Measuring presidential centrism and its effect on repression: does ideology influence whether democratic governments respect human rights?

ABSTRACT

If presidential runoff elections promote high human rights respect (Holzer 2018), and presidential runoff elections also promote the election of centrists (Haan and Vokerink 2001; Brusco, Dziubiski, and Roy 2012), are centrist presidents more likely to protect human rights? This research note aim to answer this question by analysing an original measure of presidential centrism for 14 democracies spanning from 1990 to 2011. By finding that there is a greater probability of an increase in government respect for human rights (and lower probability of a decrease in government respect for human rights) when the president is ideologically closer to the median voter, I reemphasise the ‘impact of ideology on human rights practices’ (Yazici 2018: 12).

KEYWORDS:

• Centrism
• Presidentialism
• Human Rights
• Ideology
• Repression

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. The case could be made that spatial voting actually begins with Hotelling (1929). He argues that if two companies were given a choice of location on a straight line evenly dispersed with customers, it is rational for both companies to locate at the exact middle. Hotelling (1929, 54) suggests that the same intuition applies to political campaigning, as ‘each party strives to make its platform as much like the other as possible’ so as to win over as many voters as possible.

2. While explaining repression in non-democracies is certainly of importance, repression that occurs within a non-democracy is not particular surprising. Like Holzer (2018), I am interested in why democracies, which nominally reflect the will of the people (and therefore, should be less likely to repress their people), still resort to repression. While democracies are generally more inclined to protect human rights and limit the use of repression vis-à-vis non-democracies (e.g. Poe and Tate 1994; Davenport 1995; Fein 1995; Krain 1997; Davenport 1999; Zanger 2000; Harff 2003; Davenport and Armstrong 2004; Poe, Tate, and Keith 1999; Richards, Webb, and Clay 2015), repression still occurs. Why is this the case? By limiting myself to presidential democracies, I, like Holzer (2018), can clearly establish empirical boundaries for this study, and thereby focus on differences within democratic presidential systems, rather than between democratic presidential systems and other democratic forms of government.

3. As mentioned above, this research note only looks at physical integrity rights, which are a subset of ‘human rights’. The primary reason for this decision is that I hope to build upon the body of literature that has sought to predict physical integrity rights abuse. For instance, as alluded to above, Bueno de Mesquita et al. (2003, 351) find that ‘[s]ystems with large winning coalitions engage in substantially less oppression than those with small winning coalitions’. Although they refer to human rights broadly, Bueno de Mesquita et al. (2003) only look at physical integrity rights. In a follow-up piece also mentioned above, Holzer (2018) finds that presidents elected by a runoff election (which require a large winning coalition) are less oppressive that those elected after only one round (as those winning coalitions can be comparatively smaller). Again, while he refers to human rights broadly, Holzer (2018) only looks at physical integrity rights, in particular the CIRI Physical Rights Index. Recall that the theoretical argument I outline in Figure 1 suggests that centrist presidents promote government respect for human rights because presidential runoff elections (which promote the election of centrist presidents) also promote government respect for human rights (per Holzer 2018). Given that Holzer (2018) uses the CIRI Physical Integrity Rights Index, and given that this research note explicitly relies on the findings of Holzer (2018) as a crux of its theoretical argument (see Figure 1), it seems most appropriate for this research note to also use the CIRI Physical Integrity Rights Index. An implication of this decision, however, is that the conclusions of this study may or may not extent to all human rights. Physical integrity rights cannot (and should not) be viewed as a proxy for all human rights. With that said, an excellent avenue for future research would be to apply the methods of this study to the examination of any possible relationship between centrism and alternate measures of human rights, such as empowerment rights (for which CIRI also happens to have an index).

4. For the remainder for this research note, all references to ‘human rights’ specifically refer to the four human rights measured by the CIRI Physical Integrity Rights Index (i.e. torture, summary execution, disappearance and political imprisonment).

5. Like Yazici (2018), I assume that the ideology of the political party that presidential candidates belong to is an accurate reflection of the ideology of the presidential candidates themselves.

6. Essentially I use the following formula: M = L + (50 – C) / F × W, where M is the median voter’s estimated ideology score, L is the lower end of the ‘interval’ containing the median, C is the cumulative vote share up to but not including the ‘interval’ containing the median, F is the vote share of the ‘interval’ containing the median and W is the width of the ‘interval’ containing the median. Each party’s ‘interval’ is calculated as the midpoint between that party and the one immediately left and right of it per simple Euclidean preference relations (i.e. voters choose the candidates/parties that are closest to them).

7. For obvious reasons, any elections where the president was any independent were omitted from my analysis, as independents do not have an ideology score (which are needed to estimate the position of both the median voter and the president’s distance from the median voter).

8. Note that the manifesto data that I use to compute these values (Lehmann et al. 2017, Volkens et al. 2017) does not currently cover any political parties in Africa, and that coverage is limited for the Americas. However, using the method I have outlined above, expanded coverage for the president’s distance from the median voter can be computed once new data becomes available.

9. Like Yazici (2018), I assume that the ideology of the president is constant throughout the duration of each elected term.

10. I chose this conflict variable (among others) as it was used by Holzer (2018).

11. Revisiting Figure 2, a keen reader might notice that Mexico, the state with the greatest president’s distance from the median voter value (in 2006), happens to be a state with a great many human rights violations. Note that in order to ensure that the ‘extreme’ is not driving Model l’s results, I have sought to control for this outlier by re-running my analysis absent years that Mexico’s 2006 president was in office, the results of which are reported as Model 2. As you can see, my results are robust to this alternate specification, given the strong similarities between the coefficients (and respective statistical significances) of the regressions in Model 1 versus Model 2.

12. Recall that my dependent variable (government respect for human rights) is categorical, and as such, you cannot interpret any of the probabilities corresponding to values between categories (i.e. you cannot use Figure 3 to determine the probability of a non-integer change in the dependent variable, such as a −1.5 change or +1.5 change).

Additional information

Notes on contributors

Joshua Holzer

Joshua Holzer’s current research interest lie at the intersection of electoral systems and government respect for human rights. His work has been published in Research & Politics and The International Journal of Human Rights.