By Richard Carver, Senior Lecturer in Human Rights and Governance, Oxford Brookes University
Richard Carver has more than 30 years of experience as a human rights researcher, working for Amnesty International, Human Rights Watch, ARTICLE 19, and a number of UN agencies. He has published extensively on national human rights institutions, particularly on criteria and techniques for measuring performance, impact and effectiveness. Originally an Africa specialist, his current work focuses on southeastern Europe. He holds a PhD in Human Rights and an LLM in International Law, both from Oxford Brookes University, where he is Senior Lecturer in Human Rights and Governance.
Does torture prevention work? The title of our new book presents a very simple research question but one that, oddly enough, has never really been addressed before. In this post and two subsequent ones, I will present the findings of a four-year research project directed by Dr Lisa Handley and myself with a team of more than 20 researchers.
Of all human rights, the right not to be subjected to torture appears to be the most comprehensively protected, with a web of obligations on states that go far beyond the simple prohibitions contained in the International Covenant on Civil Political Rights, the regional instruments and a number of specialized treaties. At the global level the UN Convention Against Torture and its Optional Protocol set out a series of very specific measures related to the investigation and prosecution of torturers and the monitoring of places of detention and imprisonment. The Body of Principles for the Protection of All Persons under Any Form of Detention or Imprisonment, while they lack the binding force of the treaties, set out a series of norms aimed at protecting detainees and prisoners from torture and other ill-treatment. These standards are constantly supplemented and updated by the recommendations of treaty bodies and others.
But do any of these norms actually work? If a state adopts the definition of torture contained in Article 1 of the UNCAT, does this reduce the incidence of torture? If it establishes an independent visiting mechanism, as required by the OPCAT, are prisoners or detainees any better protected?
Most of these measures (we tested 66 of them) are based upon common sense. Yet some scholars have argued that ratification of the anti-torture instruments can actually lead to an increase in the incidence of torture. So it cannot be taken as self-evident that the good intentions behind anti-torture norms will actually lead to positive consequences.
We conducted this research at the request, and with the funding of, the Association for the Prevention of Torture. Because APT clearly has its own strong interests in the field, we only agreed to undertake the project with the guarantee that the methodology, conduct and outcomes of the research would be completely independent from the APT – a guarantee that the organization was happy to give.
From the outset, it was apparent to us that we would need to combine the insights offered by qualitative and quantitative research methods. Much political science literature on human rights aspires to present a global picture using quantitative methods, but displays little understanding of the complexity of realizing human rights in practice. On other hand, most of the writing by scholars strongly favouring human rights proceeds from detailed examples and case studies, often failing to show how findings can be generalized.
We chose to conduct 16 case studies of countries reflecting as much geopolitical variety as we could encompass. While 16 is a fairly large number for qualitative comparative research, it is a very small number for quantitative analysis. However, we chose to review the experience of our 16 countries over a 30 year period (1985-2014). By making our unit of analysis the country-year (each individual country in each individual year) we were able to have enough data points (16 x 30 = 480) to make statistical analysis possible.
One of the first challenges we faced was how to measure the incidence of torture. Torture is almost invariably carried out in secret and states do not keep records. Or rather, to complicate matters, states begin to document torture cases at the point when they develop a commitment to eradicate it – a phenomenon known as the “human rights information paradox.” This means that official statistics, while they are useful source material, cannot constitute the only measure of the incidence of torture.
Most scholars who have required a measure of the incidence of torture have used one of the existing indices (such as the Cingranelli-Richards or CIRI index or the Political Terror Scale). Or they have devised their own index, as did Oona Hathaway in her important work on the impact of treaty ratification. The problem is that all these indices are founded upon content analysis of texts (the US State Department human rights reports or Amnesty International reports), not upon primary research into the incidence of torture. We concluded that there were simply too many erroneous changes in the reported incidence of torture in all these indices.
Instead we devised our own measure: the Carver-Handley Torture Score (CHATS). The CHATS is based upon a combination of primary and secondary research in the countries in our study. While contemporaneous annual human rights reports are an important source for us too, the information in these is cross-checked and triangulated with other documentary sources and through interviews.
We used similar research methods to assemble our explanatory variables – that is, the various preventive measures. Since there were so many of these, we grouped them into four indices: detention safeguards, the investigation and prosecution of torturers, monitoring mechanisms, and complaints mechanisms. Each of these in turn was divided into law and practice indices – one reflecting law on the statute book (or treaties ratified), the other actual practice in the police station or prison.
While we were able to perform bivariate regressions with each of the 66 variables separately, for a multivariate regression we could only use the composite indices (along with a number of control variables, such as democracy, economic development, and conflict).
The headline finding of the study is very clear. The preventive mechanisms that had the greatest impact were the safeguards implemented at the moment of arrest and soon afterwards. Steps such as taking a detainee to a lawful place of detention, notifying friends and family, and allowing prompt access to a lawyer and medical doctor were the most important in preventing torture. (It should be underlined that this refers to actual practice, not the mere existence of laws granting such rights, an issue that will be explored in a later blog entry.)
The other three sets of preventive mechanisms were of varying importance. Investigation and prosecution of torturers has a positive impact – again with the proviso that this refers to actual practice, not the mere existence of laws outlawing torture. Monitoring mechanisms also appear important in both prisons and police stations. The one set of measures that has no detectable preventive effect, according to our data, is complaints mechanisms.
In the next blog post I will discuss our findings in greater detail.