In the Cities Institute, where I did my PhD, I helped on cartographic projects, notably taking the lead on the spatial analysis for one project working with the CWASU and Solace Women’s Aid.
Some of the data analysis work I’ve done for the Child and Women Abuse Studies Unit (CWASU) last year has been published in a report: Finding the Costs of Freedom – How women and children rebuild their lives after domestic violence.
Mapping social and relational networks
The CWASU had interviewed 107 women and children in the UK who had been or were victims of domestic violence. One of their research interest was to identify the relevance of social and relational network to rebuilding their lives. They interviewed four times over three years, the first wave have 100 women and the last 65. Most of the loss was due to change of home and contact details. During the interviews organised through the charity Solace women’s aid, they asked the women to quantify (family members and friends), qualify (disruptive or supportive) and locate (postcode) their relatives. The relations also were quantified in terms of contact frequency, ways of contact, and length of contact. Although they performed this exercise four times, we mapped only the first (95 maps) and the last (60 maps) waves.
The data
I had access to a list of relations, with a lot of information describing the friends and relatives, and their interactions (or lack of).
The friends and relatives:
- where they lived,
- indicating if the person was a friend or family member,
- the type of relation (disruptive or supportive)
The interaction:
- the frequency of contact
- the type of contact (face-to-face or not)
- – where the contact happened (when it did; some meetings happened at their own home, at the other person’s home, or in another neutral location).
These maps challenged me in many ways. I had not been present during the research design, the data collection, or the data entry; so the data had to be cleaned before being able to analyse it.
A few women had been very secretive about the location of their social network, and so it was hard to map it. Actually, I was more surprised how well they knew their relative’s postcodes! I tried my best to pin down some locations that were very vague, at best it was the name of a city or neighbourhood, at worst it was just ‘a cafe around that neighbourhood’. The further away the person lived, the less the precise location really mattered for the analysis. In a very few cases the postcodes were wrong, but that did not happen so often.
The maps
I used ArcGIS to make the 95 + 60 maps, making one map per person. On the map I tried to show the extent of their social network as well as their interactions with each members of the network.
- Each person they knew was given two attributes: friends/family (shape) and disruptive/supportive (color)
- The relationship was described in terms of frequency (thickness) and mode of contact (color and type).
- meeting places, if they happened in neither home’s
- people without precise locations were mapped a buffer
All this was mapped onto one page each, but for some networks I had to use multiple frames to visualise their very close relationship and others living much further away.
Although the maps were not essential to the analysis, they were really appreciated by the women who got to see them three years later. It helped them coming to term with the journey they had gone through so far.
The typology
Every woman’s story was different and unique in its own way; some had completely cut ties with their families because of the situation, some had no friends, or all their friends very far away. Some kept close relationship with people they qualified as disruptive. Some had only new friends, or old friends only they never saw.
I designed a typology based on the extent of this network (see table below), and compared the changes three years later.
|
Typology
|
Description
|
|
Micro local
|
More than 50% family and friends live within 2 km of the woman
|
|
Local
|
More than 50% family and friends live within 5km of the woman
|
|
Across UK
|
More than 50% family and friends live in micro-local, local and other parts of the UK
|
|
Non-local
|
More than 50% family and friends live outside of local and micro-local but within UK
|
|
International
|
More than 50% family and friends live outside the UK
|
|
Scattered
|
None of the above
|
At first we were very surprised to find the ‘scattered’ category, but there were 7 women that fit in that category. However the category completely disappeared in the last wave, while local categories increased and global categories decreased. The women had in a way started to ‘sort out’ their lives, and invested more time to closer relationship.
What I learned
It was the first time I worked with such a large dataset, and with so sensitive data. Working in GIS, with real data, can be very intrusive; I understand that some women were very secretive about their or other people’s locations. Working with such large files required to automate as much as possible, jumping from QGIS to ArcGIS to use different tools. I learned a lot about templates!
Working on this project was an eye-opening insight into domestic violence and the burden that some women go through even once they have escaped abusive partners/relatives. I did not have access to a lot of data, only the description of their relationship, which I had to categorise and ‘fit in boxes’. Sometimes it was difficult to quantify so much real lives. The very little I had access to, however, showed there was no simple answer ‘just get out of the situation’, and that everyone around the victim/perpetrator was affected in different ways.