Citation
Correlations of Hate-Crime & Basic Linear Regression Model

Material Information

Title:
Correlations of Hate-Crime & Basic Linear Regression Model
Creator:
Leman, Ryan
Place of Publication:
None
Publisher:
None
Publication Date:
Physical Description:
Presentation

Notes

Abstract:
Hate-related crime & hate crimes, especially in 2010-2011, represent one of the most important human rights issues the US currently contends with. These issues have deeply troubled our society in many of it's facets including our media, mental health, politics, foreign relations, and even how policing is done. What would be very helpful towards creating data-based policy to aid law enforcement & our judicial system is if we could predict, using commonly gathered data, the tendencies of hate crime occurrences. The purpose with this project was to create a linear regressions model by using statistical methods such as Pearson's product-moment correlation to decide which variables were predictive for hate crime occurrences as recorded by either the FBI & Southern Poverty Law Center. The academic research question being, from the dataset published on github from "fivethirtyeight" named hate-crime.csv, do some of these variables correlate with increases/decreases in hate crime occurrences per 100k people. If so, can such variables be used to create an accurate linear regressions model for the purpose of estimating future increases/decreases of hate crime occurrences. Such model could be used to influence policy related to the allocation of resources for hate crime education, training for police & lawyers, and even influence which law-firms train & hire defense/prosecution lawyers for hate crime specialization among many other uses. What I found by running a series of person's product-moment correlations was that according to 2005 FBI records, correlated variables included: median household income, percentage of population who are not citizens, gini index score, percentage of population not white, then an inverse correlation with percentage of population who voted for previous president D.J Trump. An ANOVA analysis confirms theses correlations with the gini index being the most predictive of hate crime followed by median household income then percent of population who are not citizens and lastly percent of population who are not white. I then do the same process except for hate crime reported by the Southern Poverty Law Center. Pearson’s product-moment correlation findings show a correlation with median household income, percent of population above 25yr holding a high-school diploma, the gini index score, and again an inverse relationship with voters who voted for Trump. An ANOVA analysis confirms 3 of the 4 variables as being significant. From most to least significant are the gini index, median household income then voters who voted for trump. This work represents a start to what could be refined into a very robust predictive model for hate crime.
Acquisition:
Collected for Auraria Institutional Repository by the Self-Submittal tool. Submitted by Ryah Leman.
Publication Status:
Unpublished

Record Information

Source Institution:
Auraria Institutional Repository
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

Correlations of Hate Crime & Basic Linear Regression Model By Ryan Leman

PAGE 2

Hate Crimes & my questions Hate Crime: Represents a critical societal issue Effects many facets of our daily lives No good systematic way to handle the rise in hate crime Our questions: What demographic data may be correlated with hate crime Can these correlations be combined to create a predictive model Specifically , a linear regressions model.

PAGE 3

Correlations & Linear Regression Correlations Typically concerns only two variables Can be expanded upon to more complex statistical methods One such example being a Linear Regression . Linear Regression models Linear Regression are especially useful for creating predictive models. predictive model.

PAGE 4

Why Statistics Regarding Hate Crime Academic purposes To find correlates of increases/decreases in hate crime. A variety of demographic data will be looked at such as: average median income, Gini Index , percent population with high school diploma etc . Create a basic linear model from the correlates Data driven policy A Linear regression model can predict the increase/decreases in hate crime occurrences Projecting future hate crimes can help us fine tune policies regarding hate crimes occurrences & prevent future hate crime occurrences.

PAGE 5

Method First I started off with a data set found here: https ://data.world/fivethirtyeight/hate crimes/workspace/file?filename=hate_crimes.csv 1) I ran a pearson's product moment correlation on all variables against average reported hate crime incidence as reported by the FBI. 2) I ran the same test for hate crime reported by the Southern Poverty Law Center. 3) I include only the significant correlates in a linear regression model & verify the correlations using another correlation method using Rs built in ANOVA function.

PAGE 10

Creation of the Linear model FBI model: Hatec crime per 100k = 2.4 + median household income*(0.000019) + percent of non citizen*(20.9) + Gini Index*(17.2)+percent non white*( 5.7)+voters voted trump*(.24) SPLC model: Hate crime per 100k splc = 2.4 + median household income*( .0000078) + percent population with high school degree*(3.3) + Gini Index*(1.341) + percent voters voted trump*( 0.9)

PAGE 11

Results from final ANOVA Gini Index (p = .00042, F 6.2125, F = 14.7556) median household income (p = .017, F = 6.2125) population who are not citizens (p = .063, F = 3.64) percent of population who are not white (p = .062, F = 3.67) SPLC significant correlations: Gini Index (p = .0000018, F = 30.8 ) median household income (p = .0013, F = 11.9) voters who voted for trump (p = .0018, F = 11.1).

PAGE 12

Conclusion This analysis can be used for any agencies working with hate crime to better allocate resources for the education, prevention & recognition of hate crime Specific policies where this data is useful are policies involved in funding public programs through taxes how much funding? Should funding increase or decrease modifying how hate crime laws are taught to future lawyers in areas projected to have increases in hate crime . How much emphasis on the curriculum should be spent on hate crime laws? Should we support hate crime as a specialization within the legal field? give extra training for police in those areas projected to have high future hate crime rates. How to better identify hate crime. How to better report hate crime occurrences. How to interact with victims of hate crime.