WINSTON Churchill remarked famously in 1910: "The mood and temper of the public in regard to the treatment of crime and criminals is one of the most unfailing tests of the civilisation of any country." If the retributive rhetoric of recent months is to be believed, Ireland may be failing this test.

The important issues are brought sharply into focus by the debate on whether to broaden the grounds on which bail can be refused to allow preventive detention.

The public is understandably frustrated with bearing the brunt of crimes committed by people who have already been accused of other offences and are awaiting justice. In conflict with the need for public protection from this source of victimisation is the accused's right to be presumed innocent until the State has proved the case against him beyond reasonable doubt. The issues seem clear-cut.

But how widespread is the problem of offending on bail? In 1993 the gardai recorded 98,772 indictable offences. Of these 35,812 were detected. Of the detected offences, 9 per cent were believed to have been committed by people on bail, although fewer than half resulted in a conviction. Only 2 per cent of detected rapes and 1 per cent of detected assaults were committed by so-called "bail bandits".

Therefore, restricting the liberty of those few who have been arrested and await trial can have little effect on crime. It is misleading to suggest otherwise.

Furthermore, after its exhaustive examination of the evidence last year, the Law Reform Commission concluded there was a higher rate of offending on bail in England, even though the courts there already have the power to order a remand in custody if there are "substantial grounds" for believing that the accused will commit further offences before his case can be dealt with.

But, even accepting that the overall impact on crime levels will be marginal, there might still be some justification for changing the law if accurate identification of those likely to offend on bail was possible so that they could be selectively detained.

However, the clearest message which has emerged from the research in Britain and the United States is that both clinical approaches, (tailored to the individual case) and statistical or actuarial methods (based on aggregate assessments of risk for classes of offender) are of strictly limited utility.

Accurate judicial risk assessment is notoriously difficult. There is no straightforward association between a defendant's previous record, the circumstances and nature of the current offence, and the likelihood of offending on bail. A further complication is added when one tries to take account of the relative weighting to be given to the low risk of a serious offence compared with the high risk of a trivial one.

The key problem is what is known as the "false positive" rate, that is the number of those inevitably detained who would not have offended had they remained at liberty. Predictive methods are so crude that this is always a significant issue.

Any further restrictions on granting bail will have an immediate effect on the size of the prison population. In England and Wales, which has the kind of bail regime to which many in this State aspire, almost one in four of those in custody is held there on remand. This is four times the current Irish level. It has been estimated that a new bail regime could require the provision of at least 600 new prison places at a capital cost of some £80 million and annual running costs of £20 million.

AN examination of the data from other jurisdictions also reveals that many of those remanded in custody by the courts are discharged as soon as their case is concluded. The element of public protection associated with pretrial detention is clearly limited.

The most recent prison statistics for England and Wales show that, of males remanded in custody by the courts, fewer than half (44 per cent) received a custodial sentence. The remainder were either acquitted or not proceeded against (23 per cent), or given a non-custodial penalty (33 per cent). For females remanded in custody, only 29 per cent received a prison sentence when their case was finally heard.

Thus the current state of criminological knowledge suggests that preventive detention before trial will prevent little crime, lead to the incarceration of many who are eventually acquitted or given a community sentence, and contribute to a sizeable (and expensive) increase in the prison population. These factors are generally ignored in the call for changes to the law on bail. However, they must be taken into account if a reasoned and just decision is to be made.

It is worth repeating the words of the Law Reform Commission, which expressed the view: "Public opinion cannot be allowed supremacy over constitutional liberties if it is based on incorrect information, e.g., about the efficacy of pre-trial preventive detention, or prejudice, or simple misunderstanding of the issues and the context." In other words, fundamental change should not be driven by the temporary gains which might accrue from measures introduced for reasons of political expediency or misplaced public anxiety.

A hidden and disturbing feature of the current debate is that increased punitiveness for the few masks growing insecurity for the majority. We should learn from other jurisdictions the futility of trying to reduce crime by increasing the quantum of punishment. The ratchet effect of such policies leads to penal expansionism, misery and public anxiety.

Endorsing policies which have such effects is not an effective way of tackling crime or a sign of strong government. Far from it. Confident government in the arena of criminal justice is characterised by consideration, fairness and understanding. Embracing such values is the best way to prevent the next victim.