This web page is for error corrections and other changes to the slides and homeworks.

• (posted May 17). A bunch of corrections were made to the final exam and its solutions after they were first posted and, of course, after the exam was taken. Only two of these were actual errors. In question 7 on the test the distribution was intended to be continuous with support (0, ∞) rather than what the test form said (discrete with support 0, 1, 2, …). No one seemed to have been bothered by this, so no adjustment was made to the grading. The solution to question 8 originally posted was missing a ½. The correct answer is
Gamma(n + α, λ + ½ ∑i = 1n xi2)
In questions 3, 6, and 8 the notation X1, …, Xn was added in addition to the IID that was already there. Since this was asked about during the exam, and since IID implies X1, …, Xn was was written on the blackboard during the exam, no adjustment was made to the grading. Minor changes in wording that do not affect meaning were made in some other places in both the exam form and the solutions.
• (posted May 13). F changed to Fθ in the table about the parametric bootstrap on slide 11, deck 8. (No an actual error.)
• (posted May 1). Solution to hw 10 reposted. Not sure what the issue was but the one previously posted was behind the source code. So it may have changed some (but I don't know where).
• (posted May 1). For a while, the solution to hw 11 was posted with messed up plots (one duplicated, one in the wrong place, one missing). Fixed by 7:40 am on May 1.
• (posted May 1). Very minor changes to the solutions to homeworks 3, 4, 7, 8, and 11 to make the problem statements exactly match in the homework assignment and the solutions (no changes in the solutions themselves).
• (posted Apr 29). A bunch of changes to the proof about likelihood ratio tests. On slide 33,
ln(θ0) M
has been changed to
MTln(θ0)
so ∇ ln(θ0) is interpreted as a column vector (apparently was supposed to be a row vector before). On slide 36, the four terms in the limit have signs +, −, −, and + (where before they were +, +, −, and −). The reason is that n−12 ln(θ0) converges to − I(θ0) (not + that). On slide 38, the middle line of the three-line displayed equation is
UT (IA M [MT A2 M]−1 MT A) U
not as it was before with an extra A in there.
• (posted Apr 13). Many changes made to deck 5 of the slides. Most are small changes in wording that do not affect the sense (slides 40, 44, 50, 60, 63, 77). One (slide 14) strengthens what is stated. Before it read

If the p columns of M are linearly independent … and p ≤ n, then V is a p-dimensional vector space and the map …