How To Run a Poll: The Sample Bias Problem

Avoiding Sample BiasHow To Run a Poll: Handling Sample bias

First, let’s define the term. Sample bias is the tendency to sway the poll results and/or responses by polling groups of people who are more inclined towards a certain opinion or position.

Political Polling Example
Let’s say you are attempting to discover whether the first Presidential Debate influenced undecided voters toward Romney or Obama.

You poll stay-at-home moms of children ages 5 and under, between the hours of 10:00 am and noon the morning after the debate.

The poll results would likely indicate that undecided voters are now supporting Obama.

Reason: The sample included those more likely to have children watching Sesame Street – a PBS program – and this sample might view Romney negatively after his comment that he’d discontinue PBS funding.

Have questions about polls or surveys? Contact us today 1.877.655.7837

Bias can be intentional – through an attempt to use the results to promote a candidate, for example – or unintentional – through a lack of knowing how to conduct a reliable poll.

Eliminating Sample Bias with Pollster
The Pollster system was designed to be more effective than traditional polling using the telephone, online connections, and email in eliminating bias overall, and specifically sample bias.

Pollster calls either random phone numbers or a list of predetermined phone numbers of an organization’s membership – NRA members, for example, and asks demographic questions as part of the poll in order to clearly identify the results statistically.

Next Week: Sample Bias Danger Zones for Online, Email, and Traditional Polling