Then-Senate Intelligence Committee Chairwoman Dianne Feinstein discusses a newly released Intelligence Committee report on the CIA’s anti-terrorism tactics, in a speech on the floor of the U.S. Senate in Washington, Dec. 9, 2014. (SENATE TV/REUTERS)

Last December, the release of the Senate Torture Report shocked the world. But although the C.I.A.’s use of torture had been more brutal and extensive than previously reported, researchers who study human rights were not surprised to see more evidence that a government—even a democratic one—frequently engages in torture. In fact, there is data suggesting that in 2011, of the 107 democracies in existence, 40 frequently engaged in torture and another 41 occasionally engaged in torture.

It’s troubling that these numbers remain so high despite the fact that reducing torture has been one of the primary goals of the modern human rights movement. Among other efforts, countries have been persuaded and coerced into enshrining a right against torture in their highest and most fundamental legal document: the constitution. In recent research, however, we show that constitutional torture prohibitions have failed to reduce torture.

The first human rights document to prohibit torture was England’s Bill of Rights Act of 1689, which stated that “excessive bail ought not to be required, nor excessive fines imposed, nor cruel or unusual punishment inflicted.” The graph below shows that the number of countries with constitutional torture bans has skyrocketed in the past three decades. Today, about 84 percent of the world’s constitutions prohibit torture.

Finding reliable data on torture rates around the world is the next hurdle when evaluating whether constitutional torture prohibitions have been effective. This is obviously a problem because, as the Senate Torture Report illustrates, governments aren’t exactly transparent about their torture practices.

Social scientists have developed datasets that try to measure the frequency of torture in every country in the world. The most widely used of these datasets was created by the CIRI Human Rights Data Project. The CIRI data uses annual reports written by the State Department and Amnesty International to code countries as either frequently engaging in torture (coded as “0”), occasionally engaging in torture (coded as “1”), and never engaging in torture (coded as “2’).

Christopher Fariss has developed an alternative measure based on a statistical model that produces a single estimate of each country’s annual level of repression from a variety of data sources (see here The Monkey Cage’s earlier use of that data). The data is measured in standard deviations, so a score of -2 (North Korea’s score in 2010) means that a country engaged in 2 standard deviations more repression than the average country, and a score of +2 (Japan’s score in 2010) means that the country engaged in two standard deviation less repression than the average country.

The figure below uses both datasets to show the average levels of torture from 1990—when torture bans really took off—until 2010. The red lines show the mean torture levels (and 90% confidence intervals) for countries without a constitutional torture ban, and the blue lines show the mean torture levels for countries with constitutional torture bans.


Remember, with both datasets, lower numbers translate into more torture and repression. This means that both datasets suggest that countries without constitutional torture bans have actually engaged in less torture over this time period.

Of course, this doesn’t mean that the constitutional torture prohibitions haven’t helped. After all, countries that adopted constitutional torture bans may have made progress reducing torture, but still just not caught up to countries without torture prohibitions. Unfortunately, the data suggests that isn’t what’s going on.

For countries that have added prohibitions to their constitutions, the figure below shows the average levels of torture in the 10 years before and after the adoption. Using the CIRI torture data—shown in the left panel—torture levels appear to have actually gotten slightly worse in the 10 years after constitutional torture prohibitions were adopted. Using the Fariss repression data—shown in the right panel—things are slightly better after the prohibition is adopted, but the improvement is nowhere near statistically significant.


Even though looking at average torture rates before and after adoption is an intuitive way to examine the effectiveness of prohibitions, it is also admittedly simplistic. This approach ignores a range of factors—like a country’s level of democratization, wealth, or civil strife—that are likely to effect levels of torture. In our academic work  we try to tackle the problem using a number of statistical methods as well as alternative data sources. The results are always the same: constitutional torture prohibitions haven’t reduced torture.

If constitutional torture prohibitions really have been ineffective, the obvious next question is why. Although we cannot definitely say, our research suggests some likely answers.

Constitutional rights can remain a dead letter because countries lack the capacity to enforce them. In some cases, constitutional commitments might even be adopted under false pretenses. The Constitution of North Korea, for example grants all citizens the “freedom speech, of the press, of assembly, demonstration and association,” but as the dustup over the theatrical release of “The Interview” illustrates, free speech isn’t a core value of the North Korean government. Indeed, empirical research suggests that a substantial portion of the world has what has been described as a sham constitution.

Even among countries that have otherwise strong constitutional traditions, like the United States, constitutional torture prohibitions may fail because the use of torture often enjoys substantial popular support, especially in the face of terror threats. For example, after the Senate Torture report was released, polls suggested that Americans were still supportive of the CIA’s use of torture by a 2-1 margin. Although one reason to enshrine a right in a constitution is to prevent the government from violating it whenever there is majority support for doing so, in practice things don’t always work out that way.

Of course, it’s possible that constitutional torture prohibitions have had a positive effect, but that effect simply hasn’t shown up in the existing (and admittedly flawed) data. It’s also possible that the constitutional prohibitions may make a difference if given enough time. For now though, the best empirical evidence suggests that if we want to reduce levels of torture around the world, we have to find a better way to do it than just amending constitutions.


Adam Chilton is an Assistant Professor of Law at the University of Chicago Law School. Mila Versteeg is an Associate Professor of Law at the University of Virginia School of Law.