Fall 2014: HS 852: Syllabus
Scientific Methods for Health Sciences: Linear Modeling
Introduction course to probability reasoning and statistical inference.
Prerequisites
HS 851, or equivalent, instructor may review syllabi of previously taken courses (past 5 years) and/or require a test to assess the equivalence of the student background, as necessary.
Class Schedule
See UMSN Courses and UMich Office of the Registrar. Mondays and Wednesdays: 12-2 PM, UMSN, NIB # 2184.
Course Description
HS 852 is a general linear modeling course, building on HS 851, focusing on commonly employed scientific computing techniques used in health sciences. The primary aim of the course is to provide students with the necessary skills to determine appropriate use, carry out, and interpret general linear modeling. Statistical software will be used to manipulate data, fit models and perform model diagnostics. Introduces students to applied inference methods in studies. HS852 is a 4 credit hour course (3 lectures + 1 lab/discussion).
Objectives
Students will learn how to:- Compare and contrast advanced statistical concepts, grasp model assumptions/limitations and apply them for quantitative analyses in healthcare research
- Apply multivariate statistical modeling enabling consistency between research questions and selected advanced statistical analyses
- Critique and select appropriate advanced statistical linear models for defined healthcare issues
- Conduct multivariate statistical analyses, such as multidimensional chi squares, logistic regression, principal components analysis, survival analysis, repeated measures ANOVA, MANOVA, MANCOVA, linear mixed models, hierarchical linear models
Examples of Topics Covered
- MLR Regression
- GLM
- ANOVA
- ANCOVA
- MANOVA
- MANCOVA
- Repeated measures ANOVA
- (partial) correlation
- Time series analysis
- Fixed, randomized and mixed models
- Hierarchical Linear Models
- Mixture modeling
- Surveys
- Longitudinal data
- Generalized Estimating Equations (GEE) models
- Model Fitting and Model Quality (KS-test)
- Common mistakes and misconceptions in using probability and statistics, identifying potential assumption violations, and avoiding them
Teaching and Learning Methods
This course meets four times on campus and will use blended instructional techniques to deliver learning materials, provide instructional resources and assess student progress. Synchronous web-streaming of lectures/labs and asynchronous virtual office hour forums will be supported. Assignments will be announced on the web and will be electronically collected, graded and recorded. A variety of teaching methods will be used including lecture, Journal Club, discussion, small group work, and guest presentation.Textbooks
- SMHS EBook and additional resources will be made available through the SOCR Wiki and may include chapters, websites for review, references, reports posted online, ebooks and learning modules.
- Julian Faraway, Linear Models with R, Second Edition, 2004 (Chapman & Hall/CRC Texts in Statistical Science)
- Julian Faraway, Extending the Linear Model with R: Generalized Linear, Mixed Effects and Nonparametric Regression Models, 2005 (Chapman & Hall/CRC Texts in Statistical Science)
Software and Computational Tools
We will only use open-source software, libraries and tools including the web-based SOCR tools (which require Java and HTML5/JavaScript enabled web-browsers) and the Statistical computing Software "R" (which you need to download and install the graphical user interface (GUI), RStudio).
Assignments and Evaluation Methods
- 40% Homework Projects
- 30% Midterm Exam
- 30% Final Paper
Standard letter-grading distribution will be used:
- A: 90%+
- B: 80-90%
- C: 70-80%
- D: 60-70%
- ...
- Plus and minus grads will also be used (e.g., "B-": 80-83%; "B+": 87-90%)
Grading Policy
The lowest graded Homework assignment will be dropped. All Homework assignments must be completed by the corresponding deadline, however. No late assignments will be accepted. For students with genuine documented reasons for missing the midterm arrangements will be made. If after receiving the graded exams or HW/projects back you believe a grading error has occurred please see the Instructor or your TA, within one week. Late regrade requests may not be accommodated. Reading assignments will be given. You will be responsible for the information covered in these assignments. Lecture and discussion attendance will be recorded from time to time.
