Sign in
x
☰

Splitting text responses

Updated at 2019-06-26 16:30:51 UTC
Small icon showing a list of names including Giorgia Lupi, Mike Bostock, Tufte, Bret Victor, and others - representing thought leaders in data visualization mentioned in concatenated text format.

Survey data sometimes has multiple answers combined into single strings with a delimiter. For instance, the Data Visualization Survey by Elijah Meeks asks respondents who they consider to be a thought leader.

Example

This was elicited as a multiple response section. But, the data shows the multiple responses catenated with a semicolon:

Protobi data view showing a question "Who do you look to as a thought leader in data visualization?" with concatenated responses. Each row displays multiple names separated by semicolons with uniform 0.4% percentages, including combinations like "Mike Bostock;Tufte;Bret Victor" and "Nathan Yau;Hadley Wickham". A Filter box and Apply button appear at the top.

We might wish to see how often each person is mentioned overall, regardless of who else they might be mentioned with.

Split at delimiter

Press the blue edit icon for the element and select "More properties...". Then add a comma next to the "split" field.

Element properties dialog box for a Country element showing various configuration fields. The "split" field is highlighted in blue and contains a semicolon (;) character, which will be used as the delimiter to split concatenated text responses into separate values.

The data is now separated by the delimiter and becomes a multiple-response element. Now we can see how often each person is mentioned individually:

Protobi data view after splitting showing individual thought leaders with their mention frequencies. Mike Bostock leads at 26.7%, followed by Tufte at 17.3%, and Alberto Cairo at 15.8%. Blue horizontal bars visualize the percentages, with Filter and Apply controls at the top.