djembejohn
commented
5 years ago I've done something on this issue...
See for the code https://github.com/CampaignLab/ward-level-crime-data/tree/master/code
What was I looking for?
We were interested in looking for relationships between crime and change in Labour vote share. For instances, there have been large increases in violent crime in London over the last few years. Are those areas more likely to vote Labour?
What did I do?
We have crime data for each London ward on a monthly basis from Jan 2016 - June 2018. I looked at two variables across a number of categories.
- Rate of increase over the 2.5 years (through regression)
- Amount of crime on March 2018
The categories were
- theft-from-the-person
- bicycle-theft
- anti-social-behaviour
- vehicle-crime
- possession-of-weapons
- shoplifting
- drugs
- public-order
- criminal-damage-arson
- burglary
- other-crime
- robbery
- other-theft
- violent-crime
I looked for correlations between these variables and the pct change of the labour vote share.
What did I find?
I did not find any significant correlations between any of the categories and the pct change of the labour vote share.
Key Question: Does crime affect vote share?
Theory Do different levels of different kinds of crime affect vote share? If the number of violent crimes go up in an area does this correlate with more votes for a certain party? If there are more robberies does that affect vote share?
Key Data: https://data.police.uk/data/ - street level data by longitude and latitude.
Task Start to look for correlations between crimes, different kinds of crime and the 2018 election results. The crimes are given by longitude and latitude so they will need to be converted into ward level data.
Creating a schema and transforming into a structured csv.
Row identifier 12