Pivot designs in transport mode choice research
Generic SP attribute levels feel unrealistic to respondents making real travel decisions with specific costs and times. Pivot designs anchor levels to the respondent's actual journey.
What pivot designs are, why they produce more realistic trade-offs than generic SP designs, and how to implement them in SurveyEngine.
Knowledge Base -> Experiment Design -> Transport
Ben White, 07.07.2026
The problem with fixed attribute levels
A standard SP design presents fixed attribute levels to all respondents - travel times of 30, 45, and 60 minutes. A respondent whose actual journey takes 90 minutes finds the 30-minute alternative implausibly good.
Pivot designs resolve this by defining attribute levels as either percentages or offsets from each respondent's reported current journey. For example,a level of -10 minutes means 10 minutes less than the respondent's actual travel time.
Why pivot designs improve SP surveys
Pivot designs produce more realistic choice sets. Respondents engage more seriously with alternatives that are plausible improvements or deteriorations from their actual situation.
Pivot designs also enable joint RP/SP estimation. The revealed preference data is the respondent's current mode choice at their actual attribute levels. The SP data pivots around those levels, sharing the same preference parameters.
Building a pivot design in SurveyEngine
Step 1: Collect the reference journey at the start of the survey - mode, travel time, cost, frequency of delays.
Step 2: Define pivot offsets as percentage or absolute changes from the reference value. Confirm no level combination produces implausible or negative values.
Step 3: SurveyEngine computes actual attribute levels for each respondent in real time using reference values and design offsets.
Step 4: Display computed absolute values in the choice tasks - respondents see actual times and costs, not abstract percentages.
Step 5: Export reference and choice data together. Both are needed for joint RP/SP estimation.
Worked example - urban commuter mode choice
A mode choice study for a proposed bus rapid transit corridor collects reference journey data from 600 car commuters. The SP experiment pivots with offsets of -20%, -10%, 0%, +10%, +20% for time and cost.
Joint RP/SP estimation produces value of travel time estimates anchored to observed behaviour. The SP scale parameter is estimated at 0.68 relative to the RP scale - SP choices are approximately 50% noisier than revealed preferences.
References
Hensher, D.A., Rose, J.M. and Greene, W.H. (2015). Applied Choice Analysis. Cambridge University Press.
Rose, J.M. and Bliemer, M.C.J. (2009). Constructing efficient stated choice experimental designs. Transport Reviews, 29(5), 587–617.
Ready to build a pivot design for your transport SP study? Log in to SurveyEngine to configure offset-based levels.
Or Contact us at support@surveyengine.com — we're glad to help.