Randomising unlabelled alternatives

Experiments
1

Hi Benno,

I’m setting up a generic DCE with two unlabelled alternatives. I would like to know whether it is possible to randomly swap only the levels between the two alternatives, while keeping the column labels fixed.

For example, the column labels would always appear as “Column A” and “Column B” in the same left-right order, but the attribute levels shown under these two columns would be randomly swapped between choice tasks.

I saw that the tutorial explains how to randomise the alternatives as a whole, but I’m not sure whether SurveyEngine can support this specific approach where only the levels are swapped while the labels remain fixed. And we would also need to record whether the levels were swapped, so that the data can be analysed correctly.

Is this possible in SurveyEngine?

Thanks!

Best,

Rui

2

Rui this is entirely possible using the tutorial- but it would require specific coding

A better way is do do it in your design. That way you have full control and auditability.

Very simply

  1. open your design in excel
  2. duplicate it to the bottom half
  3. swap the columns for the bottom half duplicate set.

Now you will have double the design rows and therefore double the blocks - but now - the randomisation will happen automatically in block allocation and you’ll be able to audit this and importantly separate lexical alternative biases from ‘true’ alternative biaes.