Thursday's topics at the statistical world conference will include the use and understanding of statistical reasoning in other areas of society. The topic will be examined both from the point of health care and judicial systems. Especially health risks are being increasingly explained by means of statistical inference. These risks are not easy for a layperson to understand and an isolated random variation may be sufficient to generate anxiety about the effects of environmental factors.
The statistical nature of health risk factors also produces problems in courts of law. Statistics are being increasingly used as evidence in legal cases. This has become particularly salient with cases relating to the ill effects of smoking. Researchers Mary W. Gray and Nawar Al-Shara, from the American University, Washington DC, USA, examine in their paper, presented by Hassan Hamdan, the attitudes to the admission of statistical evidence in the US federal courts of law. The fundamental problem is that judges and juries do not have sufficient expertise in statistics. This may either lead to the rejection of relevant statistical research or, conversely, the acceptance of "junk" science as evidence in court.
US federal courts have naturally been highly cautious about admitting statistical inference as evidence. Sometimes a criterion has been that the pertinent research has been published in a scientific journal.
The problems in bringing together the legal and statistical ways of reasoning are, however, deep-rooted. Few judges question the statistics which show that killing a white person is more likely to bring a death penalty than killing a black person. Yet, this statistical fact has not influenced the legal decision-making of courts because, in the judges' view, evidence of the bias must be proven in each individual case in hand.
The two researchers conclude their paper by giving pointers for a general code of ethics statisticians should adopt in courtroom presentations, as well as in other statistical dealings with non-experts in statistics: one must define the accuracy of the results obtainable with the employed methodology, whether all factors affecting outcomes have been controlled for and whether the possible outcomes can be distinguished from each other. Good old-fashioned honesty is also a requisite.
A total of 175 presentations will be made at the conference on Thursday, on topics ranging between statistics in telecommunication, population, agricultural and enterprise censuses, statistics in pharmaceutical industry, function of central banks in the field of statistics, statistical education, customer needs and statistics, history of the field of statistics, how to measure deregulation, the information society, sampling and enterprise statistics.
Friday's sessions will cover such topics as forest inventories, human rights in the light of statistics, how to measure poverty and gender equality in statistics.
Further information about the conference programme:
Jussi Melkas, tel. +358 9 1734 3200;
Virpi Viertola, tel. +358 50 568 7444;
Kristiina Niklander, tel. +358 50 373 6510
Short English abstracts at: http://www.stat.fi/isi99/proceedings.html