What's up: Hey, guys! I wanted to share with you a project from my Quality and Reliability class (ISE 4120) last semester. This is a report a friend and I compiled by pairing different categories of crimes statistics (burglary, murder, rape, etc.) with population for the 50 states of the USA. We received all of our data from the FBI crime database corresponding to 2013, but we actually had to omit Louisiana from the plot because it was such an outlier (I know where I won't be going anytime soon).
What's Important: I wanted to share this to show you how easily you can make your own statistics with Minitab. In these times of blatant misinformation, its important to be able to collect and verify your own conclusions. This is, in fact, the same technology that weight-loss fad companies use to show significance when their product really does nothing. With a couple careful data deletions or random normally distributed numbers, one statistical analyst could pull the wool over the eyes of millions! Don't worry, however, you'd be surprised how much data is freely available over the internet. Think something is too good to be true? Go check!
What did we find?: My group actually discovered a defensible correlation between murder rates, burglary rates, and population. That implies that when one of those variables increases, they other two do as well. This could mean that if we reduce burglary rates, we could also reduce murder rates. Now that what I call killing two birds with one stone! If you're interested the Minitab plots and data are provided in the study below. Please look it over and let me know what you think!
Crime Rates in the United States: Is there a Correlation?
By: Wyatt Fox and Morgan Bernard
ABSTRACT
This study analyzes data trends pertaining to crime statistics in the United States. The statistics were collected and catalogued by the Federal Bureau of Investigations (FBI). The data was received from the FBI’s database on violent crime organized by state.The factors are population and burglary. The key output variable monitored is murder. The results of the analysis of the data indicates that with an increased population and burglary rates, murder rates also increase. Furthermore, the model and analysis pass all required diagnostic tests.
1 INTRODUCTION
The main objective of the study depicted in this paper is to develop an empirical model to help predict murder rate as a function of population and burglary rate. The FBI defines burglary as the unlawful entry of a structure to commit a felony or theft. The FBI defines murder as the willful killing of one human being by another.The data used to derive this model was obtained from the FBI database and is described in Table 1. Figure 1 shows the resulting Minitab analysis information. Figure 2 provides the normal probability plot of the residuals. The remaining sections of this report are organized as follows. In Section 2, the information received from the FBI database is presented. In Section 3, model estimates of the first order regression fit and a normal probability plot of the residuals derived from the data are evaluated. In Section 4, the fitted model derived from Minitab is displayed. Finally, in Section 5, there is a discussion of the implications of the evaluated data.
2 CRIME IN THE UNITED STATES IN 2013 DATA FROM THE FBI DATABASE
In this section, the data retrieved from the FBI crime database is described. The specific table from the database organized crime in the United States by state and displays data from both 2012 and 2013. The table also included the population of each area, a rate per 100,000 people for each crime for each area, and a percent change from 2012 to 2013. In the interest of analysis, the original table was significantly modified in order to create a flat file compatible with Minitab. All the information for 2012, the percent change values, and some non-state areas were deleted from the original table to create the flat file. The table that was used for Minitab analysis is described in Table 1 below.
Table 1: Crime Data in the United States in 2013 (organized by state) - partial table
3 MODEL SEARCH
3.1 FIRST ORDER MODEL INVESTIGATION
The murder rates per state can be predicted using the first order model found in Figure 2 with population and burglary rates as significant factors. In order for the model to be defensible, it must possess an R-squared value that is greater than or equal to 30, VIFs (variance inflation factors) that must be less than or equal to 10 for all tested factors, and the residuals must be normal, meaning that no outliers are present in the data set. Based on our results displayed in Figure 2, we can conclude that our R-squared value is quite high at 94.93% and that our VIF’s do not exceed 10. Additionally, our tested factors have p-values that are less than .05, indicating that they are significant to our model. As shown in our normal probability plot of residuals in Figure 3, no outliers are present in our data, meaning that our data passed this test. Therefore, we can conclude that our model is defensible.
Figure 1: Model Estimates for First Order Regression Fit
Figure 2: A normal probability plot of residuals for a first order model of murder rates.
4 THE FITTED MODEL
The predicted model for murder rates is defined as -8.5 + 0.000032 Population + 0.002167 Burglary. This model depicts murder rate as a function of population size and burglary rates with a constant of -8.5. Burglary is shown in this model to have a higher effect than the population. The surface plot showing population size and burglary rates as a function of murder is displayed in Figure 3.
Figure 3: Surface plot depicting murder as a function of population size and burglary rates.
5 DISCUSSION
This report analyzes the factors affecting murder rates for states across the United States. The data was extracted from the FBI’s website. The predicted results from our model indicates that murder is a function of burglary rates and population size. Of the two significant factors found, burglary rates had more of a significant effect on murder rates than population size. Moreover, an increased population size and burglary rate increases a state’s risk for higher murder rates. By decreasing burglary rates, there could be a possibility that murder rates can also be decreased. While population size cannot be altered, adding additional safety measures such as a higher number of police officers to highly populated areas could help decrease burglary rates which in turn will decrease the number of murders occurring in a given state.
ACKNOWLEDGMENTS
United States. U.S. Federal Bureau of Investigation. Crime in the United States. N.p.: U.S. Department of Justice, 2013. Web. 21 October 2017.
Morgan Bernard is an undergraduate student pursuing a degree in Industrial and Systems Engineering at Ohio State University. She is specializing in their Operations Management track and is the President of the Society of Manufacturing Engineers at Ohio State University.
Wyatt Fox is an undergraduate student pursuing a degree in Industrial and Systems Engineering at Ohio State University. He is specializing in the Manufacturing Track and is currently employed at Covert Manufacturing.
Thanks for reading! Leave a comment, let me know what you think! Follow me for more eye-opening posts to come!