Abstract
Research questions: is there strong relation between race, sex and the parents who have served in prison and juvenile crime level?
Methods: statistics methods (descriptive analysis, regression Chi-square and cross-tabulation representation of data, visual analyses).
Results: It was determined that chosen factors have effect on juvenile crime level but their weight is pretty small.
As a conclusion it can be sad that race does not affect juvenile crime level moreover chosen variables have small impact on juvenile crime.
Introduction
Juvenile crime level is a big problem in modern world. The goal of this paper is to determine characteristics of people who have committed crime or delinquent acts as juvenile. This will allow developing policies targeted on the most important factors and as a result significantly decrease Juvenile crime level.
- Univariate analysis
- Bivariate analysis
- Multivariate analysis
I assume that juvenile crime level is strongly dependent on variable listed in the table 1.
Variables used in the research.
Results
I will start my analysis from univariate analysis of dependent variable (Online Statistics Education: A Free Resource for Introductory Statistics. n.d.).
According to the table 2 only 294(or 5.74% of respondents) committed crime as teenagers.
Juvenile crime level in numbers.
Data provided in tables 2 is not enough to make any conclusion therefore we need further deeper analysis. We will start from univariate analysis.
Univariate analysis.
Descriptive statistics on researched variables provided in table 3.
Descriptive statistics on chosen variables.
Bivariate analysis
Next step is to determine dependencies between juvenile crime level and every independent value. To do this in accordance with the categorical nature of the variables I will start from cross-tabulation.
Race variable influence on juvenile crime rate. According to table 4 the smallest rate of juvenile crime is among Asian and the biggest is among American Indian with the 3.2 times range. However, I would like to point out that American Indian sample is too small compare to others. To determine whether there is any kind of relationship regressions (simple and logistic) should be build.
Cross-tabulation with chi-square of juvenile crimes dependency on race.
Simple regression. According to the summary of regression provided in the table 5 it is clear that race variable does not have significant relation to juvenile crime level. Moreover quality of developed model is bad that is shown by R (although model if valuable: F> crit F). Next step is to check the conclusion using logistic function (Online Statistics Education: A Free Resource for Introductory Statistics. n.d.).
Simple regression 1 summary.
According to received data race is not significant variable and the model is of a bad quality due to low level of R and R^2. Var1 R= .01775146 R? = .00031511 Adjusted R? = .00011937 F (1, 5107) =1.6098 p<.20458 Std.Error of estimate: .23289
Logistic regression. Summary proves that race variable is not significant.
Logistic regression 1 summary.
Sex variable influence on juvenile crime rate.
According to table 8, males commit juvenile crimes 4 times more often than women. At such strong numbers there should be some kind of relationship.
Cross-tabulation with chi-square of juvenile crimes dependency on sex.
Simple regression. According to the summary of regression provided in the table 9 sex variable is significant. However quality of the model remains bad. R reaches level of 0.17.
Simple regression 2 summary.
Dependent: Var1 Multiple R: .17988583 F = 170.9505 R? .03235891 df = 1,5112
No. of cases: 5114 adjusted R?: .03216962 p = 0.000000 Standard error of estimate: .229022711 Intercept: .186861102 Std. Error: .0104001 t( 5112) = 17.967 p < 0.0000
Logistic regression. Summary in the table 10 confirms that sex is significant variable but the model is of a bad quality due to low level of R and R^2. Constant value is excluded from the equation due to insignificance.
Logistic regression 2 summary.
H4WP3+ H4WP9 (bio mother or father has served time in prison) variable influence on juvenile crime rate.
According to table 8, almost 21% of people with parents served in prison were charged in juvenile crimes. Another tendency is that people with father in prison are 2 times more often were charged in juvenile crimes.
Simple regression. According to the results provided in table 12 variable H4WP3+ H4WP9 is significant. However quality of the model remains bad. R reaches level of 0.13. Logistic regression data from table 13
Simple regression 3 summary.
Regression Summary for Dependent Variable: R= .13108097 R?= .01718222 Adjusted R?= .01698996 F(1,5112)=89.371 p<0.0000 Std. Error of estimate: .23081
Logistic regression 3 summary.
Model: Logistic regression (logit) N of 0's: 4820 1's: 294 (Spreadsheet2.sta) Dep. var: Var1 Loss: Max likelihood (MS-err. scaled to 1) Final loss: 1091.2839492 Chi?( 1)=67.617 p=.00000
According to received data race is significant variable but the model is of a bad quality due to According to conducted regression all independent variables except race are significant.
Multivariate analysis. At this point we will combine all variables in order to determine multiple relations. Due to the dependent variable nature logistic model will be used. Logistic model results presented in (table 14).
Multivariate analysis logistic regression summary.
Model: Logistic regression (logit) N of 0's: 4815 1's: 294 Dep. var: Var1 Loss: Max likelihood (MS-err. scaled to 1) Final loss: 998.83808209 Chi?( 3)=251.92 p=0.0000
As we can see from the received data in multiple regression Var1 (Race) is not significant. P-value(0.213477) is too high.
According to received results it is clear that multiple regression has better quality than any bivariate because we have received the highest R parameter of 0.225. However quality of model remains bad.
As a conclusion it can be sad that race does not affect juvenile crime level moreover chosen variables have small impact on juvenile crime.
References.
Online Statistics Education: A Free Resource for Introductory Statistics. (n.d.). Online Statistics Education: A Free Resource for Introductory Statistics. Retrieved June 25, 2014, from http://onlinestatbook.com/2/
