Implementing pivot designs with variable piping in SurveyEngine
Pivot designs are technically demanding to implement. The reference data must be collected, stored, computed, and displayed correctly before the choice tasks begin.
How to implement pivot designs in SurveyEngine using variable piping and derived value computation.
Knowledge Base -> Survey Building -> Transport
Ben White, 07.07.2026
The implementation challenge
A pivot design requires the survey to: collect reference journey data from the respondent; compute attribute levels as offsets from the reference values; and display the computed levels in the choice tasks. Each step has implementation requirements that must be correct for the design to work.
Common implementation errors: arithmetic errors in the offset computation; display of computed levels in wrong units; failure to validate that computed levels fall within acceptable ranges.
Why implementation quality matters
Implementation errors in a pivot design are not always visible to respondents. A computation that produces a negative travel time or a cost above a plausible maximum may display as a valid number, but the choice task is meaningless.
Testing a pivot design requires testing with multiple different reference values to confirm that the computations produce valid levels across the range of plausible inputs.
TLDR Quick links
Step-by-step pivot design implementation
Step 1: Collect reference journey data using numeric entry fields with appropriate range validation. Reject implausible values at entry.
Step 2: Store each reference value as a named variable with a clear data label.
Step 3: Create derived value questions for each computed attribute level. The expression computes the pivot offset: {ref_time} * 0.8 for a -20% level.
Step 4: Validate computed levels. Add display logic that shows a warning or redirects respondents whose reference values produce extreme computed levels outside the intended range.
Step 5: Test with the full range of plausible reference values - not just typical values. Edge cases are where implementation errors most commonly appear.
Worked example - freight mode choice pivot design
A transport SP survey collects reference journey time, cost, and delay frequency. Three levels are computed for each: -20%, 0%, +20% of the reference value. Testing with reference times from 10 to 120 minutes confirms that all computed levels are positive and within the intended range.
Testing also reveals that a reference cost of £0 (respondents who travel free with a pass) produces computed cost levels of £0 for all three levels - eliminating cost as a differentiating attribute for these respondents. A minimum cost threshold of £1 is added to the computation.
References
Hensher, D.A., Rose, J.M. and Greene, W.H. (2015). Applied Choice Analysis. Cambridge University Press.
Ready to implement a pivot design in SurveyEngine? Log in and follow the choice experiment tutorial.
Or Contact us at support@surveyengine.com — we're glad to help.