Unbised estimator of Binomial Distribution

Click For Summary

Discussion Overview

The discussion revolves around finding an unbiased estimator for the power of the parameter p in a binomial distribution, specifically p^m, where m is less than n. Participants explore various approaches to derive such an estimator, focusing on the challenges associated with the power of m and the definition of expectation.

Discussion Character

  • Exploratory, Technical explanation, Debate/contested, Mathematical reasoning

Main Points Raised

  • One participant expresses uncertainty about how to find an unbiased estimator for p^m, noting that induction on m is complicated due to the power of m.
  • Another participant suggests counting sequences with two successes in a row to construct an unbiased estimator for p^2, questioning if this approach would satisfy the condition E[u] = p^2.
  • A different approach is introduced by another participant, who proposes that Y = X_1...X_m is unbiased for p^m but seeks an estimator that is a function of the sample total, leading to the construction of a random variable Z(T).
  • There is a clarification regarding the value used in the random variable Z, with one participant correcting it from 26 to m.
  • Participants engage in back-and-forth clarifications about the definitions and constructions being used, indicating some confusion over the specifics of the estimators proposed.

Areas of Agreement / Disagreement

The discussion contains multiple competing views and approaches to finding the unbiased estimator, with no consensus reached on a specific method or solution.

Contextual Notes

Participants express limitations in their approaches, particularly regarding the normalization of the random variable Z and the specific definitions used in their estimators. There is also uncertainty about the implications of the power of m in the context of the estimators being discussed.

leon1127
Messages
484
Reaction score
0
[SOLVED] unbised estimator of Binomial Distribution

I have no idea how to find such an estimator
SupposeX_1, ..., X_n \sim Bern(p)<br />
find an unbiased estimator of p^m, for m &lt; n
Induction on m was a nasty mess that should not be expected. The power of m causes some problem when I try to go from the definition of expectation. Any one have some hint?
 
Physics news on Phys.org
Statistic u is unbiased if E = p^m. Suppose m = 2. To find u, you count all sequences that have two successes in a row, then express them as a ratio of the total number of trials. Would that satisfy E = p^2?
 
I approached in a similar way.
It is clear that Y = X_1...X_m is unbiased for p^m. I was asked to find an unbiased estimator that is a function of sample total. Thus I construct a R.V which
Z(T) = 1 for T = 26; 0 elsewhere
I might have to normalise Z so that it is the same as Y. However I don't know how to sum it.
 
Where did 26 come from?

If you look at my previous post you will see that it is a function of the sample total.
 
it is not 26, it should be m...
 
Very well, my previous post applies.
 
k gotcha thx
 

Similar threads

  • · Replies 3 ·
Replies
3
Views
3K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 12 ·
Replies
12
Views
3K
  • · Replies 9 ·
Replies
9
Views
2K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 5 ·
Replies
5
Views
3K
  • · Replies 11 ·
Replies
11
Views
10K
Replies
1
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 14 ·
Replies
14
Views
6K