National Household Travel Survey Data Program

DataPalooza 6/4/2014 : Analyzing Integrated Data Part 2

Transcript

1. Data Palooza Adella Santos NHTS Program Manager Hilton Washington Regan National Airport June 4, 2014 National Household Travel Survey: Data Program 6/27/2014
2. Today’s Discussion • About the Program • Challenges to Current Methodological and Analytical Approaches • Exploratory Research needed: Long Distance Data • Technology impact on Governmental Surveys • Uses of the Data for Policy Consideration 2
3. • Authoritative Source of statistical data on the characteristics of the American Public travel • Conducted since 1969- every 5-7 years, last conducted in 2009 • Primarily Funded by FHWA and Add-on Participants NHTS Program
4. NHTS Program National Study (2009) 25,000 Sample Size Add-On Program (2009) 125,000 Sample Size
5. States and MPOs: Have an opportunity to purchase samples to the household travel survey Increased sample size can help to develop travel estimates at small geography levels Add questions that are unique to their local area Add-on Program
6. Data Content 6
7. Collect Trip Tours Trip 2 Walk Work Home Trip 3 Walk Lunch at Restaurant Trip 4 Walk Subway, car Trip 6 Car Trip 7 Car Grocery store Daycare center Trip 1 Car, subway, walk Gas Station Trip 5 Car
8. • Gather information from the public covering the entire nation using a complex survey design. • Traditionally, a probabilistic-based survey is required by OMB • Probabilistic-based design is needed in order to make scientifically-justified conclusions and inferences for the nation’s population or selected subpopulations. Challenges of Government Surveys
9. • Surveys are very expensive to implement • More expensive when the design requires estimates with desired bounds on precision and a need to make inferences with sufficient confidence Biggest Challenge: Cost
10. How can household travel survey design be improved to account for … – A changing and more diverse society? – Influx of new technology affecting measurement errors? – Changes in economy, Federal budgets, etc.? Complex Nature of a design
11. Can it lead to significant improvements in … – Data quality? – Response rates? – Reduced respondent burden? – Estimating travel behavioral parameters? Adopting New technology
12. Data Quality Considerations 12 Sample Size Approach Questionnaire Development Survey Objectives • Travel demands forecasting • Long range plans • Travel trends • Consider role of respondent and interview mode • Cognitive issues • Application type to be used • Format • Sample size • Stratification of sample • Missing data approaches • Coverage issues • Daily integration of sample data Survey Management
13. • Last Long Distance Travel survey was conducted in 1995 • 2001 NHTS had a subset of questions on Long Distance travel • Long Distance is considered a rare event for the general population. • Deal with many measurement error issues. 1995 American Travel Survey
14. • Project looked at innovative, new ways to design a survey to generate high- quality data, and to analyze those data so that the survey objectives can be still achieved despite greater cost constraints. Recent Exploratory Research for Long Distance Travel
15. • Used Smart phone • Used Face book L/D Exploratory Research
16. Long Distance (L/D) Project Focus
17. • Multi-frame and multi-mode design approaches as part of a prospective design – Address-based sampling – Dual-frame landline (RDD) and cell phone samples • New data collection apps can improve data quality, response rates, respondent burden, and bias reductions (but need to be designed in a way to attract their use) Options Considered
18. • Respondent volunteered to provide their home location address using the GPS coordinates as a starting point. • Upload the application into their android phones. • Application was designed to create a ping each time a 50 mile trip was made, • Respondent was then instructed to answer about 6 questions on their long distance travel. Smart Phone
19. • (FB) app were designed to trigger a L/D trip by considering the location “tags” that survey participants place on their posts and photos, signifying the physical location (or place that the post or plot is representing.) • FB was able to link those tags to their internal database in order to assign GPS coordinates to the location. Used Face Book
20. • Those GPS coordinates are then used to compare to the coordinates for the centroid to the region associated with the ZIP code provided by the participant during the app download and registration, and when the straight-line distance between the coordinates exceeds 50 miles, a L/D trip is triggered. FB Coordinates
21. • Accuracy of a trip purpose imputation algorithm can exceed 90%, depending on the amount of available input data to the algorithm and the number of trip purpose categories Trip Purpose Imputation Research
22. • Data fusion techniques (bringing together data collected from probabilistic samples with “opt-in” data from a volunteer panel) Data Fusion
23. Questions and Comments 23 Adella Santos 202-366-5021 Adella.Santos@dot.gov