Old tricks for new dogs. What price local knowledge?

Professor Jason Roach and Professor Ken Pease

One way to denigrate experience is to invoke the saying ‘You can’t teach an old dog new tricks’. The unstated assumptions are that the new tricks are better than the old, and that anyway the old dog is too stubborn or inept to bother learning them. Some old dogs are indeed stubborn and inept.  But there are opinionated over—confident young dogs too. Failure to teach new dogs useful old tricks is arguably even worse than failing to teach old dogs useful new tricks.

The common issue is arrogance, both from the old sweat who thinks no-one can teach him anything, and the young tyro who thinks the same. Asking either ‘How do you know that?’ evokes a withering stare of contempt, an unspoken ‘How dare you presume to challenge my assertion?’

One of the frequent side-effects of being an academic is that the longer you are the more you tend to be drawn to revisiting some of the theory and research that you were taught as an undergraduate. In our case, this has been regular conversations about the psychology that we were taught (albeit some 30 years apart), explaining why, for example, we wrote a book, ‘Evolution and Crime’ (2013). We recently scanned an introductory psychology textbook from 1990’s to see what we might revive in pursuit of crime reduction and policing today. Instead of agreeing that each chapter topic was either still popular or too outdated for any resurrecting by us, we found it to be a rich source of not just nostalgia (although admittedly a lot of nostalgia) but of psychological ideas, theories and approaches ripe for applying to 21st century policing, that had for some reason fallen by the wayside. Unless told otherwise, this is the first of our ‘resurrection’ articles.   

Another common side-effect of aging in academia is that we tend to speak about great academics that we have had the privilege of being been taught by or have met in the past. As the first author’s list is considerably shorter than the second author’s as he much younger (and more handsome) we begin with a true legend.1

The second author (KP) was privileged to be taught by one A.R.Jonckheere, a brilliant psychologist/statistician now remembered for his eponymous Trend Test, and less recalled for providing the most endearing example of nominative determinism we know. His full name was Aimable Robert Jonckheere and he was indeed lovable. He was also amazing in the depth and breadth of his talents. Professor Gloria Laycock recalls a weekend when Jonck (as he was known to colleagues, friends and students) suggested a game of Scrabble. His then girlfriend agreed on condition that it would be in English, not any of his other languages.

During a seminar in Jonck’s office circa 1964, he described a visit he had just made to the Nobel Laureate ethologist Niko Tinbergen at Oxford. Tinbergen worked a lot with sticklebacks. Wanting to know how accurate Tinbergen’s stickleback predictions were, Jonck designated a stickleback in the tank and asked him to repeatedly predict the direction in which the stickleback would turn next. Statistician that he was, Jonck said ‘I did a sign test in my head’ and concluded that Tinbergen did significantly better than chance. The point is that prediction was Jonck’s acid test of someone’s expertise, echoing the Russian proverb favoured by President Ronald Reagan ‘Trust, but verify’.

So, what has this got to do we crime and policing? Please bear with us a little longer.

The perspective from which evidence based professional practice is branched has its origin in medicine with the Cochrane Collaboration, named for Archie Cochrane, a medic who underwent radical surgery for a cancer which he did not have. The core Cochrane insight – based upon his surgeon’s confidence that he was dealing with cancerous tissue without waiting for a pathologist’s report – was that medical practitioners claimed (in good faith) expertise which they did not possess, and which had to be tested. How do you know that someone has expertise? The evidence lies in their capacity to predict. The surgeon thought he could predict what the pathologist would report. It turned out he could not. For 2000 years of medicine, confidence outran competence.

The larger scale work was pioneered by Philip Tetlock (see Gardner and Tetlock 2015) with an honourable mention going to Nate Silver (2012). Tetlock started by studying political pundits. Put kindly, their predictive powers were execrable. He went on to craft the Good Judgement Project (GJP). The predictions which he elicited were about difficult to predict real world events. He collaborated with Intelligence Advanced Research Projects Activity, recognising the centrality of prediction in making the intelligence services skilled.

Much of policing is understandably ‘reactive’, triggered by the reporting of a crime by the public. That said, policing requires an extensive skill set. All skills are tested by accurate prediction. Thus, skilled policing is discernible by accurate prediction. In football, goalkeeper capacity to anticipate where the penalty taker will direct the ball makes the save easier. Whereas for cricket the capacity to anticipate the googly enhances the batter’s capacity to remain. Prediction is what lets the best professional gambler get rich and bad prediction often letting the bookmaker.  Yet, for some reason, the aura of expertise typically leaves the capacity to predict untested. Why?

This has long troubled one of us in the policing context. For instance, police officers (designated Crime Prevention Design Advisors, Architectural Liaison Officers, or similar) have long advised planners on crime-facilitating design features. We could find no studies of whether these recommendations had any validity in identifying places that would be prone to crime. Our colleague, Leanne Monchuk, gave experienced Designing Out Crime Officers plans of an estate which had been built a decade earlier but with which they were unfamiliar. She then compared the locations identified by the CPDAs as crime prone with the locations at which there had actually been crime. The results (Monchuk et al. 2018) were as might have been expected on the basis of larger scale work on prediction. Nobody was perfect. Some were modestly skilled at prediction, some were random. A small study, but enough to persuade us that police capacity to make predictions relevant to their work should be tested, not assumed.

We see evidence-based policing to be of two kinds. The first is to identify those things worth doing in crime control. This is the Cambridge emphasis. The other is to so develop individual officers so that they are best equipped to do the things worth doing skilfully in the messy world of implementation. For this, we think the Tetlock approach necessary. It has the incidental effect of revealing senior officer hubris, just as Cochrane undermined physician hubris.

In hope of ‘starting the ball rolling’, here are some of the things which Tetlock established.

1. Some people are superforecasters, reliably beating the odds.

2. Superforecasters were not necessarily the most intelligent participants. They had these attributes

  • They did not see the world through a big theory lens (like Marxism)
  • There were conscientious in gathering relevant information
  • They adjusted their view readily when new information became available
  • There were ways of setting up groups which outperformed the predictive accuracy of its members.

One of us (JR) has conducted interviews with those police traditionally referred to as ‘ace thief-takers’, those seemingly gifted to spot a ‘wrong-un’ in a crowd and who invariably get more ‘collars’ than their colleagues. Although not yet complete, the findings suggest that said individuals are not simply good at predicting criminality based on the picking-up of various different cues, but that they have excellent memories for faces, and arguably, and most importantly, are willing to test their predictions and be wrong!

If you find yourself in sympathy with the ideas expressed here, we suggest that they could be incorporated into policing in many ways. For example, body worn camera footage for developing incidents be shown to apprentice officers, with predictions about how the incidents resolve themselves tested against actual outcomes. Not allowing experienced officers to escape, predictions made at Tasking and Coordinating meetings seem to provide scope for prediction testing.

Those familiar with the research on ‘super recognisers’ and will know that some police staff identified as being in the top 30% at facial recognition (e.g. based on CCTV images) or number plate recognition, have been put together in specialist teams (e.g. MPS). We posit that although memory will play a large part here, prediction will also play a significant part.

In hope more than prediction, we write this article in a hope that readers might access the GJP website, read Tetlock’s work, especially Tetlock and Gardner’s book, and recognise the potential of the approach in enhancing relevant policing skills.

Please do contact us if we have aroused your interest for testing some of these ideas

Thank you for your time.


1: The second author has nothing to contribute to the following two anecdotes as he was not born for another five years and makes no apology for this.

Jason Roach is Professor of Psychology and Policing and Director of the Applied Criminology and Policing Centre at the University of Huddersfield and Editor for the Police Journal. Email: j.roach@hud.ac.uk. Twitter: @jrro47

Ken Pease OBE is Visiting Professor at University College London, the University of Huddersfield, University of Loughborough, Manchester Business School and Chester University. Email: kengpease@gmail.com

One thought on “Old tricks for new dogs. What price local knowledge?

  1. Re- “interviews with those police traditionally referred to as ‘ace thief-takers’, those seemingly gifted to spot a ‘wrong-un’ in a crowd and who invariably get more ‘collars’ than their colleagues. Although not yet complete, the findings suggest that said individuals are not simply good at predicting criminality based on the picking-up of various different cues, but that they have excellent memories for faces, and arguably, and most importantly, are willing to test their predictions and be wrong!”

    I agree with what your findings might suggest. However, my experience of working with a number of police officers (uniform and detective) over a number of years, is that those that were good at catching criminals were those that were willing to put the extra work in, and get out there checking vulnerable property, doing house to house enquiries, cultivating and talking to informants, witnesses, suspects, buisness owners, shop owners and others in the community. They were willing to come back with no results but that did not stop them from keeping at it and trying again, not in a hap hazzard way but based on intelligence and knowledge (and yes with a good memory of that intelligence and knowledge which they also helped generate through their persistance). In effect they were putting the evidence base on intelligence led policing, problem orientated policing and community policing etc into practice (probably without realising it) but above all they were willing to put the effort into it and willing to keep at it when some lines they followed proved negative.

    All the best with your continued research.

    Dick Severns

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s