RNS
Browse Projects > Detailed View

Improving Household Travel Survey Weighting and Expansion

Description:

The increasing challenges associated with recruiting participants to household travel surveys and inducing the participants to complete the data collection efforts are well-documented[1]. Since lower data collection participation and completion rates increase the probability that conclusions drawn from the collected data are biased, the importance of data expansion and weighting is greatly increased. Relying on available external data sources to expand survey data, surveyors seek to address some of the biases introduced by the non-representativeness of the completed survey sample. These techniques are also important for increasing the accuracy of the estimates. However, survey population dimensions that are not included in the survey expansion can be (and often have found to be) significantly different for the survey sample and the population. These differences are likely to directly affect the quality of travel demand modeling and other survey data based analyses. When such survey data are used in the development of activity-based models[2], any biases in the expanded survey data tend to impact both the development and the calibration since household survey data are the primary data source in the overall development of such models. They also lead to decreased accuracy of the estimates.

In the U.S., surveyors have generally relied on household level demographic data such as household size distributions and household vehicle availability levels obtained from Census Bureau data sources[3][4]. Some survey data collection efforts have also included person-based demographic data from the Census Bureau sources, such as age and employment status. Recent research has been conducted to investigate how best to incorporate Census Bureau employment-based variables, including work location and industry classification, in the survey expansion process[5].

However, such research has been limited and focused on very few efforts nationally. More comprehensive research is needed to evaluate the effectiveness of different survey expansion and weighting strategies, including the selection of expansion variables, techniques for addressing multiple expansion variables across different populations (household based, person based, workplace based, and perhaps trip-based), and geographic detail levels.


[1] McGuckin, N. and H. Contrino. Hard to Reach Groups in Household Travel Surveys, Sub-Committee on Household Surveys, TRB, Washington D.C. 2012 http://www.travelsurveymethods.org/ppts/HH/Hard%20to%20Reach%20Pops%20Final.pptx. 28 Accessed May 2016. [2] Bhat, C.R., K.G. Goulias, R.M. Pendyala, R. Paleti, R. Sidharthan, L. Schmitt, and H. Hu (2012) 210 A Household-Level Activity Pattern Generation Model for the Simulator of Activities, 211 Greenhouse Emissions, Networks, and Travel (SimAGENT) System in Southern California. Presented at the 91st Annual Meeting of the Transportation Research Board, Washington, D.C., January 2012. [3] Komanduri A., and B. Selby. Use of Person-level Variables in Household Survey Expansion to support Activity-based Models, Proceedings of the 15th Meeting of the National Transportation Planning Applications Conference, Atlantic City, 2015. [4] Ye, X., Konduri, K.C., Sana, B., and Pendyala, R.M. (2009). A Methodology to Match Distributions of Both Household and Person Attributes in the Generation of Synthetic Populations. Proceedings of the 88th Annual Meeting of the Transportation Research Board, Washington, DC. [5] Komanduri A., and K.Konduri. Using Work Location and Industry Classification Information in the Weighting of Household Surveys using Open Source Frameworks, Proceedings of the 96th Annual Meeting of the Transportation Research Board, Washington, D.C., January 2017.

Objective:

This research will seek to identify the most successful household travel survey weighting and expansion strategies by demonstrating the effects of selecting different expansion variables and processing techniques to a representative group of recent household travel survey datasets. To the extent possible, expansion variables must be obtained from National or other Government data sources that are frequently updated so they may be applied across regions and time horizons. As part of the process, the researcher must review and suggest best practices for the following elements: identify a framework to prioritize variables within the expansion framework, implement best practices to develop effective and accurate control totals (and categories), choose benchmarks to compare different expansion strategies, and minimize extreme weights in the expanded dataset.

The research will help provide insights into the household survey expansion process, including the selection of variables used to expand survey data, the procedures used to combine multiple expansion variables, the comparison of survey outputs along other variable categories, and the effects of the expansion on common survey data analyses. The research will also seek to identify whether the effectiveness of expansion strategies is location specific or if the findings are likely to be transferrable to other surveys. Finally, the research will evaluate the best way to calibrate the factors to enable adjustments at a variety of geographies and establish appropriate rules for trimming large factors.

Implementation:

The researcher will summarize current household travel survey expansion and weighting practices, including the selection of expansion variables and the data processing procedures being used. In addition, the study will examine research on household travel survey expansion and weighting, as well as population and household synthesis methods that can be used for multiple variable expansion.

The researcher will select four to six recently completed household travel surveys across a variety of geographies, demographics, and urban area types, and define some common travel survey based analyses, such as trip generation cross-classification, trip length distribution, tour generation, time-of-day choice, intermediate stop-patterns, tour and trip purpose metrics, and mode choice analysis. The researcher will then apply and assess the effects of alternative expansion and weighting protocols.

The researcher will compare the resulting survey databases by applying simplified common planning applications and travel demand modeling analyses. The researcher must also outline measures to evaluate the effectiveness of various survey expansion techniques. Finally, the researcher will summarize the benefits and challenges of different expansion variables and weighting procedures.

Cost: As the research will primarily rely on previously completed surveys, expensive new data collection will not be necessary. The estimated research costs are between $150,000 and $200,000 and will be devoted to developing and implementing data synthesis (of expansion variables) and survey expansion techniques.

User Community: The proposed research will enable household travel surveyors to produce databases that more accurately reflect the survey population under study. This will improve the quality of planning applications and travel demand modeling analyses performed with the data. The improved data expansion will also reduce the need for ad hoc adjustments and assumptions usually implemented by data analysts during model development.

Effectiveness: The proposed research will enable household travel surveyors to produce databases that more accurately reflect the survey population under study. This will improve the quality of planning applications and travel demand modeling analyses performed with the data. The improved data expansion will also reduce the need for ad hoc adjustments and assumptions usually implemented by data analysts during model development.

Sponsoring Committee:ABJ40, Travel Survey Methods
Date Posted:08/29/2017
Date Modified:04/20/2018
Index Terms:Households, Travel surveys, Weighting, Data collection,
Cosponsoring Committees: 
Subjects    
Data and Information Technology
Planning and Forecasting
Transportation (General)

Please click here if you wish to share information or are aware of any research underway that addresses issues in this research needs statement. The information may be helpful to the sponsoring committee in keeping the statement up-to-date.