19 June 2011

Fudgy figures

In a recent post I expressed disquiet about potential survey bias in a report by the National Union of Students. There's a more hard-hitting analysis in the Victorian Ombudsman's 41 page report [PDF] on an Investigation into an allegation about Victoria Police crime statistics.

The Ombudsman states that -
Crime statistics are an important public interest issue. The processes for the recording and reporting of crime need to be trusted. Crime statistics are not only used to inform the Victorian public about crime, but also critically to inform operational decisions by police, such as the allocation of police resources.

In 2009 I reported to the Parliament on the issue of crime statistics and police numbers. At that time I recommended that there needed to be an independent body separate to the Victoria Police to manage the release of crime statistics.

Since that time nothing appears to have changed. Crime statistics are still managed and disseminated by Victoria Police, with known inefficiencies and other long standing concerns. There is still mistrust in the way crime statistics are used, which can be confusing and impacts on public perceptions and confidence.

In late February 2011 I received a complaint from a whistleblower that Victoria Police crime statistics released on 28 October 2010 had been manipulated for political purposes and the data was misleading. The media release stated that "... further analysis of the past three months – compared to the same three months last year - has shown a 27.5% fall in street assaults in the city and a 12.4% decrease in assaults across the state". I investigated the complaint under the Whistleblowers Protection Act 2001.

My investigation identified that the decision to release the crime statistics several days before the caretaker period leading up to the November 2010 state election was the Chief Commissioner’s and his alone. He confirms this, as do other witnesses.

I consider that the quoted reduction of 27.5% of assaults in the CBD between the July-September quarter 2010 compared to the July-September quarter in 2009 without qualification, was based on yet to be validated data. It was therefore likely that releasing the data without qualification could reasonably be perceived to be misrepresenting the fuller picture of the trends. The crime statistics were subsequently used for political purposes during a public debate by the then new Police Minister three days before the election.

The release of the quarterly crime statistics data, particularly so close to an election, was likely to be used in a political context. Senior officers also failed to pass on warnings to the Chief Commissioner that the data was not "settled" and was incomplete. In this regard the Chief Commissioner has responded that "I accept the figures in the media release, and the July-September statistics as whole [sic], can be criticised on the basis the data did not have time to settle because their release was brought forward".

There is also the perception that the public were misled by the crime statistics. After all, the media for the previous two summers had raised concerns about violence on the streets in the CBD. The data also let rank and file police down. How could for example the Superintendent in charge of the CBD argue for more resources on the basis of this information or make sound operational decisions? It is also in contradiction to their practical experience of violence in the CBD. However, the Chief Commissioner stated that "resourcing and tasking decisions are not and never would be made on the basis of this single data set. We continue to inject new resources into the city and to supplement as appropriate with resources from other units to deal with the issues in and around the city".

The independence of crime statistics is crucial. I consider there is a clear need as I indicated in 2009, for an independent body to manage the release of crime statistics. I have recommended accordingly. Victoria Police endorses this recommendation.