Is the difference between customer groups real?
The apparent 0.42-point gap fell to 0.07 once partial responses were included.
What went in
- Files
- 2
- Rows
- 1,857
- Period
- Apr 14 – May 26
- Verification
- 19 checks passed
Source files
- survey_responses.csv1,840 rows
- question_codebook.xlsx17 rows
What came back
- 1,840
- 71.2%
- 2
The finished project
survey_responses.csv
1,840 rowsCheckedThe difference you were testing for is not there. Two you were not testing for are.
The gap in satisfaction between new and long-standing customers disappears once completion bias is accounted for — unhappy new customers were far more likely to abandon the survey, which flattered the group they left behind. Two effects that were not the subject of the survey do hold: response speed predicts satisfaction, and the two reverse-coded questions were being answered inconsistently enough to distort the overall score.
1,840Responses
71.2%Completion rate
3.9×Mean satisfaction+6.0%
Unhappy new customers stopped answering
- 01The new-versus-established gap is an artefact of who stopped answering.
- 02Two questions were being read the wrong way round by respondents.
- 03Response delay predicts satisfaction better than any segment does.
Filed under
Closest to this one
Have something like this?
Drop it in and see what comes back. You only pay once you like it.
No account needed to start. You only pay when you like what you see.