CJ 605 Spring 04 Advanced Statistics in Criminal Justice:
Assignments

NOTE.  These assignments are subject to change, so it is best probably NOT to get started on a big assignment without a) checking here first for any updates and b) coming to the class prior to the due date to get details on what is expected

DIRECTIONS FOR ALL ASSIGNMENTS:
1. Typed, double spaced.
2. Bring two copies to class, one to hand in, one to keep with you in class.
3. On the upper right of EACH page of each assignment put YOUR NAME, AND THE DUE DATE. 
4. To get FULL CREDIT for each of these I need these turned in on the day of class.

DUE DATE APPEARS BELOW ON THE LEFT
2/2 GOALS:
1) To create an aggregate (in HLM parlance, a Level 2 data file) from an individual level survey file.
2) To compare the ecological variance to the total variance for a series of survey items
IN PREPARATION for this you will want to carefully read the PHMC documentation files. The Philadelphia neighborhood identifier variable is NBH. There are 45 Philadelphia neighborhoods.

Using the SPSS aggregate procedure, create a neighborhood level file of the adult  2002  PHMC data file:

hs02ar1_tract&community.sav

SAVE your work in a syntax box.
YOU WILL BE AGGREGATING MORE FILES LATER so you are going to need this syntax box.
Remember to
a) select only those cases in Philadelphia
b) turn on the appropriate weighting variable -- the household balancing weight (SEE DOCUMENTATION)
c) sort the cases by the neighborhood variable (NBH) for Philadelphia (hint: there are 45 different neighborhoods)
d) save the new weighted, Philly only individual level file
e) create and save the aggregate file

You really want to try and do this by yourself but if you get stuck >> cj605_0226.sps
Once you have the aggregate file, compare the total variance and the neighborhood level variance for the items listed in the

 following spreadsheet (Click here)

I have put in the numbers for the first variable - there is also a formula you can copy DOWN and it will calculate.
 

WHAT YOU ARE TRYING TO GET TO IS: for several specific variables, and looking only at the portion of the sample living in Philadelphia, an estimate of what percent of the total variance for that item arises from between neighborhood differences. 

Once you have these numbers: pick one variable, and write about what is going on with it, descriptively and theoretically,  IN THE CONTEXT of the LISKA reading.

WHAT YOU WILL TURN IN: your spreadsheet, with your name on it, and your attached paragraph, with your name and the date.
2/9 Working with one of the variables in your spreadsheet from last week that is NOT binary:
1. Use the Level 1 and Level 2 files you created the preceding week
2. In HLM create the  SSM file. Make sure it is weighted by the household balancing weight. It will include, in addition to the group identifier variable (NBH), only two variables at Level 1 (individual level): the variable you chose, and the weighting variable. At Level 2 (aggregate or neighborhood level) it will include only one variable: the variable you chose.
3. Bring the SSM file into HLM and complete the one-way anova.
4. BRING THE SSM FILE, THE COMMAND FILE AND THE OUTPUT FILE to class ON DISK
5. Bring 2 copies of the HLM output on PAPER and we will go through it.
6. Put your name and SSN and DATE at the top of the output you have generated. Hand in one copy to get credit. We will go through these in class.
2/16 Using your one way anova output from last week, and using plain English sentences as best you can, tell me about: the grand mean; the variance of the neighborhood means of the outcome; the Level 1 variance; and identify the proportion of the variance at L2; also, tell me what the latter means. OPTIONAL FOR THOSE WHO ARE FEELING ADVANCED: tell me what the Chi square test of the Level 2 variance means, identify the intraclass correlation and tell me what it means.
2/23 GOAL: to begin to develop your Level 1 theoretical model.

Pick an outcome from the file. Possible outcomes include
behavioral restriction (Q57)
health status (Q1)
high blood pressure (doctor ever told) (Q38)
self rated stress (Q63)
depression (elderly file only) (DSCALE1)
diagnosed mental health condition (Q60)
receive treatment for mental health condition (Q61)
days per week exercise (Q50)
Trust neighbors (Q74)
sense of community (Q75)
Total health visits (Q13a+Q13b+Q13c)
N days mental health not good (Q3)
N days physical health not good (Q2)

Develop your level 1 model statement; identify which individual-level factors will predict your outcome. For each predictor we need a clear hypothesis and a RATIONALE.

To help you begin you want to be drawing on a literature review. I have provided starter references  in the reference list. YOu should be using the Web of Science to locate additional references. It is also a good idea to include cited references to go with each of your predictors so we can see how your model is theoretically grounded.

DATA NOTES:
* If you want to use the senior file 60+, because you are interested in the depression scale, do NOT weight the file; use it in unweighted form
* You are going to need to recode variables in the L1 file so you can have "clean" dummy variables for things like nonwhite, married, female, etc. All your dummy variables should have the values 0 or 1 and the variable name should correspond to the group that scores 1.

Run an ANCOVA where you put in ONE L1 demographic predictor (for example: gender or white vs. nonwhite or age or education or marital status or household size or employment status, etc). In a short paragraph interpret the b weight and the t test of the b weight

SO YOU WILL TURN IN: your theoretical Level 1 model statement, along with a reference list; your ANCOVA output; a paragraph to go with your output. BRING 2 COPIES OF EVERYTHING TO CLASS
3/1 No assignment due
3/8 Break
3/15 Allow one of your demographic, Level 1 predictors to have a varying slope across neighborhoods. Write a rationale specifying why you expect the slope to vary. Do the run where you allow this to happen. Interpret the Chi square test of the variance of this slope. WRITE THIS UP: the rationale, description of results, interpretation of results. BRING 2 copies to class of YOUR OUTPUT and your WRITEUP
3/22 Write up your Level 2 conceptual model. For each predictor, provide a specific hypothesis, and an accompanying rationale. To help you begin your literature review, it is also a good idea to include cited references to go with each of your predictors. TRY AN ANALYSIS where you enter JUST one Level 1 predictor and one Level 2 predictor. Interpret the results shown for your Level 2 predictor. BRING 2 COPIES TO CLASS of the analysis and your write up of it.
3/29 Write up your methods section. See the tips page. Follow Criminology guidelines
4/5 Write up your introduction. See tips page. Follow Criminology guidelines
4/12 Write up your results and discussion.
4/19  
4/26  
5/3