Presentation is loading. Please wait.

Presentation is loading. Please wait.

Crime in the USA: Using SAS to Analyze Recidivism Rates

Similar presentations


Presentation on theme: "Crime in the USA: Using SAS to Analyze Recidivism Rates"— Presentation transcript:

1 Crime in the USA: Using SAS to Analyze Recidivism Rates
Philip Mayevskiy: San Francisco State University

2 Annual Parole Survey It is released by the Bureau of Justice Statistics (BJS) The latest release was the 2016 dataset This presentation will be analyzing the 2014 Annual Parole Survey

3 2014 Annual Parole Survey The observations are listed by state
Include almost 170 variables These include population sizes, demographics, types of offenses, and number of parolees who reoffend among other variables The BJS include a Codebook with descriptions of all the variables

4 Cleaning the Data To clean this dataset you must
Get rid of missing or incomplete data Make the variables easily interpretable by SAS and the reader First, convert and remove the missing values: All missing values are represented by either a ‘-8’ or ‘-9’ DATA datain.Parole; SET datain.da36320p1; format _numeric_ 9.4; IF (EXINCNEW = -8 OR EXINCNEW = -9) THEN EXINCNEW = ‘.’; RUN ; Supplementary Syntax available from icpsr.umich.edu website

5 Cleaning the Data (cont.)
Interpretability We can start with the variables in question We want to find recidivism in the parolee population We will use ‘Returned to incarceration with new sentence’ (EXINCNEW) and ‘… with no new sentence’ (EXINCREV) Since the variable is a population we have to make the population relative to the state population RRATE = ( EXINCNEW + EXINCREV ) / TOTBEG ;

6 Cleaning the Data (cont.)
Interpretability (cont.) Some of the RRATE values are also missing We can remove these states by the STATEID variable All the states that were deleted included CA, CT, NH, NM, MS, WA, WI, and DC.

7 Cleaning the Data (cont.)
Interpretation (cont.) Next, we could format the rest of the population variables in this way Starting with demographic data: Now the rest of the demographic and/or population data can be converted with this method

8 Cleaning the Data (cont.)
Interpretation – Categorical Data Binary data in this dataset is marked as ‘1’ and ‘2’ Using the GPS variable, we can format it in the following way: We do this because we need ‘0’ and ‘1’ in order to do ANOVA testing on it

9 Analyzing the Data Demographic Data
We can use forward, backward, and stepwise selection on this data Doing this on the White, Black, Hispanic, and Asian variables: The selection summary result was the following: Most significant seemed to be Black or African American The p-value for the RRATE = BLACK regression is with a negative coefficient

10 Analyzing the Data (cont.)
Demographic Data (Cont.) Fit Plot which is an output with the PROC REG function:

11 Analyzing the Data (cont.)
Demographic – Gender Next we can use PROC REG on gender We would use the model for RRATE = FEMALE This model shows that the higher the percentage of women in parole, the higher the recidivism

12 Analyzing the Data (cont.)
Variables of Interests Running the same forward, stepwise, and backwards selection none of the variables were significant The most significant variable was ‘supervised out of state’ with a p-value of 0.10 We can use a custom Fit Plot to visualize the data

13 Analyze the Data (cont.)

14 Politics of Recidivism
The Henry J. Kaiser Family Foundation offers data on political party by state We can set the values accordingly: Plotting this will give a visual to the recidivism rates by state and their political preference

15 Politics of Recidivism (cont.)
We need to plot this with respect to recidivism rate, state, and political preference:

16 Politics of Recidivism (cont.)

17 Politics of Recidivism (cont.)
We can analyze variability by Population:

18 Overall: Analyzing this data could give great insight into the problem
However, this data is very general Finding out WHY something is will take either more data or specialized data

19 Contact Information Philip Mayevskiy San Francisco State University Phone: (415)


Download ppt "Crime in the USA: Using SAS to Analyze Recidivism Rates"

Similar presentations


Ads by Google