1: About Survey Branching
Survey branching is used to route certain respondents to a question or set of questions based on specific variables, such as how they answer a prior question in the survey. This article will discuss how to use question logic to create branching in a survey.
The question in the example below will send respondents to different blocks of relevant questions depending on which of the activities they indicate they use their smartphone for on a daily basis.
2: Add the Control Question
First, add the control question. A control question acts as the starting point of the branching. Respondents will be routed to different questions based on how they answer the control question. Typically, control questions function to sort respondents into categories.
In the above example, a multiselect question (q1) acts as the control question. Respondents will be routed to questions about their shopping habits if they select “Shopping”, gaming habits if they select “Games”, and music habits if they select “Music”.
3: Add Question Blocks
After the control question is programmed, the question blocks (or individual questions) respondents will be routed to must be added. The questions or question blocks should be relevant to the response given in the control question.
The block element can be found in the Structural tab of the Element Library.
Learn More: Block Element
Questions can be added to blocks by using the “add to block” button on the stage.
Blocks can also be added by selecting multiple survey elements by holding the Ctrl or Shift keys and clicking each element in the question tree. With multiple elements selected, an “+Add to Block” button becomes available. Selected elements will be added to a block.
In the below example, a block of two questions related to the smartphone activities was added for each answer option in the control question (q1). Each block can be considered a branch in the survey.
4: Restrict Question Blocks by Logic Conditions
Once the blocks of question are built, they need to be restricted by logic conditions so that only respondents who provide one of the responses in the control question will be shown the corresponding blocks of questions.
If questions are placed in blocks, logic conditions can be set for the entire block, rather than applying them to each question individually. To set conditions for the entire block, first ensure that the block element is selected. Logic conditions can then be accessed in right options panel by expanding the “Show if” dropdown and selecting “+New Condition”
Learn More: Adding Logic to a Question or Answer Option
From the Logic Conditions menu, select the control question from the dropdown and set the appropriate conditions for each block.
The above example shows the logic conditions set for the “Shopping” block. Only respondents who selected “Shopping” in the control question (q1) will be shown the questions in the block (q2 and q3). This will be repeated for both the “Games” and “Music” blocks.
5: Test the Survey Branching
Once the survey has been programmed, the survey branching should be tested to ensure that logic conditions were set up correctly. There are two methods for testing the logic conditions: testing the survey manually and running simulated data and checking conditions in crosstabs.
5.1: Test the Survey Manually
To begin testing the survey manually, select “Test Survey” from the test menu.
Then, set up the test environment and select “Show Survey with Tools”.
Then, test the survey branching under all possible conditions to ensure that logic conditions are set up correctly and respondents are being routed to the appropriate questions.
In the above example, the “Shopping” and “Music” options were selected. As a result, our test respondent should only be shown the Shopping and Music blocks and not be shown the Games block, as shown below.
This process can be repeated to check all possible conditions in the survey branching.
Learn More: Testing an Online Survey
5.2: Check Conditions in Crosstabs
Survey branching can also be tested using a combination of simulated data and crosstabs. First, select “Simulated Data” from the Test menu and run simulated data.
Learn More: Simulated Testing
Then, create a new Crosstabs with a segment for each answer to the control question.
In the report, check each each segment against the base for each question in the corresponding branch.
The example above shows the table for q2, a question in the Shopping block. The “Shopping” segment (Segment B) matches the “Total” segment (Segment A), so we can be sure that only respondents who selected Shopping in the control question (q1) are being shown q2.
This process can be repeated for all answers in the control question and the question blocks associated with them.
Learn More: Checking Condition Logic in Crosstabs