Correlating crime and physical environment was a pet project adopted into my Bachelor’s thesis in 2012. It used 42 factors such as the literacy rates of wards, percentages of land uses present, presence of lake/park/public spaces, etc.

Introduction, Project Details and Data:

Feeling safe is a basic right and ranks 2nd in Moslov’s hierarchy of needs. When the basics of safety are addressed, we can expect growth and development in other areas as well. Increased crimes, dissuade development leading to neglect and inviting further crimes. While policing, law and order are important factors, there could also be physical aspects of the city, such as dark streets, not being illuminated by street lights, or the presence of a certain building density, etc. which could also contribute to making spaces viable for crime incidents. This was an attempt to understand associations of crimes with indicators of the physical environment. The city chosen for this was Bangalore to study this.

In 2010, Bangalore made up 1.44 % of India’s cases registered under IPC and 22.6% of Karnataka’s total registered cases while the population of Bangalore is only 13.8% of Karnataka. What makes Bangalore special is that it is the capital of the state, it is a ‘primate’ city (10 times the size of the next big city, Mysore) and growing at a rapid pace. The registered cases aren’t in proportion to its population.

Crime data of theft (highest numbers among all crimes), burglary, rape, kidnapping, murder, trafficking, gambling, narcotics and others were obtained from National Crime Records Bureau (NCRB) office, Land use maps of different planning districts were obtained from the Bangalore Development Authority (BDA), ward demographic data was obtained from Census of India, and several scores and physical/planning related indices were obtained from secondary source for indicators such as access to public toilets, % of green cover, cleanliness of roads, crowding in public transport, distance to closest park and playground, speed of travel (public and private transport), public amenities, street lighting, etc., which possibly had a relation to crimes.

The biggest hurdle was not the data but the boundaries of wards, and planning districts/police station zones not matching. Back then I did not statistical downscaling existed as a tool to disaggregate (even then it would have been a challenge - for such a short time), so I took all the wards that matched and took random samples (stratified) from three categories: high, mid and low crime incidences. Crime data, fortunately, had spatial information (in terms of which ward/police station the crimes occurred in). Along with the crime incidences, other criteria were high and low growth rates of crimes and people’s perception of safety study.

Correlation and Insights: Correlation of crimes with 42 physical/related factors was done using Excel and color-coded for easy reading. Correlation coefficients of value >0.6 or <-0.6 were colour-coded for the first stage as the correlation table was large. These were corroborated with land use maps (qualitatively) also to make sure the results obtained made sense.

Insights from the correlation coefficients and the overall study showed some commonly known associations while also giving counter-intuitive results. Some of them are shared below:

  1. Slum population had a positive correlation with the crimes - shows that while crimes occur in high-poverty areas, they are also often the victims of the crimes.

  2. population growth rate had no correlation to crime incidences.

  3. Higher negative correlation of narcotics crimes with areas with parks - places with parks are safer.

  4. Perception of safety is different than the actual incidence of crime - for a few wards, this could be as the wards area was large so there still may be pockets of safe and unsafe neighbourhoods within the ward itself; for others, it could be shaped from the media reports of crimes instilling the safety perception/fear in residents.

  5. Crimes themselves were correlated - where one of the crimes was high, others were high too. The Converse was also true for low crimes of one sort. The only exception is that of narcotics crimes not being correlated to other crimes.

  6. Areas with dead-end streets (cul-de-sacs) had lower incidences of crime - maybe due to the absence of escape routes for criminals.

Conclusion:

While a lot of factors affect crimes: from socio-economic, political (law and order situation) and urban design and planning factors, we could design better cities to make crimes less viable to occur. While this could be a great study to check especially with present methods and computation, this was a small attempt during my bachelor’s to use quantitative (mostly) and qualitative data to get to some associations using correlation.