About

This website provides data and information about several human rights measurement projects by Christopher J. Fariss and Keith Schnakenberg. The Human Rights Protection Scores section contains information about latent human rights variables estimated using a new Bayesian measurement model described in two papers (note there are 2 versions of this data). The Human Rights Dependence Scores section contains information about a network analytic approach that measures the complex pattern among human rights violations. Each project section contains links to the data, replication files, and the paper that describes the data and measurement process in detail. The projects rely on publicly available data generated by the CIRI Human Rights Data Project, the Political Terror Scale, the Ill Treatment and Torture Data Collection, the Uppsala Conflict Data Program, and several other published sources. The documentation below contains more details about these sources. The data is hosted at the Dataverse. Please contact us if you have any questions.

top

Protection Scores


Version 2

The Latent Human Rights Protection Scores (version 2) available in this section were generated using a dynamic ordinal item-response theory model discussed at length in the two papers listed below. The newest version of this model is also dynamic in the treatment of the item-difficulty cut-points of the latent variable model (labeled the Dynamic Standard Model). This parameterization allows the model to account for systematic changes to the human rights country reports published annually by the US Department of State and Amnesty International. These documents are used to code standards based human rights variables by several different coding projects. According to these indicators, human rights practices have not improved over the past 35 years, despite the spread of human rights norms, better monitoring, and the increasing prevalence of electoral democracy. This empirical pattern is not an indication of stagnating human rights practices. Instead, it reflects a systematic change in the way monitors, like Amnesty International and the US State Department, encounter and interpret information about abuses.

The new Latent Human Rights Protection Scores (version 2) demonstrate that respect for human rights has improved over time. Country-year datasets in csv format with posterior mean and standard deviation estimates for the version 2 data are publicly available. Information about the earlier version of this model and data are available in the version 1 section below. The JAGS code used to estimate the models are available in the Appendix section of the article. Other JAGS examples are available on the Code section of this website. Please contact us if you have any questions.

Citations

Version 1

The Latent Human Rights Protection Scores (version 1) available in this section were generated using a dynamic ordinal item-response theory model discussed at length in the paper listed below. The datasets contain country-year estimates of the mean and standard deviation of the estimated level of respect for physical integrity rights and empowerment rights for each year in which the constituent CIRI human rights indicators were available. See the CIRI Human Rights Data Project for more information on these indicators. The latent estimates allow for measurement uncertainty to be included in models that use human rights as a predictor variable. Datasets that contain the last 5000 draws from the posterior distribution from both the dynamic ordinal item-response theory model and the ordinal item-response theory model are also available. Detailed information about these models and other scaling techniques are discussed in the paper. We have also posted full replication files to our Dataverse page. The JAGS code used to estimate the models are available in the Appendix section of the article. Other JAGS examples are available on the Code section of this website.

Country-year datasets in csv format with posterior mean and standard deviation estimates for the dynamic latent physical integrity variable and dynamic empowerment variable are now available. We have also added binary Rdata files that contain the same variables as the csv files in addition to 5000 posterior draws for each country-year estimate. The data are publicly available, as are the replication files. Please contact us if you have any questions.

Citations

top

Data & Code

Data


This version of the latent physical integrity variable accounts for the changing standard of accountability, is based on 13 indicators of repression, and covers the period 1949-2013. We recommend that researchers use version 2 for applied research of physical integrity rights.

This version of the latent physical integrity variable is based on the 4 CIRI physical integrity indicators and covers the period 1981-2010. Also included here is the latent empowerment rights variable, which is based on the 7 CIRI empowerment rights indicators and covers the period 1981-2010.

Code

This subsection contains annotated R scripts that demonstrate how to use the data from the projects described in the Human Rights Protection Scores section and the Human Rights Dependence Scores section. There are also R scripts that demonstrate how to estimate models using JAGS. We have also added several IRT simulations for those interested in estimating their own latent variables and links to online resources for using R and JAGS. We plan to post many new examples. Please contact us if you have any questions.

R Code for Implementing Human Rights Protection Scores

R Code for Implementing Human Rights Dependence Scores

R Code for Implementing JAGS Models

Additional R Code Examples

Resources for Learning and Using R and JAGS

top

Dependence Scores

The Human Rights Dependence Scores available in this section were generated using a network analytic method described in detail in the paper below. The variables are estimates of the level of mutual dependence between one specific level of human rights abuse and all of the other levels of abuses measured as part of the CIRI Human Rights Data Project. The data are useful for testing hypotheses that are focused on one level of one right while accounting for the mutual dependence of the other right levels to that specific right level of theoretical interest. Detailed information about the network model and data are discussed in the paper. We do not plan on updating these estimates. Please contact us if you have any questions.

Citation

Christopher J. Fariss and Keith Schnakenberg. 2014. Measuring Mutual Dependence Between State Repressive Actions. Journal of Conflict Resolution 58(6):1003-1032 (September 2014).

top

Contact

Keith Schnakenberg
keith.schnakenberg@gmail.com
Assistant Professor
Martin School of Public Policy and Administration
University of Kentucky
415 Patterson Office Tower
Lexington, KY 40506

Christopher J. Fariss
cjf0006@gmail.com
Assistant Professor
Department of Political Science
The Pennsylvania State University
227 Pond Lab
University Park, PA 16802

top