Before you get started

Identify the goal of your survey - 

What question do you intend to answer?

e.g., How do people feel about minivans?

Identify your target audience - 

Who would you like to to answer your question?

Identifying your target audience is a crucial part of the survey sampling process. If we're asking how people feel about minivans, we can reasonably eliminate those under the age of 18 as potential consumers. We could further identify a household income range that would be most likely to purchase the mid-sized sedan we are researching. Finally, we could assume our target market are those with children. 

In this example, we would use the age, age and gender of child, and household income targeting.

Creating Survey Questions

The First Question!

With the first question, its generally a good idea to provide information to the respondent about the survey and to have the respondent provide consent to participate. This allows the respondent to understand the nature of the project and is an affirmative commitment to participate. 

Types of Questions

Single Select - These are multiple-choice questions where only one option can be selected. These types of questions should be used when there is a set number of possible answers and each answer is exclusive.
e.g., Please select the minivan brand you most associate with comfort.

Multiple Select - These are multiple-choice questions where more than one option can be selected. As long as the respondent chooses at least one option that is marked as "Accept", the respondent will be be allowed to continue in the survey. These types of questions should be used when there is a set number of possible answers and the answers are not exclusive.
e.g., Please select all minivan brands that you associate with comfort.

Open Ended - This type of question should be used when there is not a fixed number of answers to a question. It is important to note that open-ended questions will be more difficult to analyze as part of a data set.
e.g., Please briefly describe why you associate the brand you selected with comfort?

Avoiding Leading Questions

Creating leading questions is an easy mistake to make and can have a major impact on the data. A leading question is any question that leads a respondent to choose a particular answer.
For example, the question "Do you currently drive a minivan?" can be more effectively written as a multiple choice question such as, "What type of vehicle do you currently drive?"
In general, it is best to avoid avoid Yes/No questions. A notable exception would be the first question of the survey, when we are seeking the respondents consent to participate.

Order Matters

The order in which questions are asked is crucial to creating an effective survey and ensuring a great experience for respondents. The screening or qualifying questions should always be up front. If a question will cause a respondent to be disqualified from the survey, it is best to disqualify these respondents early. This practice reduces the amount of time that respondents waste and ensures that you are able to quickly get to those respondents you are looking for. 

Red Herrings / Attention Checks

A red herring or attention check is a question that seeks to ensure that a respondent is paying adequate attention and providing thoughtful responses to survey questions. These are not trick questions - they are straightforward questions with a clear acceptable answer.
For example, a Single Select question could ask respondents "Please select the answer 'Blue' to this question".
This is a straightforward question that establishes the respondent is not simply selecting random answers to the questions and is reading and responding to each question appropriately. If red herrings/attention checks are to be used, its best to put the first in place early.

Survey Language

Always use language tailored for your audience. When targeting younger audiences, avoid complex diction or being overly verbose. Be careful to not use industry-specific language in surveys targeting the general population.

Did this answer your question?