Sampling Low-Incidence Travel Groups in Household Travel Studies

“Low-incidence” travel behavior is difficult to capture in a traditional household travel study (where typically one day of travel is collected from a representative sample of households in a region). Behaviors or travel choices that do not occur frequently may fall into different categories:

· Behaviors that a small number of people participate in or consider in their travel choice set. This might include travel modes that are used frequently by a small number of people (bicycling, vanpool, etc.) or new or emerging travel modes that may not be widely used yet (carshare, rideshare, e-bikes, automated vehicles, etc.)

· Travel behavior in the context of complex analytical structures (complicated tour patterns, decisions that are related to household interactions)

· Behaviors that a large number of people may participate in, but may not engage in frequently enough to be captured in a one-day travel diary (long-distance travel, trip replacement behavior such as home delivery of goods and services, etc.)

It is important to be able to collect more observations of these behaviors to better analyze the current and future demand for different types of travel; to better understand how these rare or new types of behaviors affect the overall demand for transportation resources; and to understand how these emerging alternatives may affect “traditional” transportation behavior. Because these behaviors are rare, alternative methods for sampling enough observations in a regional travel study need to be identified and tested. Despite their low-incidence, many of these alternatives have outsize effects on the transportation system.

It should also be noted that some behaviors are related to more to survey design than to sampling - for example, it may be easy to sample the types of people who participate in long-distance travel, but a multi-day survey design or other survey design may be better for capturing these trips which may occur infrequently. The variety of challenges in observing different types of rare behaviors may necessitate separate and distinct research efforts to focus on methods that address the primary challenges for a given type of behavior.

Research objectives are to identify and test methods for sampling people and incidences of rare or emerging travel behaviors, including how to incorporate this into household travel survey data collection. This may involve looking at non-traditional sampling methods or survey designs, and overcoming the practical and logistical barriers to finding and collecting samples of rare behaviors. Research objectives should also consider how the data can be weighted and analyzed on its own and within the context of a traditional “representative” household travel survey dataset.

Potential research tasks may include:

· Identification of the types of behaviors that cannot be easily captured in a traditional travel survey

· Determination of which behaviors have sufficient impacts on the transportation system or on transportation model systems as to warrant changes in survey methods.

· Synthesis of non-traditional or non-representative sampling methods to capture low-incidence travel behaviors, and/or supplemental or special generator survey methods

· Pilot test of sampling or survey methods within a household travel study

· Analysis of non-traditional sampling methods (data quality, cost-effectiveness, integration of data with traditional household travel survey data)

· Analysis or recommendations for how sampling methods can be applied across different modes or different regions