A dictionary definition of ethical behaviour is behaviour that is ‘morally correct’. As such, it is inevitable that value judgements are involved in deciding whether a particular research study is or is not ethical. Also, it is inevitable that in some cases the decision will depend upon the cultural setting in which the study takes place and that it will sometimes happen that there is a conflict between different ethical principles. For these reasons, what is important is not so much to try to establish inflexible rules of ethics that researchers should follow, but instead to have some general guidelines about what to take into account when research plans are being assessed.

A number of professional bodies have devised codes of ethics for their members. To a large extent the present chapter is based on one of these, that is the International Statistical Institute's declaration on professional ethics (International Statistical Institute, 1986). This declaration recognizes four types of obligation that the statistician has in collecting and processing information: to society, to funders and employers, to colleagues, and to subjects. This chapter considers each of these obligations in turn, on the assumption that the ethical considerations that apply to a statistician involved in a research study apply equally well to researchers from other disciplines using statistical methods. Of course, some professions have their own codes of conduct that might incorporate principles that are not covered here.