Statistics and human rights
Presenter: Megan Price
When: Friday, July 15, 2016 Time: 11:00 AM - 12:30 PM
Room: Oak Bay 1-2 (Level 1)
Session Synopsis:
Statistics and Human Rights Research: Current and Historical Examples
Today's 'big data' climate has increased the demand for quantitative analysis across a wide spectrum of fields. Human rights research is no exception, and an increasing number of human rights reports now lead with statistical claims. However statistical analysis has been used to answer questions about human rights violations for decades. This presentation will provide a brief overview of several historical applications and will present detailed examples from projects in Guatemala and Syria. Work in Guatemala includes drawing a multi-stage probabilistic sample from millions of archived documents and presenting analytical results in two court cases. Ongoing work in Syria utilizes record linkage methods to identify victims reported across multiple sources and multiple systems estimation techniques to estimate the number of victims reported to none of the available sources.
Statistics and human rights
Presenter: Robin Mejia
When: Friday, July 15, 2016 Time: 11:00 AM - 12:30 PM
Room: Oak Bay 1-2 (Level 1)
Session Synopsis:
Estimating the prevalence of human rights abuses: child abductions from El Salvador's civil war
El Salvador suffered a brutal civil war between 1980 and 1992. A United Nations-sponsored Truth and Reconciliation Commission formed in 1994 reported that 75,000 people were killed and perhaps another 75,000 were disappeared, out of a population of roughly 5 million. Here, we examine a type of disappearance not specifically quantified by the commission: the abduction of children during military operations. We assess data provided by the Salvadoran NGO Asociación Pro-Búsqueda de Niñas y Niños Desaparecidos (Pro-Búsqueda), which investigates these cases, and present a characterization of known cases. Pro-Búsqueda has located over 389 children and has over 500 open cases, with new cases still being opened each year. Most of those located so far have been found alive. Some were found as youth in orphanages, some were raised in military families, and others were sent abroad for adoption. There are likely many young adult adoptees of Salvadoran origin a living in the US and other countries today who have living biological family searching for them. However, the full scope of abductions during the war remains an open question. I will discuss methods used to develop prevalence estimates in human rights investigation and present initial estimates of the prevalence of such child abductions from a capture-recapture analysis, with a discussion of the non-trivial challenges that arise in applying the method to this dataset.
Statistics and human rights
Presenter: Nicholas Jewell
When: Friday, July 15, 2016 Time: 11:00 AM - 12:30 PM
Room: Oak Bay 1-2 (Level 1)
Session Synopsis:
Counting Civilian Casualties--Statistics and Human Rights Problems
Civilian casualties are increasingly in the news as conflicts erupt at various regions of the globe. There is often confusion around the numbers of civilians who have been killed, particularly during the earliest stages of conflicts. Sometimes the range of estimates can cover several order of magnitudes, even decades after conflicts have ended, and thus estimates and claims can be hard to interpret especially when they are provided by advocacy groups on either side of a conflict. A related, but quite different human rights problem arises from the need to assess data on the number and patterns of child abductions and disappearances that occurred during El Salvador's civil war (approximately from 1978-1992). The need for accurate counts of civilian casualties (and other human rights events), particularly deaths, turns out to be a relatively modern phenomenon that has nevertheless attracted considerable scientific and political attention. I will discuss some historical background for casualty counts and note three current techniques that have been typically employed. These methods range from traditional demographic and epidemiological survey techniques, to multiple systems estimation that uses capture-recapture models, to attempts to provide full documentation including the modern technological approaches through crowdsourcing.
Statistics and human rights
Presenter: Alicia Carriquiry
When: Friday, July 15, 2016 Time: 11:00 AM - 12:30 PM
Room: Oak Bay 1-2 (Level 1)
Session Synopsis:
Statistics and forensic evidence: The challenges that lie ahead
In the United States courts, persons acussed of committing a crime are presumed innocent until proven guilty. The onus is on the prosecution, whose job it is to present evidence in support of the hypothesis that the suspect is guilty. The fair administration of justice requires that evidence presented in court be properly discussed and interpreted. In particular, evidence must be evaluated and interpreted under the two competing hypotheses posed by the defense and by the prosecution. A natural summary is the likelihood ratio (LR), defined as the probability of observing the evidence under the prosecutor's hypothesis divided by the probability of observing the evidence by chance alone, i.e., under the defense's hypothesis that someone other than the suspect was the source opf the evidence.Large values of the LR would tend to support the prosecution. We discuss some of the challenges that arise in the calculation of the LR for different types of evidence and contrast those with the case of DNA, where LRs have been in use for years. In addition to the appropriate statistical models required to calculate a LR, data to fit those models are lacking.
