Qualifying Exam MA 582 problemsApril 2015
Problem 1.
Consider the Exp() family, where > 0 (and note that E(X) = here).
- Show that no two different parameter values could give exactly the same pmf (this is “regularity condition R0”).
- Is the parameter space here open (this is “regularity condition R1”)? Why or why not?
- Is the support of the pdf here independent of the paramter (this is “regularity condition R2”)? Why or why not?
- Find the maximum likelihood estimator (MLE) of, call it Yn.
- Show whether or not Yn is unbiased for .
- Show whether or not Yn is a consistent estimator for .
- Show whether or not Yn is asymptotically normal, and if it is, identify its asymptotic normal variance.
- Find I(), Fisher’s Information for Is MLE() efficient?Why or why not?
- Find the MLE of 2. Show whether or not it is biased.
- Find a function g so that n1/2( g(Yn) – g()) is asymptotically standard normal.
Problem 2.
We say the rv X has the W distribution with parameter 0 (written X ~ W() ) if X has pdf f(x,)= 3x2/, for 0 < x < , and f(x) = 0, elsewhere.
Consider the parameterized W family {W() : > 0 }.
- For each of the “regularity conditions” R0, R1, and R2, determine if the W family here satisfies that condition or not. (Refer to problem 1 here for a reminder of what those regularity conditions are.)
- Let Yn be the maximum of the random sample of size n. Show that Yn is a consistent estimator of
- Find the pdf of Yn. (Hint: Find its cdf first.)
- Show that Yn is NOT an unbiased estimator of
- Show that n(- Yn ) converges in distribution, and find its asymptotic distribution explicitly.
- Find an unbiased estimator of call it Tn. Show that Tn is a consistent estimator of
- Show that n(- Tn ) converges in distribution, and find its asymptotic distribution explicitly. (Hint: Use parts (e) and (f) here.)