You're reading...
Operational Research

OR Society Criminal Justice SIG Meeting Presentations 26th November 2012

I attended the OR Society’s Criminal Justice Special Interest Group meeting yesterday where we had four interesting presentations. The summary of the meeting follows. Our next meeting is in West Yorkshire on March 4th.

Rebecca Endean, Director of Analytical Services, MoJ

OR in the MoJ

The OR team has a large influence on policy areas and currently comprises some 190 analysts. The team has been expanding, despite 33% back office cuts elsewhere. It includes a dedicated modeling unit which has been in demand for quick, reactive work (e.g. impact on prisons of the riots).  They also develop policy models.

Current OR uses

Criminal justice system modeling involves a suite of models that look at trends of inflows to courts, by age, crime types etc. and have an underlying assumption that past trends will continue. The team keeps a close eye on this assumption and had increasingly good data from across the CJ system and other government departments to draw on.  By comparison, data for  civil and family law, the data is poor.

The Prison projection model uses micro simulation with historical prison data and custodial projections. Their models predicted the riots to increase prison population by 1000. This was spot on. Court records were used to give good quality input data and the team are now looking at individual Court data which will further improve accuracy.

Scenario models are used to look at, for example, the impact of the spending review and CJ Reform policies.

Genetic algorithms are being used to model the Court Estate to help identify efficiency savings by establishing optimal Court space (size/location).  There are some 500 buildings currently, with a number closing and being consolidated into larger, more flexible buildings.

Data envelopment analysis is being used to help identify what does Court efficiency look like and to enable more effective benchmarking across Courts.

Conclusions

Basically there is excess demand across the CJ system, so the MoJ OR approach is to have a sensible system of models that are coherent and can meet the varied needs of stakeholders. The team is applying new tools to find better solutions and is finding that Finance & Operational people have confidence in models.

They have made a conscious decision to mix skills in the OR team: OR, Statisticianss, Economists.

Jane Parkin

Crimestoppers Simulation of Call Centre

Crimestoppers is an Independent charity that takes anonymous calls and e-mails from the public. They also provide education.

95k calls p.a. have useful information and result in 8k arrests.  The Contact Centre operates 24/7, with 8-9 hour shifts.  Calls arrive randomly and Crimestoppers was expecting 60% increase in business due to taking on new work. The question this project set out to answer was “Would longer shifts work?”

Simulation

PRISM, a simple version of Witness, was used to model calls and staffing.  Some fudging was required due to limitations of the software, but it was quick and easy to use.

The model included:

  • 3 categories of staff
  • Supervisors take calls if staff are busy
  • One year of data on call volumes, but no time pattern
  • Average hourly calls
  • No online data (forms submitted)
  • No processing time data for forms
  • Daily calls M-F virtually the same
  • Saturday busier than Sunday

The service target is 90% of calls answered in 20 seconds and the model was able to compare Service Level (SL) performance and staff utilization.  This quickly identified a problem with Saturday staffing.

What if scenarios were tested and the options were rated. The option chosen increased SL by 7%, Abandoned calls down by 47%, Staff utilisation up.

Staff didn’t like proposal for the new shift pattern and made some of their own suggestion which was also modeled. This proved to be more expensive and no better service.

Validation of the model has shown actual data closely matched predictions. Read more about this case study here.

Chris Smith, Warwick University

WASAN for efficiency at Warwickshire Police

Soft problem structuring methods were used to identify ways of reducing wasted call handler time in the Control Room.  The WASAN method originated in the nuclear waste industry (2010 Shaw and Blundell) and takes a whole system view, including upstream and downstream impacts, to identify sources of waste in the system.

The scope of this project included 999 & 101 calls in the Force Control Room and the project included these key stages:

  • Define system boundaries and purpose
  • Analyse operations via interviews with staff
    • External causes of waste
    • Calls not for Police
    • Internal causes of waste
    • Rotas
    • IT
    • How could we avoid waste calls
      • Staff ideas
      • Cluster into themes
      • Wastes
      • Actions
      • Evaluate actions vs. criteria
        • Alignment with strategy
        • Costs : Savings
        • Risk to public, staff, officers
        • How could we reduce cost of waste calls
          • Need to look at up and downstream consequences as well
          • Arrive at actions to implement

Key learning

It’s difficult to define ‘waste’ in call handling, at least initially.

Upstream issues were really important, due to their impact on the system

Munira Dossaji, MoJ

Lessons learnt from international benchmarking of CJ indicators

Value of international comparison:

The range of possible benefits includes…

  • Policy development
  • Efficiency
  • Drive performance
  • Good practice identification
  • Cost effective solutions
  • Scrutiny
    • Internal
    • External
    • Citizen
    • More informed
    • Learning

Current indicators:

The project identified 7 sources of possible data for the CJ system which included data from the EU and UN, which are published regularly.

What’s missing?

Some of the issues identified include…

  • No one source is available
  • Limited geographical coverage
  • Lack of standardization of KPIs
  • Different levels of detail available from the different sources
  • Missing indicators
  • Inconsistency of terms

Pitfalls of international comparisons:

There are different penal codes and their administration varies. These have changed over time, for example due to Legislative changes.  Crime categorization also varies and can therefore make comparisons difficult.  Other factors such as varying Age of criminal responsibility, first or repeat offenders, Impact of guilty pleas on sentence, types of sanctions and life sentences all make comparisons challenging.

The NAO also highlighted these issues in a report: Feb. 2012.

A tool for comparison:

The project identified 20 indicators, mostly quantitative, and looked at original data on international websites for 5 countries of interest.

Groupings of indicators were created, showing external drivers, internal drivers and the system relationships between the indicators. The final “outcome” indicator was the change in prison population.  The model used colour coding for No change, Increasing, Decreasing. Unsure to present a colour coded ‘story’ of each country’s CJ system.

Tips and Learning:

  • Take a system view of the CJS and look at system-wide indicators
  • Look at a few countries
  • Use locally published data
  • Select key indicators and see how they change over time (Time-series data are essential)
  • Examine penal code and how it has changed over time
Advertisements

Discussion

Comments are closed.

Connect with Ian Seath

Find us on Facebook Improvement Skills Consulting Ltd. on LinkedIn Follow IanJSeath on Twitter

Archives

Copyright Notice

© Improvement Skills Consulting Ltd. and Ian Seath, 2007-17. Unauthorised use and/or duplication of this material without express and written permission from this site’s author and/or owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Improvement Skills Consulting Ltd. and Ian Seath with appropriate and specific direction to the original content.
%d bloggers like this: