Estimating p with Bernoulli Sample

  • Context: Graduate 
  • Thread starter Thread starter chicory
  • Start date Start date
  • Tags Tags
    Bernoulli
Click For Summary

Discussion Overview

The discussion revolves around finding a method of moments estimator for the parameter p in a Bernoulli random sample, under the constraint that 0 ≤ p ≤ 0.5. Participants explore different approaches to ensure the estimator does not exceed the upper limit of 0.5.

Discussion Character

  • Technical explanation, Debate/contested, Mathematical reasoning

Main Points Raised

  • One participant suggests using the estimator P = (sum Xi) / n, but notes that this could lead to estimates greater than 0.5 in extreme cases.
  • Another participant proposes an alternative estimator: P = Max(0.5*n, Sum Xi) / n, to prevent exceeding 0.5.
  • A different approach is suggested where P = sum Xi / n if sum Xi / n < 0.5, and P = 0.5 otherwise.
  • One participant questions whether the adjustments to the estimator are too artificial and raises concerns about the probability of estimating p as 0.5.
  • Another participant argues that while the adjustments may make p = 0.5 more common, they believe this does not detract from the accuracy of the estimation.

Areas of Agreement / Disagreement

Participants express differing views on the appropriateness of the proposed estimators, with some favoring adjustments to prevent exceeding 0.5, while others question the artificiality of these adjustments. The discussion remains unresolved regarding the best approach to estimating p.

Contextual Notes

Participants acknowledge the constraint on p but do not fully resolve the implications of their proposed estimators or the assumptions underlying their methods.

chicory
Messages
3
Reaction score
0
A Bernoulli random sample :

X=1 , Pr(X=1)=p;
X=0 , Pr(X=0)=1-p;

taken X1, ..., Xn

and if it is known that 0=< p =<0.5

find a method of moments estimator of p

If I just take estimator of P =(sum Xi )/n than the estimate p may be bigger than 0.5 as in extreme case all Xi =1 .
What should I do , take estimator of p = Max (0.5*n , Sum Xi) /n ?
 
Physics news on Phys.org
chicory said:
A Bernoulli random sample :

X=1 , Pr(X=1)=p;
X=0 , Pr(X=0)=1-p;

taken X1, ..., Xn

and if it is known that 0=< p =<0.5

find a method of moments estimator of p

If I just take estimator of P =(sum Xi )/n than the estimate p may be bigger than 0.5 as in extreme case all Xi =1 .
What should I do , take estimator of p = Max (0.5*n , Sum Xi) /n ?

Just simply state that fact, that is

P= sum Xi /n if sum Xi /n < 0.5
P= 0.5 otherwise.
 
But... is it too artificial?
Pr( estimator (p) = 0.5 ) is slightly bigger ...?
 
Artificial? You like better min( sum Xi/n, 0.5) ?

No, p will not be generally bigger and you will rarely get estimations of 0.7 or 0.8, remember that they already assure you that the real p is no higher than 0.5

Adjusting the estimation makes the value p=0.5 more common but that is irrelevant and only makes the estimation more accurate.
 
Last edited:
Thank you so much !
 
You're welcome :smile:
 

Similar threads

  • · Replies 0 ·
Replies
0
Views
2K
  • · Replies 17 ·
Replies
17
Views
2K
  • · Replies 2 ·
Replies
2
Views
2K
Replies
3
Views
3K
  • · Replies 12 ·
Replies
12
Views
3K
  • · Replies 5 ·
Replies
5
Views
3K
  • · Replies 3 ·
Replies
3
Views
3K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 6 ·
Replies
6
Views
3K