How to Set Up A-B Testing
A-B Testing is powerful tool for eliminating bias, testing a hypothesis, or simply trying things out to comparing the results. A-B testing is sending respondents on one of 2 or more paths through your survey, and asking different questions depending on the path.
Fluxey allows you to include build powerful A-B testing into your surveys. How complicated your test gets is up to you, you can have multiple A-B tests, test within tests, or evn a test with 30 different paths. But they may be getting a little ahead of ourselves, to get started, lets look at how we build a simple survey with an A-B test with two pathways.
You can find the A-B test options on the ‘Link’ page, along with the other Link Logic options. Where your A-B test starts will depend on whether you have chosen to build you survey with one question per page, or multiple questions per page. If you are using one question per page you can start an A-B test on any entry, and if you are using multiple question per page you can start an A-B test on a Page Break entry.
For our example here we have created a survey with multiple questions per page. This survey will have two paths, lets call them path A and path B. Everyone will start on page 1, however after survey particants complete page 1 our A-B test begins. Path A will lead to the questions on page 2 and path B will be the question on page 3, survey particants will either be sent to path A or path B based on the weighting we assign to each path. Once everyone has completed their respective paths they will then all get sent through to Page 4.
To get started adding our paths we need to click on the Page Break where we would like our A-B test to begin. When the respondent has completed the page before this page break, they will be sent to one of the pages we select here.
Clicking the page break will open the logic panel on the right. This is where you create your A-B paths. You will be familiar with this panel if you have set up any other type of logic for you surveys.
We would like Path A to start on the next page (page 2), so we will pick “Next” from the list, and click “Add Condition”.
The condition we have added is now shown in our list of active AB conditions.
Normally, when page 2 is completed respondents will be sent to page 3, however we want this to be our second path, path B, so we need them to skip this page. To do this we click on the page break at the end of page 2 and select from the dropdown that we want respondents to go to page 4 when they finish this page.
Next we want to add path B. Path B will start on page 3, so we add that condition to the first page break, just like we did before for path A. Now that we have two conditions, we can assign each condition a weighting. A weighting allows you to send more people down one path or the other.
With our default weighting of 1 for each path we will send an even number of respondents to path A and path B. If we want to send three times as many users down path A, we can change the weighting to 3.
Weightings are calculated as a fraction of the total, in our example we have a total of 4. The value for path A is 3 so the fraction of respondents sent down this path is 3/4. Similarly, only 1/4 of respondents are sent down path B.
If you are more comfortable working in fractions, or percentages, you can use whatever you prefer. In our example entering 3 and 1 is the same as entering 0.75 and 0.25 or 75 and 25.
And there you have it, a survey with a simple A-B test. When a respondent has completed the first page they will be sent down path A or path B at random, with the chances weighted based on the weighting you choose.