Parapolitics and Roll-Call Voting in Colombia: A Bayesian Euclidean and Spherical Spatial Analysis
Juan Sosa, Universidad Nacional de Colombia, Colombia · 2025
Read Parapolitics and Roll-Call Voting in Colombia: A Bayesian Euclidean and Spherical Spatial Analysis by Juan Sosa, Universidad Nacional de Colombia, Colombia (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-22045.txt.
Introduction
Juan Sosa, Carolina Luque, and Juan Valero map Colombian Senate votes in space to expose a scandal's pull in a 2025 paper. The parapolitics scandal tied lawmakers to paramilitary groups. Roll calls from 2006 to 2010 form a new dataset. It holds 147 senators and 136 plenary votes.
Two spatial models estimate each senator's ideal point. One is a flat Euclidean map, the other a circle. Both find alignment is not the classic left-right split. Opposition versus non-opposition sorts the votes best instead. Ideal points link significantly to scandal involvement. The tie is robust across their model checks. Illegal influence seems wired into voting patterns then.
Catalog record
| CREATOR | Juan Sosa, Universidad Nacional de Colombia, Colombia |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-22045
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Parapolitics and Roll-Call Voting in Colombia: A Bayesian Euclidean and Spherical Spatial Analysis Juan Sosa, Universidad Nacional de Colombia, Colombia Abstract This study presents a Bayesian spatial voting analysis of the Colombian Senate during the 2006–2010 legislative period, leveraging a newly constructed roll-call dataset comprising 147 senators and 136 plenary votes. We estimate legislators’ ideal points under two alternative geometric frameworks: a traditional Euclidean model and a circular model that embeds preferences on the unit circle. Both models are implemented using Markov Chain Monte Carlo methods, with the circular specification capturing geodesic distances and von Mises-distributed latent traits. The results reveal a latent structure in voting behavior best characterized not by a conventional left-right ideological continuum but by an opposition–non-opposition alignment. Using Bayesian logistic regression, we further investigate the association between senators’ ideal points and their involvement in the para-politics scandal. Findings indicate a significant and robust relationship between political alignment and para-politics implication, suggesting that extralegal influence was systematically related to senators’ legislative behavior during this period. Keywords: Euclidean spatial voting model; Spherical spatial voting model, Markov Chain Monte Carlo Methods; Para-politics; Roll-Call data. 1 Introduction Spatial voting models provide a powerful framework for analyzing legislative behavior by positing that political actors make decisions based on their proximity to ideological alternatives within a latent policy space ( Luque and Sosa, 2024 ) . The central object of interest in these models is the ideal point , a latent trait representing an individual’s political preference, estimated from observed voting behavior across a set of motions. These models have been extensively applied in the analysis of parliaments, senates, courts, and councils across diverse institutional contexts ( Clinton et al., 2004 ; Poole, 2005 ; Curini, 2020 ) . Among them, Bayesian formulations offer a flexible and probabilistically coherent approach for ideal point estimation, allowing for principled uncertainty quantification and the incorporation of prior information ( Jackman, 2001 ; Clinton et al., 2004 ; Luque and Sosa, 2023 ) . In this study, we advance the application of spatial voting models by conducting a detailed Bayesian analysis of roll call data from the Colombian Senate during the 2006–2010 legislative period, a term marked by profound political realignment and the emergence of the parapolitics scandal , which revealed ties between legislators and illegal paramilitary groups ( Ávila and Velasco, 2012 ; Bakiner, 2020 ; Pardo, 2020 ) . Building upon the methodology of Luque and Sosa, 2023 , who applied Bayesian spatial models to the Colombian Senate for the 2010–2014 period, we extend the analysis in both statistical and substantive dimensions. From a political science perspective, we provide one of the first systematic mappings of latent political preferences during a critical period of institutional transformation in Colombia. From a statistical perspective, we compare two distinct geometric formulations of spatial voting models: The traditional Euclidean model and its circular generalization, the Spherical Latent Factor Model proposed by Yu, 2020 . A key contribution of this study lies in the construction of a comprehensive roll call database comprising 147 senators and 136 plenary votes from the 2006–2010 legislative period, enriched with detailed information on political affiliations, replacements, and each legislator’s involvement in the parapolitics scandal. Using this dataset, we implement a one-dimensional Bayesian ideal point model under two distinct geometric frameworks. The first assumes a linear ideological continuum embedded in Euclidean space and is estimated via a probit formulation with normal priors. The second framework replaces Euclidean with geodesic distances by embedding ideal points on the unit circle, modeling latent positions with von Mises priors. This circular formulation addresses limitations of the Euclidean model by capturing potential behavioral convergence among ideological extremes and correcting distortions that arise from projecting complex political behavior onto a linear space ( Tversky, 1969 ; Spirling and McLean, 2007 ) . To ensure comparability across geometries, we align and project circular estimates onto the tangent space at a fixed reference point, enabling direct comparison with Euclidean ideal points. We also address model identifiability through anchoring and post-processing procedures, and handle missing voting data via complete case analysis in the Euclidean model and posterior imputation in the circular model, verifying the robustness of results across both strategies. The estimated ideal points reveal a latent structure not reducible to a conventional left–right ideology. Rather, both geometric formulations uncover a broad opposition–non-opposition divide, with opposition, independent, and minority senators clustering separately from the dominant governing coalition. This non-ideological polarization reflects structural imbalances in the Senate and aligns with findings from prior work on Latin American legislatures ( Londoño Rodríguez et al., 2014 ; Luque and Sosa, 2023 ) . Moreover, we explore the relationship between estimated ideal points and involvement in the parapolitics scandal by fitting logistic regression models to posterior samples of ideal points from both models. In both cases, we find that more extreme and coalition-aligned ideal points are significantly associated with parapolitics involvement. This study contributes to political methodology by demonstrating the importance of geometric assumptions in latent trait modeling, showing that circular spatial voting models offer greater flexibility and robustness in fragmented or non-ideological political systems. It also provides novel insights into Colombian legislative dynamics during a period of exceptional institutional stress, offering empirical evidence that patterns of scandal involvement were not ideologically neutral but systematically aligned with the dominant coalition. The structure of this paper is as follows: Section 2 introduces the spatial voting models used to estimate ideal points. Section 3 describes the construction of the roll call database. Section 4 provides a descriptive analysis of the Colombian Senate during 2006–2010, with emphasis on the parapolitics scandal. Section 5 outlines the computational implementation of the models. Section 6 reports the main results, including ideal point estimates, comparisons between models, and associations with parapolitics involvement. Finally, Section 7 presents the main conclusions and outlines avenues for future research. 2 Bayesian estimation of ideal points This section introduces two spatial voting models used to estimate legislators’ ideal points from roll call behavior: the Euclidean model and its circular extension. The Euclidean model assumes preferences lie in a low-dimensional linear space, with voting driven by proximity to ideological alternatives. While widely applied, this approach can misrepresent cases where ideological extremes behave similarly. To address this, we also consider a circular model that embeds ideal points on the unit circle, offering a more flexible geometric framework. We present the formulation and inference procedures for both models, emphasizing their assumptions and key differences. 2.1 Euclidean model We consider n actors voting on m motions, where each vote represents a choice between two alternatives: Aye and Nay. The vote of the i -th actor on the j -th motion is denoted by y_{i,j} , with i=1,\ldots,n and j=1,\ldots,m , where 1 indicates a vote in favor and 0 indicates a vote against. The voting alternatives, \mathbf{p}_{j} (in favor) and \mathbf{q}_{j} (against), for the j -th motion are represented as points in \mathbb{R}^{d} , with the goal of obtaining a low-dimensional geometric representation often referred to as a spectrum or political space ( Jackman, 2001 ; Luque and Sosa, 2023 ; Lipman et al., 2023 , see) . We assume that actors are more likely to choose the option closest to their political preference