Fielding Representatively

In B2C surveys, there is often a need to understand market sizing, market penetration, and general sentiment of a specific geographic market. To achieve this a representative sample can be used.

What is a representative sample?

A representative sample involves fielding a certain population across a number of dimensions to achieve a data set that will be representative of that population. Typically, this will mean you balance based on all respondents who click into a survey, commonly referred to as click-balancing, rather than those who complete the survey.

How to conduct a representative sample

The first step is to figure out what dimensions you will be balancing on. For the US, there are 6 dimensions commonly used (i.e age, gender, etc.) and the distribution across these dimensions should be known by all panels. For other markets, the networks should be able to provide a set of dimensions along with the target distributions. The networks will handle the exercise of balancing across these dimensions, typically aiming for variances +- 3-5%.

Do's and Don't

• Ensure that the acceptable variance range is established before launching your rep sample.
• Do not set any quotas on this phase of the project
  
  • If you need to over index on a certain subset of the population (i.e users of a target brand), you can run a "over sample" or "augment" phase once the representative phase is completed.
    
    

If you intend on running a demographically representative sample, feel free to reach out to your Survey Director partners to ensure that the right questions are being asked so the networks can balance across the appropriate dimensions.