Data Quality issues have arisen in a prosecution arising from a tragic bus crash in Dublin. `

Update: Driver found Not Guilty

Root cause:

The root cause of the concerns about the quality of the information seems to be a failure on the part of the computer technician conducting the investigation to properly segregate the Dublin Bus data from data relating to a different crash on a different continent involving a different model of vehicle.

This echoes the root cause of a healthcare incident  last year where a failure to follow best practices in segregating samples and data when preparing and analysing tissue samples lead to a young man having his stomach needlessly removed.

It further transpired today that there appear, in the view of the defence's expert witness, to be inconsistencies between the data and the evidence recorded by investigating police. Prosecutors questioned whether the expert witness was misreading the data or was unfamiliar with the engine management systems of the particular model of bus - a claim the expert denied.

Other sources suggest that the investigating police had difficulty with the quality of the information that was produced from the Volvo engine management system. In evidence, a police witness said that most of the information was incomprehensible to him. The investigating Garda (Irish Police) had noticed that the engine operating hours and mileage on the data provided by volvo did not tally with information the poice had taken from the bus at the time of the accident. On investigation with his contact in the Volvo importers he was told that the "inconsistencies" were due to the "statistical nature" of the data.

And those weren't the droids he was looking for either...

Impact?

The impact here is that a breakdown in effective data stewardship has resulted in data which was inaccurate being presented in evidence, and a suspicion being raised that the 'new' information is inconsistent, incomplete or possibly corrupted. That this has been raised in a prominent court case has an impact on the life and livelihood of the accused, as well as on the understanding the families of the deceased and the survivors may have of what actually happened to cause the tragic crash.

Analysis

The expectation of legal processes is that a careful chain of evidence will be maintained, particularly in criminal prosecutions. This requires an unbroken Information Chain from whatever systems are used to obtain source data for investigative purposes. Inconsistencies must be capable of explanation. Inaccuracies are unforgivable.

The chain of evidence here has been broken. If, in a drugs prosecution, the police admitted they had produced evidence arising from a different search of a different house in a different city then the evidence would be thrown out - if not the entire case.

On the question of interpretation of information, if there is such variance in the data formats and outputs from engine management systems which seem to baffle expert witnesses and police investigators alike, perhaps a requirement should be raised in law that all vehicle manufacturers develop EMS systems that produce data in a defined standard format for the purposes of accident investigations.

Standardisation of format and standardisation of process removes the need for expert witnesses to be expert in the specifics of individual manufacturers engine types and would allow the courts to focus on the information itself, and a knowledge of the laws of physics, to reach a determination as to what the root cause of the accident was.

Non-comprehension due to the "statistical" nature of information is a red-herring. In this context it absolutely has to be. All statistics are built up from raw data using formulae or algorithms. These may require some explaining and some documentation but there is no reason why the data could not be explained and verfied. For information that would be used in evidence in a criminal prosecution to be "incomprehensible" is... well... incomprehensible and suggests an absence of data about the data (aka meta-data).

Of course, there is also the possiblity that the investigating officer had not been adequately trained in statistical analysis and might have been stretched to understand the information. That's why W.Edwards Deming taught that we should "institute training and retraining" in the job to continuously improve skills.

As Information Quality professionals, members of the IAIDQ and the IQ Network must work to raise awareness of the need for high standards in information management to ensure that the quality of what is produced meets or exceeds the expectations of all information consumers - including the legal system. If, as information quality professionals, we fail to achieve and maintain the highest standards then we let down not just ourselves and our companies but potentially society as a whole.

IQ Matters. Join the IAIDQ, become part of the IQ Network and learn how to improve the quality of management of information.

[UPDATE]

Coincidentally Larry English, a co-founder of the IAIDQ, has an article on B-Eye Network this week about a less serious but otherwise similar issue in the US. Take a look.... or PodCast it..

As Larry says in this article... "“Reputation” of an information provider is not a guarantee of the quality of information provided."

Other links:

Irish Examiner, 2007-02-15: http://www.irishexaminer.com/breaking/story.asp?j=13718258&p=y37y83x4&n=13718346&x=

Irish Examiner, 2007-02-16 @ 15:15: http://www.irishexaminer.com/breaking/story.asp?j=283455300&p=z83456z5x&n=283456287&x=