STA 2201S: Applied Statistics II Spring 2014

Final exam with solutions.

HW 4 deadline is extended to Monday, April 14, 5 pm. No extensions can be given beyond this date and time.

Some clarifications for HW 4:

for Q1, it is best to work conditionally on \(w_i\). If you make the substitution \( x_j = w_j - u_j\) in the regression model, note that the error no longer has mean zero, and it is correlated with the covariate.
for Q3, there are several ways to plot the sex coefficient as a function of other smoothing coefficients, and any version you choose is fine, as long as you're clear about what you did.

HW4, Q3, refers to Peng et al. (2006), referenced in the slides of March 21: Here is the paper.

HW 2 Solutions

Homework 4 due before 5 pm, April 14
Latex source (amended March 21, 1.49pm)
Homework 3 due March 21, 2 pm. Latex source
Homework 2 due February 28, 2 pm.
Homework 1 due February 7, 2 pm.

April 4

Slides. You are not responsible for any topics shaded in gray.

March 28

Slides (first part)
Slides updated 1.55

March 21

Slides
LA Times biomarker story
Paper
Current population tables at Statistics Canada
Previous versions: Harris found these using Wayback Machine

March 14

March 7

February 28

Slides (updated 1.30, Feb 28)
Annotated
Yi Lu's slides
CBC News on cheating
Paper by Christensen Hughes on survey form 2002-2003.
For the PSID data, the degrees of freedom in the output are a bit mysterious. After much exploring, it seems that the calculation of these 'denominator' degrees of freedom is anything but straightforward. There are good discussions here and here, and Faraway has updated his book with a post here. A nice tutorial on mixed effects models by John Fox is here.

February 14

Slides (Correction to Slide 10 Feb 27)
Annotated

February 7

Slides (Final)

January 31

Slides (updated 31/1 after class)
Annotated
Science article on Medicaid experiment
R code for toxoplasmosis example (somewhat scrappy)

January 24

Slides
Annotated
chart on drinking and voting
Globe and Mail
National Media

January 17

Slides
A few annotated pages from the slides
LaTeX code for HW 1
Harvard Med School report on the nocebo response (thanks to Ali Al-Aradi)
The Scientist Magazine article on the nocebo response (Ditto)
Globe and Mail article on the migraine placebo study
Science Translational Medicine report on migraine study
Supplemental Material for the migraine study

January 10

Slides (Typos corrected Jan 12)
Brain and reading study:
There are links to the Washington Post article, and to the Wired Science blogpost, embedded in the slides, but I have also uploaded
- Washington Post story
- Research article (from Brain Connectivity, 3, No.6, 2013)
Reproducible Research
- Slides from Aaron Baggett including a brief introduction to knitr
- Yihui Xie's knitr page
- Economist article on unreliable research
- Ionnidis, 2005, "Why most published research findings are false".
Statistics Jobs in the News

Course Information

Text

Statistical Models by A.C. Davison.

Computing

You are welcome to use the statistical computing package of your choice, but I will refer exclusively to the R computing package. Some online resources that I've found helpful are:

RStudio, an IDE for R that has many useful features
Some tips and tricks for RStudio by Paul Chang
R commander, a simple GUI based interface to R
Introduction to R by Longhow Lam
John Verzani's online book simpleR.
If you already know SAS/SPSS/Stata, you may find this Quick-R guide helpful.
Thomas Lumley's R course notes are often recommended on the LinkedIn R Project