Getting Started: Questionnaire Design for Inguo

Share on facebook
Share on twitter
Share on linkedin
Share on email

Your questionnaire should be designed to understand cause without bias. Here are some basic tips on how to do exactly that.

 

What Is Your Key Variable?

You should know what you’re trying to understand. Your key variable is typically an outcome that you’re trying to model for, like brand favorability, customer satisfaction or sales.

 

Are Your Questions Quantifiable?

We can accept Continuous, Categorical, and Binary data. When designing your questionnaire, it is okay to collect qualitative information at the same time, however, you will need to remove any data that is not relevant to causation (like text open-ends).

 

Avoid Confirmation Bias

Are your questions leading the respondent too much?  Are they introducing hidden meanings or assumptions that will influence or skew how respondents answer, either intentionally or unintentionally?

If you introduce bias into your survey, Inguo will show causation based on that bias. Working in teams, or having several people review your survey before it is launched, is a smart way to root out confirmation bias. If the causal results are markedly different from your other, more traditional metrics, dig deeper with confidence! You will be sure to unearth insights that would have otherwise been hidden in your dataset. 

 

Qualified Audience

The better and more qualified your audience, the better your results are likely to be. Inguo doesn’t require a ton of data. We need a sample size that is approximately 10-times the number of variables (questions) you will load into our tool.

That said, some multiple choice questions will be converted to multiple variables, which is important for researchers, analysts, and data scientists to consider when determining a suitable sample size for the study.

 

Question Types

Multiple choice questions can be converted to continuous or binary data points, depending on the nature of the question. 

Multiple choice questions that capture sets of ranges, such as income or years or experience, are prime examples of a multiple choice question that can be treated as a continuous scale and a single variable.

Multiple choice questions that allow discrete (different) responses or even allow for more than one selection will be converted to binary. In this case, each option becomes a variable. So if you have 5 options in a single multiple choice question (like brands of peanut butter consumed), each will count as a different variable and will be represented as a discrete node on the graph.

Scroll back to the top of the page