increase in the overall level of Crime in the US

The study's main goal was to look into the elements that contribute to an increase in the overall level of crime in the United States. The study looked precisely at how the unemployment rate in the economy affects the overall level of crime in the country. However, the model incorporated a number of other variables such as education, ethnicity, GDP per capita, income level, gender, and political party.

The study takes into account cross-sectional and aggregated data obtained from all 50 states in the United States. The data was collected and aggregated to obtain national averages. It should be noted that one of the variables (Party Dummy) included in the model is a dummy variable. It indicates 1 for a year when the ruling party president was a Republican and 0 when the ruling party president was a democrat. The data used in the study was obtained from World Bank database, US Census website and Fed website.

Variables in the Model

Dependent variable: US Crime rates

Independent variables:

Education: Percentage of individuals that have attained at least tertiary education

Sex: The percentage of Male people in the population

GDP Per capita: The overall level of GDP divided by the entire productive population.

Income: The overall disposable income of individual households in the US

Party: The variable indicating the party affiliation of the ruling pity for instance Republican or Democrat.

Hypotheses

Null Hypothesis

There exists no significant relationship between Crime rates and unemployment in the US

Alternative Hypothesis

There exists a significant relationship between Crime rates and unemployment in the US

Level of significance: α = 0.05

Regression Results

From the description of the variables in the data above, it can be seen the data has a total of six independent variables against one dependent variable. Consequently, the best statistical tool to use in such kind of analysis is regression analysis. Using STATA (an advanced statistical tool), the six independent variables were regressed against the variable of Crime rate. The results of the regression analysis are showed subsequently.

Table 1: Results of the first regression with all the variables in the model

Source: STATA regression output

Table 2: Variance Inflation Factors (VIFs) for the first regression model

Source: STATA output for VIF

Table 3: Results of the second regression with all the variables in the model

Source: STATA regression output

Table 4: Variance Inflation Factors (VIFs) for the second regression model

Source: STATA output for VIF

Interpretation and Explanation

From the results of the first regression model, it is evident that the two variables of Income and Male (Sex) are highly collinear because they were associated with very huge VIF figures as shown table (1) and table (2) above. When the data has high collinearity, the results obtained in the model may be erroneous since they will be associated with huge confidence intervals. Thus, these two variables were dropped and a new regression model was run.

The final regression model only contains the variables of Unemployment, GDP per capita, political party dummy and the Education variable. The last regression model is the one adopted in the subsequent analysis because the independent variables it encompasses have negligible levels of VIF as indicated in table (4) above. From the regression model results in table (3), there exists a non-significant positive relationship between unemployment and crime rate in the country (B = 33756.14, P-value > 0.05). From this result, there is no sufficient evidence for rejection of the null hypothesis since the relationship between unemployment and crime rate is not significant enough at 95% level.

Furthermore, the data indicates that there is a very significant negative relationship between GDP per capita and Crime rates in the country (B = -115.0673; P-value < 0.05). That indicates a decrease in GDP per Capita would lead to an increase overall crime rates in the country. The data also indicates that governments with a republican president are likely to record atleast 327240 crimes more than those under a Democratic president and this relationship is significant (B = 327240.9; P-value < 0.05). Lastly, the data indicates as more people get tertiary education, the level of crime in the country is set to lower down (B = -69978.22; P-value > 0.05) though that relationship is not significant. The results also indicate that unemployment, GDP per capita, party affiliation and Education background explain 97% of the overall variation in overall crime rates in the country.