PubH 6470 introduces students with a background in statistics to programming, graphics, and data analysis using SAS. The course concentrates on data-step programming, data editing and reformatting, as well as statistical applications.

Instructor: William Thomas, Mayo A-467, 625-0651
Office hours: 2:15 - 3:15 Wednesdays or by appointment

In addition to Mayo C-381, PC-SAS is installed at these computing labs:
Diehl Hall Biomed Library
Coffey Hall, Coffman Union

Many documents on this website are PDFs (portable document format). For software to read them, download Adobe Acrobat Reader here.

I recommend doing the coursework with PC-SAS on your own computer; it's not required but it's much more efficient.
Get PC-SAS through the University of Minnesota for $75 per year. See the course information for other options.

• UCLA SAS Help Page: examples, videos
• Stanford SSDS Getting Started Guides (Click "Software Services")
• Phil Spector's Intro to SAS (slides)
• CSCAR Univ of Michigan SAS Help
• Yale Statlab SAS Help
• Univ of Texas SAS Help

• SAS Help documentation
• Google: Search for "SAS name of problem", for example: SAS change character to numeric
• SAS customer support: code samples, search papers presented at SAS User Group conferences
• SAS L-archives search: Q & A
• UCLA SAS Help Page: advanced examples
• SAS Press: code and data from SAS books

• HW 1, Child IQ.xls, Lead-data.xls.

• HW 2; OGTT.xls; HW2 NHANES3 Adult spreadsheet, HW2 NHANES3 Exam spreadsheet, example demographics table.

• HW 3; Florida 2000.xls, Grades.xls.

• HW 4; OGTT2.xls, Collection Dates.xls.

• Midterm Exam; midterm-2011.xls. Example SAS code.

• HW 5; HW5 NHANES3 Exam.xls.

• HW 6; satiety.xls.

• Final Exam; disease-events.xls, GFR.xls.

---

---

2011 Syllabus: Class Notes, Data, Programs

(* revised from 2010)

• Running SAS, editing code, importing and checking data, basic programming (weeks 1-3)
• 1. Intro to PC-SAS* ; Child IQ.xls, SAS program
• 2. Procedures (Ttest, NPar1Way,Freq), data step, IF, missing values*; SAS program
• 3. Character variables, SET, MERGE, using output datasets*
• 4. SAS limitations, more on merge, data set options, dates and times* ; Editor keyboard shortcuts, D Morgan: "Essentials of SAS Dates and Times", SAS program.
• 5. Data checking, Graphics: Plot, Insight, SGplot, Gplot*
• 6. Simplifying repetitive code with arrays and macros*; P Grant: "The SKIP macro" (comment macro)
• General Linear Models: regression and ANOVA models (weeks 4-7)
• 7. Linear models - Minn math scores, 2000; Grade-8-mathscores.xls
• 8. Checking code, Proc REG, math scores example*; M. Yee: "Debugging SAS code"
• 9. Proc GLM, categorical predictors, class variables*
• 10. LSMeans, interaction plots, adjusting*
• 11. Multi-factor ANOVA, adjusted LSMEANS*
• 12. Reading bad spreadsheets, program structure, GLM models*, bad_spreadsheet.xls, SAS program
• 13. ODS select, ODS output, confounding, mediation, segmented regression*, segmented regression.xls, SAS program
• Bootstrap, missing values (week 8)
  • 14. Bootstrap confidence interval with Proc Surveyselect*; D. Cassell: "Bootstrapping in SAS"; for SAS bootstrap macros, documentation, and examples using those macros, see 2008 Lectures 12 - 14.
  • 15. Bootstrap t-test, multiple imputation*
• Logistic regression, ordinal regression, log-binomial regression, propensity scores (weeks 9-11)
  • 16. Risk, odds, smoothing, Proc Logistic*, PL Flom: "Proc Logistic: Traps for the Unwary".
  • 17. Logistic regression: fitted probabilities, lack-of-fit test*, NCHS Hypertension data (SAS dataset hypertension_2008.sas7bdat).
  • 18. Odds-ratio and relative risk, log-binomial regression, ordinal regression*; SAS code, age-BMI sample data (SAS dataset age_bmi_sample.sas7bdat).
  • 19. Propensity scores*; matching macro, adapted from M. Coca-Perraillon (1987) "Local and Global Optimal Propensity Score Matching."
• Longitudinal data, crossover trials, hierarchical linear models (weeks 12-13)
• 20. Imbalance and adjusting, longitudinal data and correlation, area under the curve*.
• 21. Reshaping longitudinal data: long vs wide data, writing out CSV and Excel files*, SAS code, family income data (SAS Dataset econ_long.sas7bdat); Zirbel (2009) "Learn the basics of Proc Transpose", Tilanus (2007) "Turning the data around: Proc Transpose"; my reference SAS code.
• 22. Longitudinal data, correlation matrix, Proc Mixed*; SAS code, Alzheimer Trial.xls ; Judith Singer: "Multilevel Models in Proc Mixed"
• 23. Random effects example, multilevel/hierarchical models*; SAS code, Minnesota-radon.xls (from Gelman and Hill, 2007).
• 24. Proc SQL, fuzzy merge, crossover designs*; SAS code, Family Economic Data.xls ; TJ Harrington: "Intro to Proc SQL" , W Hu: "Top Ten Reasons to Use Proc SQL" .

---

---

2010 Syllabus: Class Notes, Data, Programs

• Running SAS, editing code, importing and checking data, basic programming (weeks 1-3)
• 1. Intro to PC-SAS ; Child IQ.xls, SAS program
• 2. Arithmetic, missing values, basic tests
• 3. Character variables, SET, MERGE, standardizing
• 4. Dataset options, more on merge, data checking, Insight
• 5. dates, macros, arrays, computing change from baseline; P Grant: The SKIP macro (comment macro)
• Graphics: SGplot, Gplot, Insight, ODS (week 4)
• 6. Graphics SGplot, Gplot ; SAS code
• General Linear Models: regression and ANOVA models (weeks 5-7)
• 7. Linear models - Minn math scores; SAS code, Grade-8-mathscores
• 8. Proc REG, math scores example ; SAS code
• 9. Proc GLM, categorical predictors, class variables.
• 10. Factorial ANOVA, interaction plots, LSmeans; Rat-Diets.xls, SAS code.
• 11. Proc GLM: LSmeans.
• 12 Proc Corr, Back-transformation with ODS, interactions, program structure.
• 13 Confounding, Mediation, reading bad spreadsheets; bad_spreadsheet.xls.
• Bootstrap, missing values (week 8)
  • 14. Bootstrap, Proc Surveyselect; bootstrapmacros.sas, SAS bootstrap documentation; for examples using these macros, see 2008 Lectures 12 - 14.
  • 15. Bootstrap tests, missing values, imputation.
• Logistic regression, ordinal regression, log-binomial regression, propensity scores (weeks 9-11)
  • 16. Risk, odds, Proc Logistic, PL Flom: "Proc Logistic: Traps for the Unwary".
  • 17. ROC curves, subset selection, conditional logistic regression, NCHS-hypertension.xls.
  • 18. Percent classified correctly, log binomial regression, ordinal regression (corrected version).
• Longitudinal data, crossover trials, hierarchical linear models (weeks 12-13)
• 19. Longitudinal plots, long vs wide data, Transpose, Zirbel (2009) "Learn the basics of Proc Transpose", Tilanus (2007) "Turning the data around: Proc Transpose".
• 20. SQL, fuzzy merge, response feature, AUC; SAS code, Family Economic Data.xls ; TJ Harrington: "Intro to Proc SQL" , W Hu: "Top Ten Reasons to Use Proc SQL" .
• 21. Longitudinal data, correlation matrix, Proc Mixed; SAS code, Alzheimer Trial.xls ; Judith Singer: "Multilevel Models in Proc Mixed"
• 22. Mixed model example; Crossover designs
• Survival data: survival function, competing risks, proportional hazards (weeks 14-15)
• 23. Survival data.
• 24. Proportional hazards regression
• 25. Propensity scores; matching macro, from M. Coca-Perraillon (1987) Local and Global Optimal Propensity Score Matching.

---

2008 Lecture Notes and Examples
1. Intro to SAS I; SAS program.
2. Intro to SAS II, reading Excel spreadsheets; Workbook1.xls, Workbook2.xls, SAS program.
3. Data checking, Proc Insight, SAS Manual, basic tests
4. Missing values, graphics, reporting in MSWord, SET, MERGE; "Fix SAS output" MSWord macro, SAS program
5. Merging, data set options, GLM.
6. GLM: residual & interaction plots, means, LSmeans, dates, arrays; SAS program.
7. GLM: LSmeans, estimate; missing values.
8. MI and MIanalyze with GLM; smoothing, jitter; SAS code, HAMD2 data.
9. Correlation, partial correlation, regression: Proc REG; SAS program, Grade 8 data (SAS permanent file).
10. Regression example, VIF, plots, subset selection.
11. Making CLASS variables for Proc Reg, predictions, sample size; SAS program.
12. Macros and Bootstrap; SAS program, bootstrapmacros.sas, SAS bootstrap documentation.
13. Bootstrap confidence intervals: correlation, kappa, agreeement.
14. Bootstrap prediction error, t-tests.
15. Longitudinal data: graphs, area under a curve (AUC); SAS program.
16. Within-person correlation, covariance matrix.
17. Proc Mixed: repeated measures, random effects.
18. Crossover designs.
19. Logistic regression.
20. Log-binomial, repeated binary observations.
21. Conditional logistic regression, ordinal logistic regression.
22. Survival data, Kaplan-Meier estimates, randomization log-rank test, SAS code, macro file from Cantor: SAS Survival Analysis Techniques, 2nd ed..
23. Reporting comparisons of survival curves, proportional hazards regression.
24. Checking proportional hazards, subset selection, time-varying predictors.
25. Competing risks: cumulative incidence, SAS code for lecture, CumIncid macro (from www.sas.com), BMT data from Klein & Moeschberger, Survival Analysis,2nd ed, sec 1.3, App D.

---

Updated 13 Dec 2011