4 Replies Latest reply on Oct 28, 2015 7:32 AM by Simon Runc

    Likert Scale Charts

    Brooke Schwartz

      Hi,

       

      I have read/watched all of Steve Wexler's materials on survey data/likert scale charts, and I cannot get my workbook to look like his. Has anyone had any success with this? I set up all of the calculated fields and parameters like his, but still having issues. My workbook is attached. Please help!

       

      Thanks!

        • 1. Re: Likert Scale Charts
          Simon Runc

          hi Brooke,

           

          The only thing you hadn't done was to set the compute using on the Table Calcs (Gantt Percent & Percent) to Score.

           

          ...I've also done a bit of formatting...I love these divergent bar charts, but in my line of work I never get any Likert data (I'm always trying to get my DevOps/Marketing team to create 1-5 surveys...if only to get a chance to create a few with real data)...so I've taken out my 'likert frustration' by colouring your Viz how I would, given the chance!!

           

          Hope the explanation makes sense, but if not please post back.

          • 2. Re: Likert Scale Charts
            Brooke Schwartz

            THANK YOU SO MUCH! Super helpful.

            • 3. Re: Likert Scale Charts
              Brooke Schwartz

              Simon Runc

               

              Simon--

               

              I just compared the percentages in the likert charts you created yesterday with the numbers that are in the Qualtrics report I downloaded from my survey, and they don't match up, but maybe I am not reading the chart correctly. Do the percentages in each segment of the bar represent the percentage that chose that answer option? For example, in the chart you created 12% chose 1, but in reality, only 3 people, or 2% chose 1 in the survey.

               

              Let me know, thanks!

               

              Brooke

              • 4. Re: Likert Scale Charts
                Simon Runc

                hi Brooke,

                 

                I didn't actually check the data!...bad data analyst!!...so, wrongly, assumed there were same number of respondents for each question (looking at the data I can see some Qs were answered by 125 and some by 123)

                 

                In this set up the %age for each question always add up to 100%. Below is a table showing a cross-tab for the 1st 2 questions

                 

                Likert.PNG

                 

                I've also added the Counts to the Viz (both for each reply, and total respondents for each question).

                 

                If you want the Viz to be on counts, not percentages you can change each of the calculations to be Absolute numbers and not percentages.

                 

                So for example. Gantt start would now be

                 

                -[Total Count Negative]

                 

                instead of

                -[Total Count Negative]/[Total Count]

                 

                ...etc.

                 

                When the number of respondents is very close for each question, there will be very little difference, but if they vary dramatically you'll need to take care that the Viz isn't misleading (eg. if for one question there were only 20 respondents who all scored it a 5, say, the blue Bar for 5 wouldn't extend further than another question where only 20% of respondents scored 5, but that count was say 50 people)

                 

                Hopefully that makes sense, but please post back if not.