CRJU 525: Advanced Statistics for Criminal Justice

Presentation Instructions (worth 25% of course grade)

Syllabus Description:

In order to facilitate learning through the application of course concepts, you are to complete a 15 to 20-minute presentation applying univariate, bivariate, and multivariate statistics to a dataset of your choosing. You will need to choose a research question(s) and apply the appropriate statistical techniques to answer your question(s). External research on your topic of interest will also be essential. You will need to turn your Presentation in for grading.

Presentation Instructions:

Step 1: Choose your topic of interest. This will likely be related to your thesis topic if you are completing a thesis, or similar to previous papers you have written if you are completing a comprehensive exam.

Step 2: Identify your dataset. If you have already collected your thesis data, or you have obtained your thesis dataset from a secondary source, then you are ahead of the game. If not, then visit https://www.icpsr.umich.edu/icpsrweb/landing.jsp and begin searching for datasets that are related to your topic of interest. You might also find a good database using the Google Database search tool: https://datasetsearch.research.google.com/. Please make every attempt to locate a dataset that is less than 20 years old. If you have any trouble downloading the SPSS data, please contact me. Note that some SPSS datasets will be saved as .por data. You may need to instruct your computer to open these files with SPSS, or open SPSS first, and then click to open .por files.

You should also download the codebook as well. You may use the GSS dataset if you like.

Step 3: Construct a few possible research questions.

Step 4: Identify variables that are appropriate operationalizations for your dependent variable (DV). For example, if my topic of interest is police officer use of force, then I would look in my codebook and search for variables that measure police officers using force on a citizen. I might choose one of the following:

V56 In any of these contacts with a police officer, did any officer warn you that he or she would use physical force such as: a nightstick or baton, a firearm, a chemical spray, a flashlight, a police dog, or any device other than handcuffs to restrain you or to take you into custody?

V57 In any of these contacts with a police officer, did any officer actually use any form of physical force against you including using any of the items just mentioned?

V58 How many times did the police actually use or threaten to use force against you during the last 12 months?

For this class, your DV must be scale level or dichotomous, so recoding might be necessary if you have a nominal or low-order ordinal DV (less than 7 categories).

Step 5: Identify variables (IV) that are appropriate operationalizations for your independent variable(s). Your independent variables should be chosen with your guiding theory in mind. You can be very deductive and test a theory by choosing variables that your theory suggests effect your DV, or you can be more inductive and play around with your data and see what variables significantly associate with your DV and then pick a theory that helps explain these associations.

Remember that your IVs need to be at the scale or high-order ordinal level of measurement to include in a regression model. You may need to dummy code nominal or ordinal variables to suit this requirement.

Step 6: Identify your control variables. Control variables are variables you think might associate with your DV, but are not part of your guiding theory. These often include demographics which have been shown to associate with your DV (i.e., gender, age, employment, ethnicity, etc.) or variables that represent alternative theoretical explanations. Dummy coding may also be necessary for control variables.

Step 7: Construct your hypothesis or hypotheses. Based on your guiding theory, construct hypotheses predicting the relationship between your DV and IVs.

Step 8: Univariate analysis. You need to describe each of your variables through frequency tables and/or graphs.

Step 9: Bivariate analysis. Next, you should conduct bivariate analyses to see how your IVs and control variables associate with your DV. This will take the form of a correlation matrix (refer to your notes, readings, and lab chapter from Week 6 & 7). During this step you also want to check for intercorrelations, or IVs and control variables that have a Pearson’s correlation of over .70. If the correlation is over .7, then they are too closely related (measuring essentially the same thing) and will steal each other’s explanatory power in your multivariate model. In this instance, you must pick one of the pair and remove the other.

In your presentation report on the significant correlations. Provide explanations referring back to your guiding theory.

Step 10: Multivariate analysis. In this very important step, you must choose the appropriate multivariate regression technique. For this class, you must pick between multiple linear regression or binomial logistic regression (your choice depends on your DV level of measurement). Correctly interpret your findings. Refer to your notes, readings, and lab readings from Weeks 10 and 12.

Step 11: Discussion. After your analysis is complete, you must discuss your findings referring back to your guiding theory. Did your results support your hypotheses? Was your guiding theory supported? What are some possible policy implications of your findings? Are there limitations to your research? Where would you like to take future research?