From: Deviations from typical paths: a novel approach to working with GPS data in the behavioral sciences
Formalizing questions, definitions, and assumptions | ||
Step 1 | Establish a research question | The research question facilitates the process of formalizing the definition of a typical travel behavior and the purpose for studying variation in travel behavior. |
Step 2 | Establish theory and formalize assumptions of the target dynamics of travel behavior | Theory is used to characterize the behavioral dynamics under question including the timescale over which the travel behavior unfolds and behavior cycles. |
Step 3 | Operationalize target dynamic and determine a priori groupings of data | Develop the timeframe of interest (e.g., morning commute or combined daily commutes) and grouping (e.g., weekday commute and weekend trips) units based on theorized habitual travel behavior and potential day to day variability in that behavior. |
Analysis of location tracking data | ||
Step 4 | Defining and determining typical paths | Estimate or define a typical path for each grouping within person—e.g., if the target dynamic is theorized to differ on weekdays and weekends, two separate typical paths should be obtained for each person. |
Step 5 | Calculate deviations | One deviation is calculated per observation using a distance metric to find the distance from the observed point to the typical path. |
Step 6 | Analyze deviations | Analysis should reflect the research question, theory, and groupings established. Multilevel models are well-suited to intraindividual studies of variation, with time-varying covariates allowing for study of covariation over theorized timescales. |