LECTURER:
Karen Bandeen-Roche, PhD
Department of Biostatistics, Hygiene
E3624
Johns Hopkins University
Bloomberg School of Public Health
phone: 410-955-1166
e-mail: kbandeen@jhsph.edu
fax: 410-955-0958
Office Hours: Thurs: 1:15 - 2:30 pm
LECTURES:
10:30 AM - 12:00 PM
Tuesday, Thursday
Room W4030
LABS for
review, auxiliary material questions, and help with the problem sets:
|
|
Tuesdays, |
12:15 PM - 1:00 PM |
W4030
|
TEACHING ASSISTANTS/LAB INSTRUCTORS:
-
Yong Chen
-
Marie Thoma
-
Hao Wu
OFFICE HOURS for Teaching Assistants:
WEB SITE:
TEXTBOOKS:
-
FEH: Harrell, F.E.
(2001), Regression Modeling Strategies, With Applications to
Linear Models, Logistic Regression, and Survival Analysis, New
York: Springer.
Suggested Supplemental
Books:
-
Breslow, N.E. & Day,
N.E. (1980), The Analysis of Case-Control Studies, Oxford
University Press.
-
Breslow, N.E. & Day,
N.E. (1987), Design and Analysis of Cohort Studies, Oxford
University Press.
-
Cox, D.R. and Snell,
E.J. (1981), Applied Statistics, Principles and Examples, New
York: Chapman and Hall.
-
Dobson, A.J. (1983), An
Introduction to Generalized Linear Models, New
York: John Wiley & Sons.
-
Hosmer, D.W. &
Lemeshow, S. (2000), Applied Logistic Regression, 2nd edition,
New York: Wiley.
http://www3.interscience.wiley.com/cgi-bin/bookhome/109855848
-
McCullagh P. and
Nelder J.A. (1989), Generalized Linear Models, 2nd. Ed.,
Chapman and Hall.
-
Santner T.J. and Duffy
D.E. (1989), The Statistical Analysis of Discrete Data, New
York: Springer-Verlag.
GRADING
Homework assignments
Same policy as for Biostatistics 653–
Best 3 out of 4 EXCEPT that Homework 4
MUST BE SUBMITTED |
40% |
|
Project (1) and Final (1) Exam
|
60% |
Weighted to higher of: 100% Final OR
50% Final, 50% Project
e.g. Project is optional
NOTE: Project is
mandatory for
Biostatistics
degree students |
|
Guaranteed grades: |
|
| A = 90% on both components |
|
| B = 80% on both components
|
|
| C = 70% on both components |
|
|
Guaranteed grades are as for Biostatistics 140.653.
Curve may also be implemented.
There will be no extra or make-up credit, except as may
occasionally be offered on homework assignments or exams |
PROJECT
DUE DATE: May 15 at noon.
A data analysis project may be submitted for 30% course credit. The
primary analytic outcome(s) should be binary or counted, so that the
project will draw primarily on the first two thirds of Biostatistics
654. The project consists of selecting a data set (preferably related
to your own research or field), posing a substantive question of
interest, analyzing the data to address the substantive question, and
writing up findings
in a report. The report should include sections:
1. Introduction/Background
(1-2 pages): describe (i) the scientific problem of interest; (ii)
how the data set you will analyze arose and why it well addresses the
scientific problem; (iii) motivation for specific potential
confounders, mediators or effect modifiers; (iv) references to other
work.
2. Aims (1/2 page):
motivation and statement of the specific question(s) that you will
address in your analysis. This section should make clear whether the
primary goal is descriptive, inferential, or predictive; state any
hypotheses.
3. Methods (2 pages): (i)
operationalization of the problem within a statistical model or
sequence of models; (ii) description of analyses to be applied,
including how each addresses the scientific question(s) or ensures
meaningful interpretation.
4. Analysis (2-3 pages
text, plus supporting tables/graphs): a report of analyses conducted,
including description/graphs and formal inference.
5. Conclusion (1-2 pages):
summary/interpretation of findings, discussion of study limitations
and implications.
GRADING CRITERIA: Each section of the project will be graded as A, B,
or C
level on the criteria: clear/engaging narrative, correctness,
completeness. The analysis
section will count for 50% of score and the other sections equally for
the other 50%. I
will deduct credit for a trivial project topic; if you are concerned
whether your project has
sufficient content, please discuss it beforehand with
Dr. Bandeen-Roche.
ETHICS POLICY:
Homework assignments
Please study together, and feel free to talk to one another about
homework assignments. The mutual instruction that student colleagues
give each other by doing this is among the most valuable that can be
achieved. However, it is expected that homework assignments will be
implemented and written up independently. Specifically, please do not
share analytic code or output. Please do not collaborate on write-up
and interpretation. Please do not access or use solutions from any
source before your homework assignment is submitted for grading.
Thanks.
ETHICS POLICY:
Project
The project must be your own work. Papers that involve research in
collaboration with others is permissible provided that all colleagues
are acknowledged, you conduct all analyses you report independently,
and you write up the work entirely on your own. The paper must follow
ethical standards of scientific publication. Please cite references
appropriately. Any narrative that is not your own must be placed in
quotes and attributed to the source. Thanks in advance.
LATE POLICY:
Course requirement due dates for the term are provided below;
occasionally they are modified for all based on course progress.
Homeworks must be submitted on time to receive credit. Except for
extraordinary crises, there will be no exceptions.
In general exams must also be taken at the scheduled time. At the
instructor’s discretion, exceptions will be made for unforeseen
personal illness, family health emergency or other crisis, or for
unavoidable conflicting trips that are agreed at least three weeks in
advance of the exam at issue.
CALCULATOR:
-
Basic
functions (+, -, x, /), logarithms and exponents, simple memory and
recall, factorial key.
REGISTER FOR COURSE
e-MAIL:
COURSE DESCRIPTION:
Biostatistics 140.654 is a
course in generalized linear regression analysis. Foundational
topics of the course include: generalized linear models and their
uses; maximum likelihood estimation and inference; and model
assumptions, diagnosis, and interpretation. Specific topics
include: logistic and Poisson regression, grouped and
individual-level data, analysis for unmatched and matched
case-control studies, analysis for cohort studies, and introductory
survival analysis.
COURSE OBJECTIVES:
Biostatistics 140.654
acquaints students with:
-
the definition, statistical assumptions, and interpretation of
generalized linear regression models, specifically including
logistic and Poisson regression and loglinear modeling
-
maximum likelihood (ML),
conditional likelihood, and partial likelihood estimation, including
the iteratively reweighted least squares implementation of ML
-
standard methods for
making inferences on model parameters, including Wald testing and
confidence interval construction, and likelihood ratio / deviance
testing
Students will develop skills
to:
-
build and fit generalized
linear regression models and survival analyses using standard
statistical software;
-
diagnose the
appropriateness of their own and others' models for description,
inference, and prediction;
-
analyze case-control,
rate, & cohort data, recognizing special features of each;
-
sensibly interpret fits
and inference for statistical and scientific importance.
OTHER LINKS:
Home |
Schedule |
Classes |
Problem Sets |
Labs |
e-Mail
|
