How Busy is Too Busy? Investigating the Participation of “Busy” Households in Metro Area Household Travel Surveys 14th TRB Planning Applications Conference 5-9 May 2013 Columbus, OH Authors: Jason Minser, Abt SRBI Tim Yeo, Abt SRBI Randal ZuWallack, Abt SRBI Mindy Rhindress, Ph.D, Abt SRBI Jonathan Ehrlich, Metropolitan Council Kimon Proussaloglou, Cambridge Systematics
21
Embed
How Busy is Too Busy? Investigating the Participation of “Busy” Households in Metro Area Household Travel Surveys 14th TRB Planning Applications Conference.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
How Busy is Too Busy?
Investigating the Participation of “Busy” Households in Metro Area Household Travel Surveys14th TRB Planning Applications Conference 5-9 May 2013Columbus, OH
• Sponsoring agencies include MPOs, DOTs, and other planning agencies
• Comprehensive inventory of households’ 24-hour travel
• Two phase study designRecruitment: Inventory of household, vehicle and
person characteristics
Follow-up: Inventory of individual household member travel for a 24 hour period
• Data used for travel demand forecasting
Abt SRBI | pg 3
Typical HTS Protocols
Advance Letter(Unmatched only or Both)
Recruitment(Phone and/or Web)
Reminder to Travel(Phone and/or Mail)
Follow-up/Retrieval(Phone, Web, Mail)
Abt SRBI | pg 4
Points of Response / Non-Response in HTS
Advance Letter(Unmatched only or Both)
Recruitment(Phone and/or Web)
Reminder to Travel(Phone and/or Mail)
Follow-up/Retrieval(Phone, Web, Mail)
?
?
?
?
Abt SRBI | pg 5
Factors Affecting Non-Response Rates
• Trust of sponsoring government agency/agencies
• Ability to reach household representative(s)
• Perceived importance of survey
• Burden of reporting
• Household composition
• Travel day specifics (e.g., day of week, planned activities)
• Busyness?
Abt SRBI | pg 6
What is Busyness?
• Is actual or perceived influencers that obstruct a household from reporting on their travel day
• Influencers could include, but not be limited to:Hours workedTypes of activitiesHousehold compositionHome ownershipPresence of childrenEmployment statusOccupation status
• Filling out diaries is not an “essential task” for a household, if busyness is perceived, little to no recourse
Abt SRBI | pg 7
Importance of Understanding Busyness
• Helps transportation researchers: Determine the most effective corrective measures in order
to improve study participation Better predict trip characteristics of non-respondents Evaluate possible correlation between busyness and quality
of respondent-provided travel data
• Research Questions How do households’ travel days differ? What are they doing
to be so “busy”? Who are these households? What do they look like? What we know from who responded, can we predict what
kind of travel we missed?
Abt SRBI | pg 8
About the Data
• Address-based sampling – three tiered stratification by region, household size, and number of vehicles
• Multiple Methods Recruitment – phone and web Retrieval/Follow-up – phone, web, mail back
• 25,000+ households were recruited to participate in 24-hour travel diary (all persons 6 years of age or older)A total of 20 activities were available to choose from14,000+ households returned travel diaries
• Households were randomly assigned a weekday and distributed evenly throughout the week
• Anybody can have a day like any of these, but there are household characteristics that we can use to predict busyness
• Build a model to estimate the probability of having a hectic day, routine day,day out,an “all work, no fun day”, oran easy day
Abt SRBI | pg 15
Logistic Regression Model
Hectic Routine Day outAll work, no fun Easy day
Kids between 6-17 x x x xKids under 6 x x x x xOwn house xNumber of drivers x x xNumber of vehicles x x x xPresence of a person with a disability x x x xHomemaker present in the house x x x x
At least one person telecommutes one or more times a week x xNumber retired in hh x x xNumber employed full time in hh x x xNumber unemployed in hh x x xMax education attainment in hh x x x x xMin education attainment in hh x x xYoungest adult in hh x x xHH Type x x x x
Abt SRBI | pg 16
Logistic Regression Model, cont’d
• Logistic regression model provides household probabilities of having each class of day
• Use probabilities as weights:
i.e., HH 1 would count more toward “Day out” and “Easy day”; HH 2 would count more toward “Hectic” and “Routine”
HH 1 HH 2
Hectic 10% 65%
Routine 15% 25%
Day out 40% 3%
All work 5% 5%
Easy day 30% 2%
Abt SRBI | pg 17
Logistic Regression Model, cont’d
• Sum of the probabilities for the responding households = estimated distribution of days
• Applied model to non-responding households
• Busyness is clearly driving non-response for at least some households
Hectic Routine Day outAll work,
no fun Easy day
14.7% 10.0% 25.4% 20.7% 29.3%
19.2% 11.6% 21.5% 22.0% 25.7%
Abt SRBI | pg 18
Discussion• Stark differences between busyness classes
Trip making Number of trips
• Large households are indeed the busiest Especially when older children are present
• Seniors are primary demographic in Day out and Easy day Represents two leisure day types: active and less active Both are overrepresented No universal truths in this group
• Busyness is obstructing participation in at least some households Missing these households is driving down trip totals
• Better consideration given to how much we ask households to tell us about their day
Abt SRBI | pg 19
Next Steps
• Apply busyness model to other regional HTS data to look for differences and similarities
• Apply model to population statistics to pre-determine potential make-up of travel prior to fielding Examine the impact on the recruitment survey
• Offer incentives based on multiple characteristics of households
• Identify data reporting issues across the different classes