Interpreting small quantities of DNA: the hierarchy of propositions and the use of Bayesian networks

The dramatic increase in the sensitivity of DNA profiling systems that has occurred over recent years has led to the need to address a wider range of interpretational problems in forensic science. The issues surrounding questions of the kind "whose DNA is this?" have been the subject of considerable controversy but now it is clear that the emphasis is shifting to questions of the kind "how did this DNA get here?" Such issues are discussed in this paper and new insights are provided by two particular recent developments. First, the notion of the "hierarchy of propositions" that has arisen from a project called Case Assessment and Interpretation (CAI) that has been running in the British Forensic Science Service (FSS). Second, a technique for drawing inferences in the face of many interacting considerations, known as "Bayesian networks"--or "Bayes' nets" for short--that has been the subject of an earlier paper in this journal (1). The discussion is carried out by means of case studies, based on actual cases. It is clear that, whereas the inference in relation to the source of the DNA in a crime sample might be overwhelmingly strong, the inference in relation to the propositions that a jury must consider relating to the identity of the actual offender may be much more tentative.     

Do Juries Meet Our Expectations?

Surveys of public opinion indicate that people have high expectations for juries. When it comes to serious crimes, most people want errors of convicting the innocent (false positives) or acquitting the guilty (false negatives) to fall well below 10%. Using expected utility theory, Bayes' Theorem, signal detection theory, and empirical evidence from detection studies of medical decision making, eyewitness testimony, and weather forecasting, we argue that the frequency of mistakes probably far exceeds these tolerable levels. We are not arguing against the use of juries. Rather, we point out that a closer look at jury decisions reveals a serious gap between what we expect from juries and what probably occurs. When deciding issues of guilt and/or punishing convicted criminals, we as a society should recognize and acknowledge the abundance of error.     

PARENS PATRIAE

This article discusses the role of the court in protecting children in Australia and provides a means for comparing this process in other jurisdictions. It addresses the different perspectives of mental health and legal professionals in obtaining insight into case law and precedent in the English-speaking countries.     

Sudden death in hemodialysis patients

Hemodialysis patients may die suddenly and unexpectedly from a number of causes. These may be divided into those deaths due directly to and occurring during hemodialysis, those deaths occurring while the patient is not undergoing dialysis, and those deaths that may occur at any time. The first group includes brain herniation, air embolism, acute hemorrhage as a result of machine malfunction or fistula rupture, electrocution, cardiac arrhythmia caused by hypokalemia, complications of subclavian intravenous catheter insertion, third-degree heart block as a result of triglyceride emulsion, and disseminated intravascular coagulation (DIC) or hyperkalemia caused by overheated dialysate. The second group includes deaths due to pericardial tamponade because of effusion and suicidal causes of death (exsanguination, electrolyte imbalance as a result of excessive intake of salt, fluid, or potassium) as well as more conventional methods of suicide. The last category includes people dying of arteriosclerotic cardiovascular disease, hypertensive cardiovascular disease, and internal hemorrhage. Investigation of these deaths, including pertinent historical, laboratory, and autopsy data and investigation of dialysis equipment, is discussed.