RTC
DVRPC 2012-2013 Household Travel SurveyTMIP Peer-ReviewOctober 29, 2014
Ben Gruswitz, AICPOffice of Modeling & Analysis
Sarah MoranOffice of Modeling & Analysis
Delaware Valley Regional Planning Commission
SURVEY BACKGROUND
• 1-day paper diary survey• 10,000 households goal, 9,384 actual
complete good surveys (almost 900,000 HHs contacted)
• 3 day GPS sub-sample (500 HH goal, 380 actual)
• Address based sampling frame• 12 month roughly equal sample, weekdays• State, area-type, HH-size x income, and HH-
size x auto ownership as control variables• Diary data retrieved by either phone, web, or
• Stratified random sample by State & Area Type
• (CBD, urban, suburban, rural, etc.)• Oversample with higher transit propensity• Transit Score - population density, employment
density, & carless households• Quintile Approach – oversample tracts by an
individual county’s top scoring tracts (top 20%)• Incentive program: 0-vehicle; Low income
(<$35,000); and Spanish speaking
SAMPLING PLAN
SAMPLING PLAN
• GPS data was originally desired to correct for under-reporting
• Conclusions so far:• Data is interesting • Can be good check
against diary data• Has it’s own flaws
DATA QUALITY - GPS
GPS SAMPLE
6
Income RangeGPS Sample
$0 to $9,999 10$10,000 to $24,999 37$25,000 to $34,999 26$35,000 to $49,999 47$50,000 to $74,999 92$75,000 to $99,999 71$100,000 to $149,999 74$150,000 to $199,999 37$200,000 to $249,999 14$250,000 or more 18Don't know 1Refused 44Total 471
Household Size
GPS Sample
1 1182 1993 764 575 146 6
10 1Total 471
Area Type
GPS Sample
CBD 9CBD Fringe 1Urban 93Suburban 260Rural 95Open Rural 13Total 471
• Transponders not always on (take a while to connect)
DATA QUALITY - GPS
• Data processing can under-report trips
DATA QUALITY - GPS
DATA PROCESSING
Data Quality Assurance & Control• Verbatim response
recoding• Geocoding• Tour identification and
classification• Missing data and trip
imputation• Formatting• Misaligned fields
Weighting• Data weighted and
expanded to reflect demographics by county and area type
• Household and Person weights
9
QUERY SUMMARY
Pass Passable Fail
Result Data Table DescriptionDelivery
CountDelivery Percent
Contract Standard
Household Are there at least 9,500 complete HHs? 9,502 100.02% 100%Household Are all HH locations geocoded? 9,502 100.02% 100%Household How many HHs provided their income? 8,856 91.98% 90%Person Do # of people in each complete HH = # of complete person records for that HHID? 9,502 100.00% 100%Person Do all complete HHIDs in Household and Person tables match? 9,502 100.00% 100%Person Are household worker counts less than HH size? 9,502 100.00% 100%Person Are all people over 15 employed or unemployed? 17,962 99.82% 100%Person How many work locations of employed people at fixed work sites geocoded? 8,557 93.86% 95%Person Are all people a either a student or non-student? 21,266 99.89% 100%Person Do all students have a school type? 4,228 100.00% 100%Person How many students have coordinates for their school? 4,037 95.48% 95%Vehicle Do # of vehicles in each complete HH = # of complete vehicle records for that HH? 9,502 100.00% 100%Vehicle Do all HHIDs in Vehicle table match Household table HHIDs? 9,502 100.00% 100%Trip Do # of trips in complete households = # of complete trip records? 9,502 100.00% 100%Trip Do all complete HHIDs in Household and Trip tables match? 9,502 100.00% 100%Trip How many HHs have trips with coordinates or have a household size exemption? 5,719 60.19% 100%Trip How many trips are geocoded among non-exempt households? 59,892 93.17% 100%Trip Is there a mode for each trip? (non-exempt) 64,684 94.79% 100%
WHAT’S AN “OTHER” RESPONSE?
Work Status Codes & Categories
1. Retired2. Disabled/on disability status3. Homemaker4. Unemployed but looking for
work5. Unemployed and not looking
for work6. Student7. Volunteer97.Other98.Don't know99.Refused
Person IDWork Status
Work Status - Other
UUUUU 6 <null>
WWWWW 3 <null>
XXXXX 97 Elderly
YYYYY 1 <null>
ZZZZZ 1 <null>
COMPARING RESULTS
• ArcGIS Model Output
• “XY to Line” tool gives point to point distance
• Model created fields to flag records where Google/Bing geographies disagree
GoogleBing
CHANGE IN PERSON TRIP RATE
13
Household Trip Rate Person Trip Rate
County Households Population Trip Rate Trips
(HHRate*HH) Trip Rate Trips
(PRate*P)
Burlington
165,620 449,117 7.70 1,274,834 2.73958 1,230,391
Camden
188,861 513,660 7.90 1,491,732 2.92433 1,502,110
Gloucester
104,091 288,187 7.52
783,176 2.93160
844,848
Mercer
130,292 366,442 7.98 1,040,044 3.43389 1,258,322
NJ Counties 588,864 1,617,406 7.79 4,589,786 2.98977 4,835,671
Bucks
229,933 625,485 7.57 1,739,928 3.01902 1,888,352
Chester
183,793 499,548 7.81 1,436,169 2.82283 1,410,137
Delaware
206,021 558,874 7.53 1,550,553 2.91042 1,626,561
Montgomery
308,083 799,886 7.58 2,334,522 2.88910 2,310,950
Philadelphia
580,509 1,525,811 6.29 3,653,757 2.71217 4,138,252
PA Counties 1,508,339 4,009,604 7.10 10,714,929 2.83675 11,374,252
DVRPC Total 2,097,203 5,627,010 7.30 15,304,715 2.88073 16,209,923
CHANGE IN PERSON TRIP RATE
14
Burlin
gton
Camde
n
Glouc
este
r
Mer
cer
Bucks
Chest
er
Delaw
are
Mon
tgom
ery
Phila
delp
hia
-1.00
-0.80
-0.60
-0.40
-0.20
0.00
0.20
0.40
0.60
0.80
Motorized Non-Motorized All Trips
Ch
an
ge
in
Tri
p R
ate
CHANGE IN MODE SHARE
Walk Bike Personal Vehicle
Private Transit
Transit School Bus
Other-2.5%
-2.0%
-1.5%
-1.0%
-0.5%
0.0%
0.5%
1.0%
1.5%
2.0%
0.0%0.2%
-2.0%
0.0%
1.6%
-0.3%
0.4%
Ch
an
ge
in
Mo
de
Sh
are
15
Top Related