In addition to Juliette's suggestion, I'm wondering if a Table Calculation (SIZE()) could be used to get what you want--see the link above for an example. If you post your example, I can take a stab at it for you.
Or if anyone else has a similar set of raw data they can post in a .twbx file, I'd love to take a look at this.
Hi Matthew, and thanks for your reply and wanting to take a stab at it!
Here is a simple table with test data that should generate a chart where the value/column 2 is missing. As mentioned, the scale used in the survey is:
1 = To a small degree, ..., 5 = To a large degree.
RespondentNr Question Answer 1 Question 1? 4 1 Question 2? 4 1 Question 3? 4 1 Question 4? 5 2 Question 1? 3 2 Question 2? 1 2 Question 3? 1 2 Question 4? 1 3 Question 1? 5 3 Question 2? 4 3 Question 3? 5 3 Question 4? 4 4 Question 1? 4 4 Question 2? 4 4 Question 3? 3 4 Question 4? 3 5 Question 1? 4 5 Question 2? 4 5 Question 3? 4 5 Question 4? 4
Please extract the data in a Tableau packaged workbook (.twbx file) with this view created, and I'll take a look.
I don't believe the method I was considering will work for this data set--sorry.
I was having high hopes that your solution would do the trick as mine seemed to work except when you added the question number in and then it fell apart!!
I may or may not have a solution to your problem, but I wanted to first ask you a question to make sure I'm on the right track.
In regards to the situations where you receive no responses, are you looking to replace those non-responses with a "1" because you assume no response means they would have answered "1"? Therefore, you survey data set responses would have responses from "1 to 5", thus eliminating the "0" responses and making your survey data the way you want it before you put it into Tableau.
Please advise. Thanks.
I am still struggling with this issue myself. I had the same feedback from support to add dummy data as a solution. I have had this work with relative ease in the past, but this time I used Tableau Prep to clean my survey data, and I am concerned about the clean-up needed on the Viz side of things as a result of the dummy data as highlighted previously.
For me, the Table Layout -- Show Empty columns did not work.
Does anyone know if they've been working on this?
Do we have feedback on this issue yet? I'm trying to process survey data and have the same issue but still couldn't fix it....
As far as I know there is no other solution than to join a data-set with all possible answers to the actual data-set.
The resulting dummy data interferes with other calculations, so it is handy to generate a validation ID for real cases, e.g. as [ValidID] which can be used instead of [Number of Records].
Similar to how you would get a weight in your data. See: https://www.datarevelations.com/working-with-weighted-survey-data.html
If you have to use both you end up with something like [Number of Records] = [ValidID]*[weight].
It works, but it's still far from working well and requires a whole series of manual adjustments when needed.
To make it a bit more convenient, I use [ValidID] as Context Filter for all parts, where I don't need this workaround.
Unfortunately, most of the time it is easier to show a comment under the question like: "'Do not agree at all (1)' was not mentioned.". Or to rebuild a shown scale.