Data Lab: Base N

Virtual Questions (VQs) allow you to explore your data further by combining or splitting the responses to questions you did ask in order to answer questions that you didn't actually ask in your survey.

Depending on what you want to achieve with your VQ, you may want to set custom bases for the percentage analysis rather than calculating out of your total sample (N). Regardless of whether you use the default or set a custom base for your VQ, response count data will remain constant.

Base N (Default)

When you create a VQ from scratch, it starts off using the Base N option, which means the analysis is using your total sample size as the baseline for any percentage calculations. This is good for comparing responses or demographics across all respondents, independent of the questions they saw or the choices they made within the questionnaire.

Customizing Your VQ Base

Changing the base affects the summary percentages by filtering responses and only calculating out of the respondents to whom the question applies, rather than the entire sample size.

A common use case for custom bases would be analyzing monadic legs - when you want to calculate based only on the set of respondents who were exposed to a monadic leg instead of the total survey respondents. This feature can be applied anytime you want to focus on sub-populations within a survey.

Example: if you have [skip], [show if], or [group] logic in your survey and Q5 was only answered by 200 respondents out of the total 400 respondents collected, you can adjust the base of your VQ to match that sub-population.

VQ Presets

You have the option of selecting from presets when creating a VQ, and in some cases, these presets will automatically set a custom base. For example, the platform autodetects group logic and Smart Loops and will create presets so that you can quickly and easily compare the different runs or legs of your survey. In the example below, the VQ preset for top 2 box purchase intent of three different products is selected in one click. Note how the bases are automatically customized for each leg at the answer level.

Manual

You can customize the base at the answer level and the total VQ level whether or not you use a preset option.

Answer level bases are set just above the text field as shown here:

• Click the "base" text on an answer variant.
• Custom bases can be defined by logic (Q2) or with numbers (400), but expressing it with logic is a better way due to its dynamic nature - it’ll stay relevant if some of the respondents are added/rejected or filter settings of the report are updated. Using Number, on the other hand, is static and could lead to misinterpretations later.
• After you click the “base” link we’ll pre-populate the field with “base =”. Enter the logic expression describing your desired base or total N for this answer.
• Changes are saved by pressing the checkmark or canceled by hitting the X.

Using a base on the overall VQ, rather than at the answer level, is also easily accomplished by clicking “Base: N” below the answers:

You can use logic, variables, demographics, and more to create your custom bases.

Sum of all VQ answers

This setting allows you to quickly calculate the percentages shown in the chart out of only respondents who fall within at least one of the defined answers within the VQ.

With Base: N active, the percentage shown in the chart will be calculated out of the total responses. In the example below, that means n=750, with percentages 53.6% and 19.7% as shown below.

To calculate the percentages based only on the number of respondents relevant to the VQ, click the drop-down and change "Base: N" to "Sum of all VQ answers."

Total response counts for each answer have remained the same (402 and 148), but the percentage is calculated out of a total of 550 respondents.

Using custom bases can be helpful when you are focused on a specific subset of your respondents, or only want to calculate percentages based on exposure to items. However, keep in mind that this explicitly excludes a portion of your sample whose responses might provide important context to your overall research aims.

"Everyone else"

When you create a VQ, there is a checkbox labeled “Everyone else” located under your answer variants.

Use this option to quickly and easily create a comparison between the subset of your sample to whom the VQ answer variants apply, and the rest of your respondents. Using “Everyone else” with Base: N is a great way to ensure you have sufficient context when making interpretations of your data.

Questions about using custom bases or Virtual Questions? Don’t hesitate to reach out!