GPS-based household surveys are becoming the preferred means of conducting household travel surveys to support travel demand modeling. The merits of these survey efforts are well understood and have been discussed at length in multiple forums.

However, what has not been discussed widely are the detailed analyses required to convert a GPS point-file to a “ready-for-use” model dataset. This paper aims to address these questions by documenting analytical and research findings from an analysis of GPS-efforts in Cincinnati and Minneapolis. Key topic areas include:

1. Outline Data Needs at the Outset. Every agency should outline specific data needs before conducting a GPS-based household travel survey. This is vital to streamline the questionnaire; design an appropriate prompted recall approach; and analyze the data to support its own policy decision making.

2. Identify Target Markets. A suite of methods and an approach of combining data from different sources are necessary to target the regional population and specific market segments in each region. Hard-to-reach segments are different than those in a traditional survey and may include seniors unfamiliar with the technology, school age children who cannot carry GPS units in schools, and large households. If diaries are used for specific segments, they should be designed to be compatible with the dataset derived from the GPS survey.

3. Develop Clear Instructions. Respondents must have a clear understanding of when the GPS-devices must be used, by whom, and for what duration. This will prevent commonly observed issues such as “forgetting to take my device during lunch” and households carrying only one GPS during trips with other household members making it difficult to determine joint trips.

4. Collect Location Data. Traditional household surveys collect socioeconomic data during recruitment. However, additional questions pertaining to common locations and activities conducted at these locations are necessary to improve information about activity and travel behavior in GPS surveys.

5. Perform Detailed Validation Checks. The wealth of knowledge pertaining to activity locations and durations must be used when synthesizing GPS-patterns and converting them into modeling-ready databases. For instance, calculating the number of non-home based trips can be a good way to identify if too many traffic stops being coded as activity stops.

The authors believe that practitioners across the country will benefit from the information outlined in this paper by addressing critical issues early on in their survey process; thereby freeing up time to focus on advanced analytics.