In research and data analysis, it’s common to ask survey participants questions in which they can select multiple answers. For instance, a question might ask, “What are your favourite dog breeds?” and respondents may choose multiple options such as Beagle, Dachshund, and Shiba Inu. This results in each respondent contributing multiple pieces of information to a single question. An example of such a question can be seen above.

These types of questions generate what are called multiple response sets.

Before performing the analysis, you need to ensure your data is correctly structured. Each possible answer for a multiple response question should be coded into a separate variable (column). Each variable should also be coded as a dichotomous variable, meaning it only takes two values: “1” for a selected response and “0” (or blank) for a non-selected response.

If you export your data from Qualtrics directly into SPSS, this will be done for you, as seen below:

From here, you can simply define your multiple response set in SPSS.

However, many of our clients export their data to Excel first, perform cursory checks and cleaning, and then import their data into SPSS. This is where things can take a turn for the worse!

The first “gotcha” is ensuring that the data is exported as numeric values and not as choice text in the Download a data table screen:

 

As SPSS is a numbers-based application, it would much prefer a “1” instead of “Beagle”, and while you can recode this in SPSS, it’s an unnecessary step that could lead to introducing errors in your data.

The second “gotcha” is that by default, Qualtrics will export the multiple response question data to a single column:

As previously mentioned, analysing multiple response sets in SPSS requires each variable to be in a column of its own. While this data can be manipulated (and we often do this for clients) it’s a somewhat laborious and unnecessary process.

To ensure this doesn’t happen, choose ‘More options’ in the Download a data table screen:

Next, ensure that a tick is placed next to “Split multi-value fields into columns”:

This will result in each possible response (or dog breed) having its own column, which can now be imported and analysed as a multiple response set in SPSS:

While getting multiple response set data into the correct format can be a bit tricky, a few extra steps can save considerable time when it comes to analysis and gaining meaningful insights.

Share this entry