In order to quantify the relationship between crime severity and each of the poverty, food insecurity, income, and unemployment indicators, we calculated a crime score.
Calculation of Crime Score
To calculate the crime score, we factored in the severity of the crime and the weapon implicated in the crime report. Each type of crime was assigned a value based on the perceived severity of the crime: homicide (10), assault with dangerous weapon (7), sex abuse (6), arson (5), robbery (3), burglary (3), motor vehicle theft (2), theft/other (1). We reasoned that types of crime that involved direct bodily harm were more severe than thefts or bulgarlies, which do not involve person-person interactions. We also factored in the weapon involved in the crime report, which were identified as “Gun,” “Knife,” or “Others” and given a score of 5, 3, and 1, respectively.
We decided to bin each attribute in somewhat unconventional way. We determined the unique values of each attribute in the whole dataset and ordered them from least to greatest. We decided to bin the unique values into 5 equal parts. We wanted to consider the difference in the data entries for each bin, but did not want to analyze a skewed dataset where most of the entries were allocated in the lower bins. The bins represent the different tiers of median income, poverty rate, unemployment rate, and food insecurity rate in increasing order (1 being the bin with crimes that occurred in areas with the lowest values for each attribute, 5 being those with the highest). For every crime report, the score was incremented by one. For more severe crimes, the score was incremented by more based on the aforementioned values assigned to the type of crimes and the type of weapon used. Then, we determined how much each bin contributed to the total crimes score by computing the bin crime score over the total crime score for each attribute.
Pie Charts: Relationship Between Crime Reports and Attributes
We hypothesized that the highest bins for poverty rate, unemployment rate, and food insecurity rate would contribute more to the total crime score, while the lowest bins for median income would contribute more to the total crime score. If the proportion of each bin’s crime score of the total crime score is higher, we concluded that there is a greater correlation between the level of poverty, food insecurity, income, and unemployment associated with that particular bin and the severity of the crime.
The bins with the lowest median income comprise the largest proportion of total crime score (50.9% for bins 1 and 2). For unemployment rate, bins 4 (22%) and 5 (21.6%) contribute the most to the total crime score. These results indicate that bins with the highest unemployment rate comprise the largest proportion of the crime score, which suggests that these bins contain more severe crimes. For poverty rate, bins 3 (22.3%) and 4 (25.3%) contribute the most to the total crime score. For food insecurity rate, bin 4 (27.6%) contributes the most to the total crime score. The relationship is not as clear with respect to poverty rate and food insecurity, given that we expected to see the highest bins with the highest crime score, but the results indicate that the the higher bins (3 and 4 for poverty rate, 4 for food insecurity) still do contribute more to the crime score compared to lower bins. These results provide quantitative evidence that there is a relationship between the most severe crimes and high unemployment, poverty, and food insecurity rate and low median income.
Standard Deviation of Crime Scores
We then calculated the standard deviation for each attribute. Within each attribute, we calculated the standard deviation of the crime scores for each of the bins (1-5) compared to the mean crime score. Higher standard deviation indicates there is more variability among the crime scores and that there are likely greater differences between the crime scores for each of the bins. The following is the result:
MEDIAN_INCOME STDEV: 5306.159882626983 (5.83%)
UNEMPLOYMENT STDEV: 2264.0393547816257 (2.49%)
POVERTY_RATE STDEV: 4546.297801508387 (5.0%)
FOOD_INSECURITY_RATE STDEV: 4400.371825198412 (4.84%)
Median income recorded the highest standard deviation rate of 5.83. Although standard deviations are not drastically high, we can observe that there are differences among each bin, where median income seems to show the most variability among the other attributes, suggesting that median income may be a better metric to assess severe crimes compared to other attributes. However, there are limitations to our analysis; our calculations of the crime scores are very subjective, given that we designated certain offenses as more severe than others based on our own perceptions and arbitrarily assigned values to each of the offenses. With the results from the crime score calculations and the standard deviation of crime scores, we can infer that there seems to be a correlation between severity of crime reports and higher unemployment, poverty, and food insecurity rate and lower median income.