In the mid-1990s the police domestic violence unit in Plymouth UK, a city with a population of around a quarter of a million, was receiving 1,000 notifications a year.

In 1999, Ahimsa (meaning non-violence), an independent voluntary agency treating domestically violent men, approached the Centre for Social Policy at Dartington for help in independently evaluating its treatment program.

It was established that Ahimsa could not ethically accept an RCT design because of the continued risk of life-threatening violence to a woman whose partner might be allocated to an untreated control group.

The evaluators’ response was to collect individual family and neighborhood data, before and after treatment, to provide a firm evidence base to meet SPR standards for work where RCT was not possible.

Between 1997 and 2000, 420 men were referred or asked for help and were offered an initial assessment. Some 83% were seen and data were collected on age, residence, civil status, work and health and parenting.

Quantitative analysis of local census data by public health staff permitted independent indicators of poverty and deprivation for each of 64 neighborhoods in the city, with the most deprived having the highest domestic violence referral rate to Ahimsa and, later, the highest treatment drop-out rate.

Analysis of age groups showed that most men were in their 30s, and were experiencing high rates of unemployment and sickness absence. Three quarters were parents or stepparents.

Men’s compliance with treatment was monitored during three stages of individual assessment and counseling as a preliminary to group work of 48 weekly sessions when they reported back on the impact of the previous session. Drop out from treatment in the first two stages before group work was heavy with just 29% of all initial referrals completing all three parts. Once in the group, men usually completed.

This loss was in line with many similar reports from elsewhere in the UK and from the US where drop out is considered a serious problem.

In Ahimsa, the pre-post design attempted follow up two years after treatment ended. A postal survey to the last address of all men initially seen had a low response, and the Ahimsa Women’s Service’s contact with partner victims found that half had left with no further contact with the violent partners. Consequently neither mail survey nor partners could produce reliable evidence as to whether the violent man had changed.

It was decided to make a direct approach and to focus a qualitative study on men completing treatment, aiming to explore any subsequent violence within the family or with a former partner.

As a pilot, follow-up home visits were offered to 18 men from two treatment groups, nine of whom were found to be still with the original victim partner. Two other men lived alone and three younger men in their twenties had returned to live with there own parents after further violence to their partners. Two more refused visits and two could not be traced, leaving 14 men to be seen either in their own or their parents home.

The assessment home visit used a tested procedure that codified standard social work assessment practice and in previous delinquency research had been shown to be reproducible and valid.

On these visits, with all of the family present including children, there was usually a lengthy unstructured discussion. It enabled relationships between partners and with children to be observed and noted. The research worker saw home and neighborhood conditions at first hand to augment evidence on the quality of family life.

Eight women independently claimed that their partner’s behavior was much improved and there had been no physical violence. Four men admitted further violence and two living alone had occasional contact.

In the absence of random allocation, this indication of outcome was based on positive contact at least two years after treatment finished. The small series of home visits permitted limited conclusions on treatment effectiveness.

In circumstances where an untreated control series was ethically unacceptable, collecting baseline data at referra, followed by monitoring of the treatment process and independent indicators of environmental deprivation, all helped to limit the influence of intervening variables which, unbeknown to researchers, might have influenced outcome. The domicillary follow-up provided improved data about family life.