The Listener Survey Toolkit
Chapter 7
How to know which responses are significant

TABLE OF CONTENTS
Overview
Why Conduct a Listener or Member Survey
Research Myths and Misconceptions
Building the Survey
Survey Methodology
Survey Questions
Processing the Data
Data Analysis and Reporting
Applying the Survey Results
Station Service and Support for the Listener Survey Toolkit
Sample Questions
Significance has both a statistical and generic meaning.

Statistically, a result is significant when its value is greater than the standard error between the respective results. For example, if we ask "Should WXYZ play more music by Scottish composers?," the responses may be 78% yes and 22% no. Due to the sample size of the survey, the standard error of this question was plus or minus 5%, which means the "yes" answer is somewhere between 73% and 83%, and the "no" answer is somewhere between 27% and 17%. Clearly, this result would be significant because the lowest possible "yes" answer is far greater than the highest possible "no" answer. In the most basic of terms, this is an example of statistical significance.

Generically, a result is significant if it is such that it can cause an increase in the station's understanding or a change in its programming, policy, or other behavior. A station would not be wise to take action solely on the result of a question that was 52% for and 48% against. It's not enough of a difference of opinion to make a clearcut difference in behavior.

[ Chapter 7 Table of Contents | Next Section | Toolkit Home ]