Since very few of us have the option to conduct a census of our customers, prospects or panelists we need to rely on the sampling process. For most consumer and B2B market research studies, one of our goals is to provide data that mirrors the population(s) of interest along the lines of these classification variables. There are two primary methods for meeting this goal – 1) use quotas to ensure representation and 2) use weighting to adjust proportions to mirror the population. Method one (quotas) is done on the front-end while data is being collected whereas weighting is done post-facto.
Quotas can be set in most online survey platforms. Quotas are a method of limiting who responds to the survey (or portions of it) based upon their responses to screening questions. Typically screening questions include demographics (e.g. age, gender, income, etc.) or firmographics for B2B marketing research (e.g. industry, company size, headquarters vs. field office, etc.) along with some measure of brand awareness or usage.
The number of quotas used can vary from one to many. However, it is best practice to limit the number to a maximum of five to seven. This will help to avoid several issues including inability to find the sample that meets the criteria. Quotas can become incredibly complex, therefore, it is a goal to be as thorough as possible, yet keeping an eye on simplicity. Sample quotas would be:
- 33% of respondents must be male under age 50
- 25% of respondents must be first time buyers with incomes of less than $50,000
- Respondent distribution should match the percentage of sales by territory
To use quotas effectively, we must have a priori knowledge of the variable distribution of interest, or at least of the percentage we desire to see in the data set. Panel profile questionnaires can serve as a source for this foundational knowledge. Other sources include data from CRM or corporate financial systems.
Should we always sample in accordance with the proportions found in our source data? The answer is no. There are times when we would want to oversample a particular population (new buyers for example) if we wanted to give their opinions greater weight in our decision process.
When quotas are filled – the percentage of respondents desired in a particular group is reached, the cell closes. New participants meeting these criteria are either screened out of the survey or re-directed to another questionnaire. Quotas are effective tools used to ensure that response patterns match a desired set of groups.