The Net Promoter Score (NPS) is computed as the difference between two sets of percentages (most commonly, the proportion of people to rate a brand as and 8 to 10 less the proportion to rate the brand as a 6 or less).
A data set in Q containing standard Net Promoter Score data (i.e., ordinal data on a 0-10 scale where 0 is not all likely to recommend and 10 is extremely likely to recommend).
Method 1: Manually computing NPS
The Net Promoter Score (NPS) is computed as the difference between two sets of percentages (most commonly, the proportion of people to rate a brand as and 8 to 10 less the proportion to rate the brand as a 6 or less). It is computed as follows:
- In the Blue Drop-down Menu select the likelihood to recommend question. It should be a Pick One or Pick One - Multi.
- Right-click on the table and select Values to open the Value Attributes dialog box.
- Fill in the Value column as follows and press OK
- Replace values 0 through 6 with -100.
- Replace values 7 and 8 with 0.
- Replace values 9 and 10 with 100.
- Right click on the table and select Average from either Statistics - Right or Statistics - Below.
The table will now show the NPS.
Method 2: Automatically computing NPS
Use this QScript in Q's online library Create New Variables - Recode Net Promoter Score (NPS) Variable(s).
Significance tests on NPS questions
A common misconception is that the above is a way of "tricking" Q into computing NPS. However, the above is not really a trick as such, but is based on a recognition that the NPS can be computed either as a difference in percentages, or, as the average of a 3-point scale, and that both are, in a mathematical sense, identical. A key benefit of viewing the NPS as the mean of a three point scale is that Q's automatic significance testing is automatically testing for differences in NPS score (i.e., although NPS is described as a difference in percentages, from a statistical testing perspective it is just a three point scale).