Field notes from the travel app frontier: Build me, break me, impute me

Dissertation on how missing location data in smartphone-based travel surveys can be classified and addressed through imputation, using a 2018 Statistics Netherlands field test among 674 participants as its empirical foundation.

Participants in the field test logged an average of just 8.2 hours of location data per day — roughly one third of the target coverage. Short gaps can be handled with linear interpolation without meaningful bias, but longer gaps require more advanced methods. The dissertation introduces Dynamic Time Warping–Based Multiple Imputation (DTWBMI), which reconstructs missing mobility data by drawing on a participant's own travel patterns as a template, and integrates it into a hierarchical imputation framework alongside conventional interpolation and multiple imputation at the day level. Applied to the field test data, the framework yields substantially more stable and less biased mobility estimates than listwise deletion or naive interpolation.

McCool, D.M. (2026). Field notes from the travel app frontier: Build me, break me, impute me. Dissertation, Utrecht University, doi:10.33540/3285.

Downloads