The book starts with one of the earliest randomised tests, in 1747, by the British naval surgeon James Lind. He was trying to find an antidote for scurvy, a condition so lethal to seamen that it was to kill more than seventy per cent of British sailors who served in the Seven Years’ War between 1756 and 1763. Lind tested six pairs of sailors and established that the pair given citrus fruit responded positively. Lind wrote up his results, but these were largely ignored until 1795 when lemon juice became standard issue on British ships.

The kind of randomised tests that most readily spring to mind are those for pharmaceutical drugs. These normally start with animals and then proceed in various stages through humans, with one group being given the drug and the other – the control group – given no treatment, or a placebo. These can be ‘double-blind’ tests where neither the doctors nor the patients know who is getting what. The book tackles head-on the obvious ethical question of why the control group should be deprived of a drug that may benefit their particular condition. Leigh’s answer is that the point of the test is to determine the efficacy of the drugs: the effects of the drug are still uncertain, so the control group can end up receiving the best current treatment. This seems right, but there are clearly difficult decisions to be made by researchers in this area.

The same kind of ethical questions arise, although perhaps not quite so dramatically, when these types of tests are used to assess the value of social programs. The book uses the example of a Melbourne program designed to assist homeless persons, which produced little positive results over a period of three years, and looks at American examples of efforts to improve the condition of unemployed individuals. There have been similar assessments of the programs to improve learning at all levels of education, from pre-school to tertiary, although assessors need to be aware of distorting factors, such as the influence of family life on school students as opposed to what happens in the classroom. Leigh emphasises the importance of the correct use of statistics by persons conducting these tests. It can be added that responses to surveys can be heavily influenced by the leading character of the questions asked.

When looking at efforts to improve the criminal justice system, the book notes some reasonably predictable results, such as regular police presence being a factor in the reduction of offences. There is some scepticism on the part of the author about the utility of prisons, but this view arguably overlooks the fact that most crimes are committed by a relatively small group in society which has no real interest in rehabilitation. It is true that this group is not deterred by longer sentences, but these terms of imprisonment are the only protection that the community has against serial offenders.

Dr Andrew Leigh MPLeigh has a particular interest in the Third World and argues that randomised tests have an important role to play in assessing government initiatives by these countries in agriculture, business incentives, health, and education. There is also, however, a detailed examination of their use in conventional, first-world fields of pricing, marketing, and advertising. It is said that Google added US$200 million dollars to its bottom line by finding the perfect colour for its toolbar. In the case of political advertising, much of the evidence suggests that it is largely ineffective, but this does not stop it from being the principal cost of political campaigning in Australia and in other parts of the world. If it could be eliminated, there would be little need for laws governing donations to political parties, but Australia’s High Court has ruled out this solution, divining a doctrine of freedom of political communication under the Constitution and using it to protect paid political advertising.

The book discusses social science laboratory experiments, with university students usually the test subjects, and points out that not only is there likely to be a difference between these students and the general community, but the students who volunteer for the experiments are self-selecting and are likely to be different from the student population at large.

There is a wealth of information in this work and even more in the notes to the various chapters, although there is almost a degree of overkill in Leigh’s fervent advocacy of randomised testing, given that most readers would either accept the utility of these exercises or be persuaded by the early stages of the book.