Violent Crime in America ECON 240A Group 4 Thursday 3 December 2009 Table of Contents Introduction Data Descriptive Statistics Statistical Analysis QuickTime og en dekomprimerer kreves for se dette bildet What are the causes crime of Our team hypothesized that there may be five factors contributing to the prevalence of violent crime in a specific jurisdiction Public Expenditures on Law Enforcement and Public Safety Public Firearm Ownership Education Income Ethnicity Our Measures of Violence Violent crimes per 100 000 people QuickTime og en dekomprimerer kreves for se dette bildet Data From the 50 States and DC Education Percentage of public high school freshman going on to graduate high school Poverty Per capita income Public Spending per capita expenditures on state and local law enforcement and corrections Ethnicity percent of population that is non white Firearms percent of households that own guns Freshmen that Graduate HS Cost of State and Local Law Enforcement Guns Per Household Income Per Capita Percent Non White Minorities Freshmen that Graduate HS Cost of State and Local Law Enforcement Guns Per Household Income Per Capita Percent Non White Minorities Negative Correlations k e r e v Q u c i k i T me o g e e d k mp o r me i e r r s f o r s e d e t e i b d l n k e r e v Q u c i k i T me o g e e d k mp o r me i e r r s f o r s e d e t e i b d l k e r e v Q u c i k i T me o g e e d k mp o r me i e r r s f o r s e d e t e i b d l k e r e v Q u c i k i T me o g e e d k mp o r me i e r r s f o r s e d e t e i b d l t e t e t e t e n n n Freshmen that Graduate HS and Percent Non White Minorities Expenditures on State and Local Law Enforcement and Guns Per Household Guns Per Household and Income Per Capita Guns Per Household and Percent Non White Minorities Positive Correlations Cost of State and Local Law Enforcement and Income Per Capita Cost of State and Local Law Enforcement and Percent Non White Minorities Correlations to Violence Positive Percent Non White Minorities Income Per Capita State and Local Law Enforcement Expenditures Negative Guns Per Household Freshmen that Graduate HS All negative and positive correlations are statistically significant Violent Crimes vs Ethnicity H 0 t0 025 50 2 009 n 51 H 1 t0 025 50 2 009 5 Percent of population that is non white is a significant explanatory variable for violent crimes per 100 000 capita Dependent Variable VIOLENTCRIMEPER Method Least Squares Date 12 02 09 Time 13 25 Sample adjusted 1 51 Included observations 51 after adjusting endpoints Variable Coefficient Std Error t Statistic Prob PERCENTNONWHITE 9 250446 1 547537 5 977530 0 0000 C 191 0509 47 71586 4 003929 0 0002 R squared 0 421698 Mean dependent var 431 6078 Adjusted R squared 0 409896 S D dependent var 238 3361 S E of regression 183 0855 Akaike info criterion 13 29621 Sum squared resid 1642495 Schwarz criterion 13 37197 F statistic 35 73086 Prob F statistic 0 000000 Log likelihood Durbin Watson stat 337 0533 1 972195 Violent Crimes Regressed Against Possible Factors Data Possible Factors R squared t Statistic Non whites 0 421698 0 0000 Income per capita 0 098054 0 0253 Guns per household 0 108723 0 0181 Expend on public security 0 333322 0 0000 Freshmen to graduate HS 0 322653 0 0000 Multiple Regressions Average freshman grad and expenditure per capita are significant Depende nt Variable VIOLENTCRIMEPER Method Least Squares Date 12 02 09 Time 14 16 Sample adjusted 1 51 Includ ed observations 51 after adjus ting endpoints Coefficient Std Error t Statistic Prob Households with guns are no longer significant AVGFRESHMANGR 1402 599 AD EXPEDITURECAPIT 0 570637 A HOUSEHOLDGUNS 252 7796 C 1285 674 314 2518 4 463295 0 0001 0 174902 3 262603 0 0021 193 8526 309 1428 1 303978 4 158834 0 1986 0 0001 R squared 53 5 Variable R squared Adjusted R squared S E of regression Sum squared resid Log likelihood Durbin Watson stat 0 535028 0 505349 167 6252 1320615 331 4913 2 134995 Mean dependent var S D dependent var Akaik e info criterion Schwar z criterion F statistic Prob F statistic 431 6078 238 3361 13 15652 13 30804 18 02712 0 000000 Regression Diagnostic 12 Series Residuals Sample 151 Observations 51 10 The Jarque Bera statistic suggests that the residuals plot are normally distributed 8 6 4 2 0 300 200 100 0 100 200 300 400 500 Mean Median Maximum Minimum Std Dev Skewness Kurtosis 1 69E 13 19 93613 538 9753 346 9 274 162 5186 0 502499 4 262888 Jarque Bera Probability 5 535431 0 062805 Multiple Regressions Dependent Variable VIOLENTCRIMEPER Method Least Squares Minority group is still significant in explaining violence per capita Income per capita is not a significant explanatory variable Regression is significant Prob Fstatistic 0 000 Date 12 02 09 Time 14 10 Sample adjusted 1 51 Included observations 51 after adjusting endpoints Variable Coefficient Std Error t Statistic Prob PERCENTNONWHITE 8 679604 1 597096 5 434616 0 0000 PERCAPITAINCOME 0 006128 0 004682 1 308896 0 1968 C 1 632381 154 6450 0 010556 0 9916 R squared 0 441628 Mean dependent var 431 6078 Adjusted R squared 0 418362 S D dependent var 238 3361 S E of regression 181 7674 Akaike info criterion 13 30036 Sum squared resid 1585891 Schwarz criterion 13 41399 F statistic 18 98207 Prob F statistic 0 000001 Log likelihood Durbin Watson stat 336 1591 2 082096 Regression Diagnostic 12 Series Residuals Sample 1 51 Observations 51 10 8 Mean Median Maximum Minimum Std Dev Skewness Kurtos is 6 4 2 J arque Bera Probability 7 13E 14 25 67716 503 4726 581 7 378 178 0950 0 079576 4 774282 6 743485 0 034330 0 600 400 200 0 200 400 The Jarque Bera p statistic suggests that the residuals are not normally distributed Data Issues Residuals Regressions Residuals plotted against the fitted violent crime per capita Results from the White Heteroskedasticity for violent crime per capita regressed against expenditures on state and local law enforcement per capita White Heteroskedasticity Test F statistic 20 93164 Probability 0 000000 Obs R squared 23 75862 Probability 0 000007 Crime vs Expenditures DC Dummy Dependent Variable VIOLENTCRIMEPER Method Least Squares Date 12 02 09 Time 23 17 Sample adjusted 1 51 Included observations 51 after adjusting endpoints Variable Coefficient Std Error t Statistic Prob DCEXPENDITUREDUM 619 4809 185 5877 3 337941 0 0016 EXPEDITURECAPITA 0 770615 0 046868 16 44227 0 0000 R squared 0 456163 Mean dependent var 431 6078 Adjusted R squared 0 445064 S D dependent var …
View Full Document
Unlocking...