and analysis of roll-call voting in Brazil. Section
“Conclusion” concludes the article.
To continue reading
Request your trial