
Advanced Statistics Advanced Probability and Statistics Math Forum 
 LinkBack  Thread Tools  Display Modes 
March 8th, 2012, 09:00 AM  #1 
Newbie Joined: Feb 2012 Posts: 18 Thanks: 0  bias estimator
Before the election, a polling firm asked 1,000 people and an another poll firm asks 2000 (other) people about what they did vote. A newspaper proposes to take the mean of the Labour Party's average level of support of the two measurements as an estimate of the true level of support to the party. questions: Show that this is an bias of an estimator. Why is this not optimal? Suggest a better estimator. 
March 8th, 2012, 12:14 PM  #2 
Senior Member Joined: Feb 2012 Posts: 144 Thanks: 16  Re: bias estimator
so see that this estimator is unbiased you need to compute its expectation (easy). to see why it is not optimal you need to calculate its variance. I suggest you calculate the variance of a more general estimator, namely a*I1+b*I2 where a+b=1 a>0 b>0 and I1 and I2 are the results of the two polls. You'll see that ther is e unique choice of a (and b=1a) that minimizes the variance, and it is not 1/2,1/2.


Tags 
bias, estimator 
Thread Tools  
Display Modes  

Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Unbiased estimator proof  nadroj  Advanced Statistics  7  September 12th, 2014 12:59 AM 
Test for bias  coolcat  Advanced Statistics  1  November 30th, 2013 02:44 PM 
bias coin density question  pzhao  Advanced Statistics  1  November 13th, 2013 03:32 PM 
coin tossing bias  baxy7  Advanced Statistics  5  July 3rd, 2011 09:34 AM 
nonlinear variance estimator  kandre  Advanced Statistics  1  November 14th, 2009 02:16 PM 