For anyone concerned about crime levels, Statistics Canada has been offering reassuring news: Crime rates across the country are dropping. And not only are volumes declining, but so is the severity of offences.

Normally, that would be the end of it. StatsCan's numbers are usually accepted at face value.

But the Macdonald-Laurier Institute, an Ottawa think tank, has published a scathing dissent. The institute complains StatsCan understates the prevalence of crime. The agency counts only those offences reported to the police. But that paints too rosy a picture, it says, because twothirds of criminal acts go unreported.

In addition, drug-related offences are excluded. Yet these are a significant part of crime in many communities. And drug offences are on the rise: They climbed 27 per cent over the last decade. Leaving them out of the national picture creates an artificial sense of well-being.

The agency gives various technical reasons for these omissions, all of which have merit. Yet the combined effect could be misleading.

The institute also questions whether the severity of offences is really declining, a claim StatsCan bases on sentencing trends. Over the past few years, the average length of sentences has fallen. The agency equates this with fewer serious crimes.

Yet there might be a number of reasons sentences get shorter. Judges might take a more lenient approach. Or Parliament might mandate a different policy. That actually happened in 2003, when a significant portion of youthrelated offences were diverted away from the criminal justice system. Those offences still occur, but they no longer appear in the crime data.

Lastly, the institute complains that StatsCan does not report offences committed by individuals on probation or out on bail, although the information is available. The omission is significant, because there is a controversy around this issue.

Last week, Parliament passed a law that prisoners convicted of non-violent crimes must serve at least one-third of their prison sentence before being paroled. The Liberals and NDP wanted the minimum jail time kept at onesixth, where it has been for some years.

In weighing these competing views, it would help to know how much crime is committed by prisoners on bail or early release.

We disagree with the suggestion, hinted at by the institute, that a political agenda underlies StatsCan's crime reporting. Most of the policies in dispute have been around for decades.

Yet there are, perhaps, still grounds for concern. StatsCan bases its reporting, to a considerable extent, on advice from the Canadian Association of Chiefs of Police.

Like any other professional group, police chiefs are naturally concerned about their image. A litany of statistics showing burgeoning crime, or ineffectual efforts to combat it, would obviously be unwelcome. In view of that fact, it would be wise for StatsCan to take a more independent stance.

One reason Canadians worry about crime is the apparent disconnect between official statements and their own experience. When only seven per cent of car thefts or eight per cent of burglaries are solved -and those are the figures in B.C. -they see an entirely different reason why crime reports are falling. Victims have given up on law enforcement.

Of course StatsCan is merely the message carrier. It has no responsibility for policing.

But it does have a duty that goes beyond technical accuracy or carefully worded footnotes.

Its reports should fairly capture a sense of the national experience. Bland statements about declining crime rates that leave out whole categories, or deal only with the tip of the iceberg, don't do that. Some rethinking is in order.