The Utah Colleges
Exit Poll Sample Design Committee has the responsibility of creating
a sample design that will provide accurate predictions on election
night. The goal of this committee is to design a stratified multistage
cluster sample with a small margin of error. A major responsibility
of the Sample Design Committee is to design the sampling process
to be implemented in selecting the polling places and to determine
the sampling interval to be used at the specified polling places,
throughout the state of Utah.
First, "certainty"
counties are selected to be a part of the sample. These are the
counties that are automatically included because they contained
a participating school and/or have a large population. The next
task is to stratify the districts. The purpose of the stratification
is to ensure that the sample design provides a good representation
of all Utah voters and minimizes the margin of error. After the
counties in each district are stratified, the next step is to select
the number of counties per district and stratum. The goal is to
sample the fewest number of counties while maintaining a reasonably
small margin of error. Therefore, the committee determines which
of the remaining counties should be added to the certainty counties
already selected. The non-certainty counties are randomly selected
using the Probability Proportional to Size (PPS) formula based on
the expected voter turnout of each county. The next task for the
committee is to find out the optimal number of polling places per
county to examine. While doing this, the committee also determines
the number of people to interview at each polling place. PPS is
also used to select the polling places.