The Counted

IS 607 Final Presentation

Matthew Farris

counted

Question and Reasons:

  • Does the size of a Police Force influence the number of Police related killings?

  • Reasons behind this topic

    • Strong movement for social change
    • Need for Accountability
    • Data can provide valuable resources

freddie

Data Sources

  • The Counted

    • Database for Police Fatalities
    • Crowdsourced
    • Journalists Verify Data

  • Census Data

    • Government Estimates
    • Accurate based on 5-Year Trend

  • Police Size

    • FBI data by most populous city
    • Summarized by Governing (for ease)

scott

Data Sources

  • Counted - CSV
  • Police Size - Webscrape (Table)
  • Census Data - API

Data Transformation

  • The Counted

    • Summarized by state
    • Joined with Census Data

  • Census Data

    • Converted State to Abbr.
    • Cleaned white spaces
    • Factor Transformation

  • Police Forces Size

    • Separated State Name from City
    • Summarized the data by state
    • Number of police per 10k Citizens

Summary Info

##      State      Fatalities       Population       Fatal_Rate_Per_Mil
##  AK     : 1   Min.   :  1.00   Min.   :  575251   Min.   :0.9494    
##  AL     : 1   1st Qu.:  5.00   1st Qu.: 1855617   1st Qu.:2.2709    
##  AR     : 1   Median : 17.00   Median : 4601049   Median :3.4988    
##  AZ     : 1   Mean   : 21.69   Mean   : 6384632   Mean   :3.6305    
##  CA     : 1   3rd Qu.: 22.00   3rd Qu.: 6899123   3rd Qu.:4.3765    
##  CO     : 1   Max.   :194.00   Max.   :38066920   Max.   :9.6888    
##  (Other):43                                                         
##  Officer_Per_10k
##  Min.   :11.17  
##  1st Qu.:15.10  
##  Median :18.89  
##  Mean   :19.31  
##  3rd Qu.:22.03  
##  Max.   :42.60  
##

Visualization

## Loading required package: lattice
## Loading required package: plyr

plot of chunk unnamed-chunk-2

Modelling

plot of chunk unnamed-chunk-3

Summary

## 
## Call:
## lm(formula = Fatal_Rate_Per_Mil ~ Officer_Per_10k, data = counted_combined)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2.5785 -1.5646 -0.2569  0.8348  5.8116 
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)    
## (Intercept)      4.99707    1.23727   4.039 0.000202 ***
## Officer_Per_10k -0.07219    0.06383  -1.131 0.263903    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2.042 on 46 degrees of freedom
## Multiple R-squared:  0.02706,    Adjusted R-squared:  0.005906 
## F-statistic: 1.279 on 1 and 46 DF,  p-value: 0.2639

Interpretation

  • Modelling

    • There was not enough evidence to reject the Null Hypothesis.
    • Linear Model was not the "Best Fit"
    • Alternate models would have made a better choice

  • Looking Forward

    • State vs. City Data
    • Cross referencing Communities
    • This is Not Over