Recent Additions
Lecture notes & handouts
Useful Links
Readings, Assignments
References are to Lachin, unless otherwise noted. Sections
in square brackets are for self-study (i.e. I won't lecture on the
material) and discussion in class.
- Errata from Lachin
- Ch. 1: find definitions for medical terms
- Ch. 2, Sec. 1-3 [2.1.3.1-2.1.3.4]: Do problem 2.1
- Ch. 2, Sec. 4,5, 2.6.1-2.6.3 [2.8-1,2.8.3]:
- Conditional inference for the odds ratio
- Ch. 3, Sec. 1-4, 5.3
- Altman, Ch. 1, Ch. 5, Sec. 1-8 [Sec.9-14].
- Single Sample Binomial Tests
- Lenth, R.V., Some Practical Guidelines for Effective Sample Size Determination
- Altman, Ch. 14, Sec. 1-4 [Sec.5-7].
- [Altman, Ch. 15]
- Exercise 2.1: Provide a justification for the approach to analysis of
crossover trials described by Altman in section 15.4.10 based on the
model described in class on Jan. 27 by relating Altman's notation
to the "y bar dot dot" notation used in lecture.
- Exercise 2.2: Consider the usual set up for a 2x2 table, with rows denoting
test results (+/-) and columns denoting disease status (present/absent),
under the assumption of independent sampling
conditional on disease status. We can (based on the Wed. Feb. 3 lecture)
estimate the positive predictive value (PPV) of the test in question as long
as we have an estimate of disease prevalence. Suppose that in an
independent sample of size m the observed prevalence of disease
was p.
Derive a confidence interval for PPV based on this information,
using the logit transformation.
- Exercise 2.3: Do Problem 4.2 (p. 160, Lachin) parts 4.2.1, 4.2.2, 4.2.3.
- Assignment 2 = Exercises 2.1, 2.2, 2.3 above is due Wed. Mar. 3 in
class.
- Assignment 3 is due Mon. Mar. 22 in class.
- Assignment 4
- ICU.csv
- Ch. 7 reading: 7.1 (omit 7.1.3), 7.2, 7.3.1, 7.3.4, 7.5,
7.6.1-7.6.3
- hint on Assignment 4, Q3 b.: Use the predict() function in R.
Create counterfactual data-set of A patients by
df[df$hosp!="A","hosp"] <- "A"
- Hemant Ishwaran's function for ROC analysis
- Problems with Stepwise Regression
- Estimating relative risks in binomial models
Other resources