As the name suggests, the classification section is where you would place questions used for classification purposes. What are classification questions? Age, gender, income and education level are examples. However, classification is not limited to the demographic questions only – for example, you should think in terms of how you would want to slice your data. If the data variable (question) you need for data slice is not anywhere else in the survey, this is the section where you can put it.
Over the years, there has been a trend to place additional questions in this section that are nice to have answers to but are not necessary. If you plan to do that, group those questions at the very end, after you have captured all “must-have” questions. It is always a good practice to inform respondents that these “nice-to-have” questions are optional. Most sample vendors would not let you ask questions to confirm respondents’ email or phone, etc. However, you are free to do that with your own panel.
Still, we encourage you to not make this a regular practice. If you are doing a raffle, then you must capture pertinent information to inform winners in this section. Raffle for respondents acquired via a sample vendor are handled differently. Most vendors will ask you to do the drawing and provide them with unique respondent IDs (this is an ID issued by sample vendor to track their panel members).
Important Note: If you have screeners or quotas based to demographic questions that
would need to be placed in the opener section. Most commonly these are gender and
Separating myth from fact
Classification questions also have a lot of myths surrounding them, resulting in surveys that fail to gather the exact data that the study needs. Here are some of the most common ones we’ve heard.
All classification questions should appear after the main survey body.
Fact: The primary purpose of classification questions is to help you understand how different demographics have different opinions and/or perceptions. It is irrelevant where you place these questions. However, it is a common practice to have some classification sections inside the screener section.
All classification questions must be answered.
Fact: Not necessarily. The question you have to ask yourself is: How important is this classification question to my overall study objective? If it is a key component, e.g., “what is the make/model of your car” in a vehicle quality study, then it needs to be a must-answer question. If having a few respondents skip over a certain classification question doesn’t alter your overall analysis, then you could flag it as an optional question.
Asking all classification questions at the beginning is a bad practice.
Fact: This depends upon the nature of your study and how critical the response rate is. The primary reason for not starting with classification questions is the negative impact on response rate this choice might have. Some respondents have a tendency to drop out when presented with classification questions. In some cases, every classification question is a key component of the study, and without those questions, the survey would need to be dropped. In such a scenario, it would be prudent to have it happen up front. Certain healthcare and financial surveys tend to require more precise classification (health issues, etc.) and lack of those answers would render the survey meaningless.
If I already have answers to all classification data (especially demographics), I should never ask those again.
Fact: It is a good practice to not ask questions when you already have the answer. However, things change. People move, have kids, change jobs, get married, etc., so it is also good practice to confirm those demographics once or twice a year, depending upon your respondent base.
How to Write and Test a Market Research Survey, Part 1: Introduction
How to Write and Test a Market Research Survey, Part 2: Writing the Opener/Screener
How to Write and Test a Market Research Survey, Part 3: Writing the Main Body
How to Write and Test a Market Research Survey, Part 5: Testing the Survey