BioStat 571 Homework

Week 1 Exercises - Due Wednesday January 13, 2010
- PDF version of Exercise #1
- Lindsey-Jones-1998.pdf -- overview of GLMs with discussion of one model-selection criterion.
- For the first part of the exercise you are asked to locate and summarize an article that discusses model selection -- the following file has some options: ModelReferences.txt
- KingCounty2001.data Data for question #2 -- birth data
- KingCounty2001.txt Data Documentation for question #2 -- birth data
- FirstSteps.pdf Information regarding the intervention program.
- Solution Information
Week 2 Exercises - Due Wednesday January 20, 2010
- PDF version of Exercise #2
- Some Comments: HW2comments.txt
- MN-chapter5.pdf -- McCullagh and Nelder, Chapter 5
- Ananth-Kleinbaum-1997.pdf -- review article regarding ordinal regression
- To perform ordinal regression you can use the S+ function found in the MASS package (Modern Applied Statistics with S-PLUS) polr.q .
- You will also need to "source" the following support functions to get standard errors using polr(): vcov.q and misc.q .
- tenhave.data Data for question #2 -- child psychology experiment
- tenhave.txt Documentation for question #2 -- child psychology experiment
- Solution Information
  - exer2-key-2010.pdf
    - tenhave.results.txt Example results using my own S+/R function with intercept parameterization that differs from polr().
    - tenhave.sas Example SAS code
    - tenhave.lst Example SAS output
Week 3 Exercises - Due Wednesday January 27, 2010
- PDF version of Exercise #3
- CT-chapter2.pdf -- Cameron and Trivedi, Chapter 2
- Liang-McCullagh-1993.pdf -- review article regarding overdispersion
- Bee-data.txt Data for question #1 -- bee activity data
- Note: For question #1 you will want to use the splines library in R, and the function "ns" to fit natural splines.
- New Note: (22 Jan 2010) For question #1 if you are not familiar with some of the spline models then the following code should help orient you to the basic ideas -- shown here for a linear model, but the regression structure can be used for any GLM -- spline-tutorial.q -- Save this, review the code/comments, and run the code to show some model fits.
- Note: The file zip-single.q contains the mixture model setup that we are studying in question 2.
- Note: The example in the notes on pages 143-1 through 143-6 should be helpful in question #2 (code also in Handouts section).
- Solution Information
  - exer3-key-2010.pdf
    - zip-sim2.q R/S+ code for sims
    - zip-summary.out Results
Week 4 Exercises - Due Wednesday February 3, 2010
- PDF version of Exercise #4
- Solution Information
Week 5 Exercises - Due Wednesday February 10, 2010
- PDF version of Exercise #5
- Note: -- the data are in the "Data" section of the course web page
- cluster.min.q
- cluster.mean.q
- GenCRM.q
- stacked2wide.q
- infjack.q ----- infjack.demo.q ----- infjack.demo.out
- Solution Information
  - exer5-key-2010.pdf
  - Information regarding Q1 and Q2 can be found in the following files:
  - Information regarding Q3 is found in sections of the following book chapter. Specifically -- see pages 5, 8, 11, 14, 17, 20.
    - LDAchapter.pdf
Week 6 MIDTERM - Due Friday February 19, 2010
- PDF version of midterm exam
- Note: -- the data are in the "Data" section of the course web page
- Comments on Midterm ----- last updated: 12 Feb 2010 at 12:01pm
- New Note: (07 Mar 2010) -- here is my R program that computes the power (for different sample sizes) for Question 2(b) -- Count Data Sample Size Calcs
Week 7 Exercises - Due Wednesday February 24, 2010
- PDF version of Exercise #6
- Note: -- the data are in the "Data" section of the course web page
- Littell-StatMed2000.pdf ----- article on choice of covariance for LMM.
- geepack-paper.pdf ----- article on a GEE package for R (one way to use WLS).
- Solution Information
  - exer6-key-2010.pdf
Week 8 Exercises - Due Wednesday March 3, 2010
- PDF version of Exercise #7
- panss.data ----- PANSS data in "wide" format.
- panss.txt ----- documentation
- panss-data-stacked.data ----- PANSS data in "stacked" format with baseline.
- panss-data-stacked.txt ----- documentation
- Solution Information
  - exer7-key-2010.pdf
  - exer7-key-g-2010.pdf
Week 9 Exercises - Due Wednesday March 10, 2010
- PDF version of Exercise #8
- madras.data ----- Madras schizphrenia symptoms data.
- madras.txt ----- documentation
- gauss.hermite.q ----- function that returns ordinate values and weights for Gauss-Hermite quadrature. The function has a single argument: q = number of quadrature points. A list is returned with: z = evaluation points; w = weights.
- Solution Information
  - exer8-key-2010.pdf
  - exer8-key-extra-2010.pdf