Caledonian System Evaluation: Analysis of a programme for tackling domestic abuse in Scotland

Evaluation of the Caledonian System: a programme to tackle domestic abuse in Scotland.


1. Defined as 'any form of physical, non-physical or sexual abuse, which takes place within the context of a close relationship, committed either in the home or elsewhere. This relationship will be between partners (married, co-habiting or otherwise) or ex-partners.'

2. Although experience of partner abuse remains widespread, analysis of the SCJS over time indicates that it has fallen since 2008/9 - from 18.2% saying they had 'ever' experienced it in 2008/9 to 14.1% in 2014/15. Similarly, the proportion reporting having experienced partner abuse in the last 12 months fell from 4.2% to 2.9% (Murray/Scottish Government, 2016).

3. The predecessor to the Scottish Advisory Panel on Offender Rehabilitation ( SAPOR)

4. With the exception of one Women's Worker who was interviewed individually. The researchers ran two groups in each area - one involving Men's Workers and one involving Women's and Children's Workers. One delivery manager participated in a group discussion, while two were interviewed at a Delivery Managers event. The report also draws on additional discussions with staff - particularly Delivery Managers and Data Champions - during the course of fieldwork, including during a Caledonian practitioners' event in early March 2016.

5. By asking Hubs to submit anonymised lists of participants and suggesting who they should approach first.

6. The evaluation team only received consent to contact two Sheriffs, and repeat attempts to set up interviews with Sheriffs and with Police Scotland proved unsuccessful. The timetable for interviews was fairly tight, however, which may have been a barrier for some stakeholders.

7. 'System-generated risk' refers to the ways in which involvement with a service might actually increase rather than reduce the risks to women - for example, as a result of men trying to control women's engagement with the service, or as a result of information sharing leading to breaches in women's confidentiality.

8. For further detail on each of these roles and their key responsibilities, see the Caledonian System Synopsis manual.

9. It is perhaps worth noting, however, that the usefulness of some of the early paperwork for monitoring and evaluation purposes (e.g. psychometric tests) is dependent on them being collected as early in the pre-group stage as possible.


11. It is worth noting in light of these comments that the Judicial Studies Committee did recently run two-day training sessions for Sheriffs on domestic abuse. A short session on the Caledonian System formed part of this training.

12. Millon Clinical Multiaxial Inventory - a tool for identifying personality disorders and clinical syndromes - see for further details.

13. See Kropp and Hart (2000) for details.

14. For example, Aberdeenshire estimate that there are 17 cases erroneously coded as not having completed, because the database would not let them enter a 'successful completion' outcome as they had not completed a specific psychometric with them (since it was only introduced in their area later), and a further 4 successful completions that were miscoded. Lothian and Borders own data indicates a successful completion rate of 60% - higher than the 55% derived from this analysis of the monitoring data.

15. While there are fields in the database that allow reasons for breach to be recorded, these are incomplete and difficult to interpret - for example, of the 99 cases where a reason for not starting group work was recorded, 36 are coded 'no', with no further explanation.

16. Although those completing the programme were slightly more likely to be recorded as 'White', this largely reflected a higher proportion of non-completers whose ethnicity was recorded as 'not known'.

17. Which includes offences relating to threatening or abusive behaviour and stalking. Offences that would have been recorded as 'Breach of the Peace' pre-2010 would have been recorded under the Criminal Justice Licensing Act from 2010 onwards.

18. Assault is defined as 'an attack upon the person of another'. 'Assault to injury' is a more serious charge than simple assault, and means the assault caused actual bodily harm.

19. A fact that makes it difficult to use the monitoring data to assess whether or not those with alcohol and drug problems are indeed more difficult to keep engaged - most men are recorded as having had some degree of alcohol (71%) or drug (55%) problem when they start on the programme, and similar proportions of those with and without problems go on to complete the programme. However, the monitoring data does not tell us whether these problems became more or less serious or well-managed as they progressed through the programme.

20. Based on all women with a ID entered on the monitoring data

21. The women's monitoring data also includes some measures that might, in theory, be used to assess men's progress - particularly, the partner behaviour checklist and women's behaviour checklist. However, as noted in section 2.2.1 and in Appendix B, there was a particularly high level of missing data in fields based on these checklists, so the evaluation team were unable to use it.

22. Initially by the assessor, and at Maintenance stage by Men's and Women's Workers using information gained from contact with both the man and the women, wherever possible.

23. An assessment procedure designed to provide objective measures of one or more psychological characteristics (such as beliefs, values and attitudes), based on standardised conditions which have known reliability and validity (i.e. they provide a reliable and appropriate way of comparing one person with another).

24. Two additional assessments - MCMI III and the Relationship Style Questionnaire are also included in the System. However, as these are not intended as outcome measures - and there is no data for either in the Monitoring Data after the assessment/pre-group stage - we have not reported on these here.

25. At pre-group stage, current alcohol and drug problems are measured using a scale of 0 to 3, where 0 is the most severe level of problem and 3 is no current problem; at maintenance stage the level of each problem is recorded in the monitoring data using a Yes or No response

26. Most recently in a UK context, Kelly and Westmarland (2015) concluded they could not use their control sample since it differed in too many key respects from the intervention sample.

27. 'Yes' based on all those with a score of 0, 1, or 2. 'No' based on all those with a score of 3.


Back to top