Skip to main content

Mathematical Statistics II · Bootstrap, permutation, and simulation-based inference

The sample is x=(1,2,4) with n=3

Problem

Mathematical Statistics II practice problem

The sample is x=(1,2,4)x=(1,2,4) with n=3n=3. The functional of interest is the squared sample meanθ^=xˉ2.\hat\theta=\bar x^2.Write θ^(i)\hat\theta_{(i)} for the same functional computed after deleting the iith observation, and write θˉ(⋅)=n−1∑iθ^(i)\bar\theta_{(\cdot)}=n^{-1}\sum_i\hat\theta_{(i)} for the leave-one-out mean.

1. Prove xˉ=7/3\bar x=7/3 and θ^=49/9\hat\theta=49/9. Compute the three
leave-one-out valuesθ^(1)=9,θ^(2)=254,θ^(3)=94,\hat\theta_{(1)}=9,\qquad \hat\theta_{(2)}=\frac{25}{4},\qquad \hat\theta_{(3)}=\frac94,and their mean θˉ(⋅)=35/6\bar\theta_{(\cdot)}=35/6.

2. The jackknife bias estimate is
(n−1)(θˉ(⋅)−θ^)(n-1)\bigl(\bar\theta_{(\cdot)}-\hat\theta\bigr). Prove that this equals 7/97/9, and that the bias-corrected estimator θ^−bias^\hat\theta-\widehat{\mathrm{bias}} equals 14/314/3.

3. The pseudovalues are
θ~i=nθ^−(n−1)θ^(i)\tilde\theta_i=n\hat\theta-(n-1)\hat\theta_{(i)}. Prove(θ~1,θ~2,θ~3)=(−53,236,716)\bigl(\tilde\theta_1,\tilde\theta_2,\tilde\theta_3\bigr) =\Bigl(-\frac53,\frac{23}{6},\frac{71}{6}\Bigr)and that their average is again 14/314/3. The jackknife variance isV^=n−1n∑i=1n(θ^(i)−θˉ(⋅))2=1n(n−1)∑i=1n(θ~i−θ~ˉ)2.\widehat V =\frac{n-1}{n}\sum_{i=1}^n \bigl(\hat\theta_{(i)}-\bar\theta_{(\cdot)}\bigr)^2 =\frac{1}{n(n-1)} \sum_{i=1}^n\bigl(\tilde\theta_i-\bar{\tilde\theta}\bigr)^2.Prove V^=553/36\widehat V=553/36.

4. Audit both sentences: (i) “the bias factor is nn rather than
n−1n-1, or the functional is the unsquared mean, giving bias 7/187/18 or 00”; (ii) “the variance uses (n−1)/n(n-1)/n times the sum of *pseudovalue* squared deviations, giving 553/9553/9, or this is a nonparametric bootstrap of the sample maximum.”

Stay inside the delete-one jackknife of xˉ2\bar x^2. Do not replace the functional by xˉ\bar x or by a bootstrap of the maximum.

Hint

Hint

xˉ=7/3\bar x=7/3 and θ^=49/9\hat\theta=49/9. The three leave-one-out means are 33, 5/25/2, and 3/23/2.

Check your work

Work the problem yourself first. Then open it in Training to check your answer and read the full worked solution.

The answer check and full solution for this problem come with ProofAnvil Practice membership ($19 USD monthly). See membership. Or start with the free Mathematical Statistics II sample problem: Try the free sample problem.

More Mathematical Statistics II practice problems