Friday, March 4, 2011

Sampling: Design and Analysis – Sharon Lohr – Chapter 3 Ratio and Regression Estimation

 

-          Often the sampling frame gives us extra information about each unit that can be used to improve precision of estimates

 

 

Ratio Estimation

 

-          For ratio estimation to apply, 2 quantities y and x must be measured on each sample unit

-          X auxiliary / subsidiary variable

-          Why use ratio estimation

o   Sometimes we want to estimate a ratio

§  Average yield of crop

o   Want to estimate population total, but population size N is unknown

o   Increase precison of estimated means and totals

o   Adjust estimates from sample so they reflect demographic totals

o   Adjust for non response

 

Regression Estimation

 

-          Ratio estimation works best if data are well fit by a straight line thru origin

-          Difference estimation

o   Used in auditing

 

 

Estimation in Domains

 

 

 

Models for ratio and Regression Estimation

 

-          Differences between model based and design based estimates

 

 

No comments: