|
Activity Number:
|
75
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 12, 2002 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Survey Research Methods*
|
| Abstract - #301753 |
|
Title:
|
Sampling Methods Related to Bernoulli and Poisson Sampling
|
|
Author(s):
|
Dhirendra Ghosh*+ and Andrew Vogt+
|
|
Affiliation(s):
|
Synectics for Management Decisions, Inc. and Georgetown University
|
|
Address:
|
1901 North Moore Street, Suite 900, Arlington, Virginia, 22209, USA , Washington, District of Columbia, 20057-1233, USA
|
|
Keywords:
|
conditional Poisson ; inclusion probabilities ; independence ; rejective sampling
|
|
Abstract:
|
In Bernoulli and Poisson sampling, each element of the population is selected independently with a prescribed probability. Then common estimators have easily calculable variance, but sample size is a variable that can assume any value.
Sampling methods that permit the sample size to be restricted include rejective sampling, also called conditional Poisson sampling, in which Poisson sampling is performed but the sample is rejected unless it has the desired size. Rejective sampling violates the independence property. One way to realize a Poisson sample is to choose independent uniform random variables in [0,1] for each population element, and select the element if the corresponding uniform random variable takes a value less than or equal to the stipulated inclusion probability. A refinement of this is to adjust the inclusion probabilities by a scale factor so that the sample size is exactly some desired size. This method also violates independence. There do, however, exist methods that preserve pairwise independence, whilst ensuring that variances are easily calculated and that sample sizes can be restricted to just a few values.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |
