A Smartphone-based Travel Survey Trial Conducted … · • Sampling targets were limited in most...

2
Emergence of smartphone-based travel survey as a promising alternative to paper-based or web- based self-reporting survey Challenges in this new survey method should be carefully examined Sampling targets were limited in most of existing smartphone-based survey study Introduction Survey Overview & Method Objective (1) Describe the smartphone-based travel survey trial in Kumamoto, Japan (2) Examine the attributes of the smartphone-based survey’s participants We collected 97 samples out of 13,279 households with zero incentive. Data is free of incentive bias iPhone Ver. Android Ver. Participants touch the tap when they start and end a trip Travelers’ location is collected every 100 m. Participants note their travel mode in addition to the start and end of their trip. Location data are collected every 10 second. Acceleration data are also collected. Results With/Without invitation letter With/Without paper/web- based response Device Survey Period Total 1 Oct 10-11 2 Oct 16-18 3 Oct 23-25 Supplemen tary Nov 20-21 With invitation letter attached With Android 20 18 8 46 iPhone 6 7 1 14 Subtotal 26 25 9 60 Without Android 7 8 1 16 iPhone 6 5 1 12 Subtotal 13 13 2 28 Subtotal 39 38 11 88 Purely voluntary participants (without invitation letter) With Android 1 2 3 iPhone 1 1 Subtotal 1 3 4 Without Android 3 1 4 iPhone 1 1 Subtotal 4 1 5 Subtotal 5 3 1 9 Total 5 42 39 11 97 28 (56.0%) 33 (43.4%) 18,432 (49.5%) 344,291 (46.9%) 22 (44.0%) 43 (56.6%) 18,799 (50.5%) 390,183 (53.1%) 0% 20% 40% 60% 80% 100% Participants in both smartphone-based and paper/web-based surveys Members of households that contain the smartphone-based survey participants who failed to complete paper/web-based survey Smartphone-based survey targests (household members) 2010 Population census (Kumamoto City) Male Female Conclusion 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 0-9 10-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89 90-99 100- Proportion Age 2010 Population census (Kumamoto City) Smartphone-based survey targests (household members) Members of households that contain the smartphone-based survey participants who failed to complete paper/web-based survey Participants in both smartphone-based and paper/web-based surveys Participation Rate: Male > Female 0 2 4 6 8 10 12 14 16 1 2 3 4 5 6 7 8 9 10+ Number of participants (person) Number of trips reported during paper/web-based survey (trips/day) Female Male Male Female Total Smartphone- based survey Number of samples 28 22 50 Average trip rate: A 4.14 4.59 4.34 PT survey (Paper/Web- based survey) Gross trip rate: B 2.98 2.61 2.78 Net trip rate 3.26 3.03 3.14 t-statistics for statistical test (difference between average A & B) 1.44 (No statistical difference) 4.04* 3.14* (*: Statistically significant difference at 1% level) KUMAMON A local mascot of Kumamoto Prefecture, and now he is the most famous character in JAPAN. We used him in these apps, named “SUMAKUMA”. Departure & Arrival Tap Departure Tap Arrival Tap Map Tap End Tap ©2010 kumamoto pref. kumamon #3901 ACKNOWLEDGEMENT Research collaborator: the Urban Planning Bureau, Kumamoto Prefecture, and the Urban Policy Bureau, Kumamoto City Graduate students of Kumamoto University: Mr. Kotaro Nohara, Mr. Keisuke Matsuda, and Mr. Shotaro Imura for their development of the iPhone-version Apps and the data analysis Mr. Yoshihiro Sato, Kumamoto University for this poster design Trans Field Inc. for their assistance in the development and preparation of the Android-version App and server management Fukuyama Consultants Co., Ltd. for their support on several matters on the survey Research fund by “Committee on Advanced Road Technology” in the Road Bureau, Ministry of Land, Infrastructure, Transport, and Tourism, Japan Initial results of the first smartphone-based travel survey conducted jointly with a large-scale household travel survey in Kumamoto, Japan Created smartphone-based travel surveys apps for Android & iPhone 13,279 Invitation letters were mailed to households Ultimately, 97 joined the smartphone-based survey with zero incentives Males were more likely to join the smartphone-based survey than females Female and 20s were more likely to fail to complete the survey People employed in manufacturing, energy, medical and public- related service occupations showed higher participation rates The average trip rate for smartphone-based was higher than that for traditional household travel survey Future works: -A discrete choice analysis of smartphone-based survey participation The model could be explained by some demographics described here Sampling strategies and removal of sample selection bias - Increase the sample size Our research team conducted another smartphone-based survey in Kumamoto downtown area with rewards ($5) during Nov. &Dec. 2013 and collected 1,086 samples. Details will be reported soon. Smartphone-based Travel Survey System Survey System Comparison of Travel Survey Method Purely voluntary participant learned of the survey by reading a newspaper article(Oct. 9, 2012) or accessing the survey committee’s website Gender Age Occupation Response Methods Trip Rates Comparison of trip rates among smartphone-based survey and paper/ web-based survey participants Android, the Android logo, Google Play™, the Google Play™ logo are registered trademarks or trademarks of Google, Inc. Apple, the Apple logo, Mac OS are trademarks of Apple Inc., registered in the U.S. and other countries. iPhone, iPad are trademarks of Apple Inc. In addiGon, other logos and names may be trademarks/registered trademarks of other companies. Smartphone-based survey participants tend to recall every trip An active person who travels frequently during the day is more likely to join a smartphone-based survey Trip rate distributions of smartphone- based survey participants. Possible Reason Higher trip rates by smartphone-based survey ExisGng study Consistent finding with existing GPS-based studies Bricka et al. (2009), Stopher et al. (2007), Bricka & Bhat (2006), Stopher & Shen (2011),Du & Aultman-Hall (2007) 0% 5% 10% 15% 20% Agricultural, forestry & fisheries Construction Manufacturing Electricity, gas & water supply Information & communication Transportation & postal Wholesale & retail trades Finance & insurance Real estate and goods rental & leasing Professional & technical services Accomodations, eating & beverage services Living-related & personal services and… Education, learning support Medical, health care & welfare Services, not otherwise classified Government Others Proportion 2010 Population Census (Kumamoto City) Participants in smartphone- based survey - Half of smartphone-based survey participants responded household survey by web - Significantly higher than other participants People in manufacturing industries tend to be more familiar with high-tech devices such as smartphones. People employed by public utilities and government might be expected to participate in this public-services related survey. High participation Rate: 30s & 40s 20s are relatively low besides their high smartphone ownership Summary of Smartphone-based Survey Participants Speculation: Some of female participants became anxious about privacy issues related to the survey and failed to complete the survey 20s participants often fails to complete (=paper-based & smartphone-based) the survey Tracking & Acceleration Data Analysis of tracking & acceleration data are ongoing along with auto-mode detection analysis Kumamoto Paper GPS + Web Smartphone Respondent burden Large Small Small Survey operation cost Large Middle (distribution cost of GPS instruments) Very Small Initial & other cost Basic and traditional New but available Development of app Accuracy & reliability of data Low High High Duration of survey Typically autumn weekday Multi-days: weekly, monthly, yearly Multi-days weekly, monthly, yearly Path/route of traveler Not Available (N.A) Available Available Bus stop, interchange, trip timing information Low accuracy Low response Precisely Available Precisely available Trip purpose Written Select (by web diary) Tap on smartphone or auto voice input Travel mode Written (un-linked-trips are often neglected) Select or auto detect by acceleration data Select or auto detect by accelerometer or auto voice input Selection bias Property of sample Young people are reluctant to reply Monitored person only Limited only to owner of smartphone but largely distributed Person Trip (PT) Survey (Household Travel Survey) in Kumamoto Metropolitan Area (pop= 1 million) in Japan, in the Autumn 2012 Mail-out/mail-back paper-based self-reporting survey (opt. web-based) Final response rate 38.9% Several strategies to increase More than 120 thousand household were asked to join the base survey Out of them, 13,279 household were asked to join the smartphone-based survey as well as paper/web-based household survey No rewards (incentive) because of governmental constraints Current system just records the track, but modification of record by participants’ devices are now being developed Most participants mention battery life problem People familiar with the Internet are more likely to join Approval of Privacy policy & use of terms The response rate for 20s were lowest for the traditional paper- based survey. If smartphone-only-survey are developed, they will be the potential candidate “SUMA” means “SMArtphone”, and “KUMA” means “KUMAmon” or “KUMAmoto” 72.7% 80.6% 81.1% 82.7% 89.9% 95.7% 97.9% 97.4% 95.8% 50.9% 27.3% 19.4% 18.9% 17.3% 10.1% 4.3% 2.1% 2.6% 4.2% 49.1% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Web-based Paper-based Age of head of household survey participant in total ⑥⑦ ⑩⑪ Time Acceleration (m/s 2 ) A Smartphone-based Travel Survey Trial Conducted in Kumamoto, Japan: An Examination of Voluntary Participants’ Attributes Takuya MARUYAMA*, Shoshi MIZOKAMI*, and Eiji HATO** (*Kumamoto University; **the University of Tokyo, Japan)

Transcript of A Smartphone-based Travel Survey Trial Conducted … · • Sampling targets were limited in most...

•  Emergence of smartphone-based travel survey as a promising alternative to paper-based or web-based self-reporting survey

•  Challenges in this new survey method should be carefully examined

•  Sampling targets were limited in most of existing smartphone-based survey study

Introduction

Survey Overview & Method

Objective (1) Describe the smartphone-based travel survey trial in Kumamoto, Japan (2) Examine the attributes of the smartphone-based survey’s participants

We collected 97 samples out of 13,279 households with zero incentive. Data is free of incentive bias

iPhone Ver. Android Ver. Participants touch the tap when they start and end a trip Travelers’ location is collected every 100 m.

Participants note their travel mode in addition to the start and end of their trip. Location data are collected every 10 second. Acceleration data are also collected.

Results With/Without

invitation letter

With/Without paper/web-

based response Device

Survey Period

Total 1 Oct 10-11

2 Oct 16-18

3 Oct 23-25

Supplementary Nov 20-21

With invitation letter attached

With Android

20 18 8 46

iPhone

6 7 1 14 Subtotal

26 25 9 60

Without Android

7 8 1 16

iPhone

6 5 1 12 Subtotal

13 13 2 28

Subtotal

39 38 11 88

Purely voluntary

participants (without invitation

letter)

With Android 1 2

3

iPhone

1

1 Subtotal 1 3

4

Without Android 3

1

4

iPhone 1

1 Subtotal 4

1

5

Subtotal 5 3 1 9 Total 5 42 39 11 97

1

28 (56.0%)

33 (43.4%)

18,432 (49.5%)

344,291 (46.9%)

22 (44.0%)

43 (56.6%)

18,799 (50.5%)

390,183(53.1%)

0% 20% 40% 60% 80% 100%

Participants in both smartphone-based andpaper/web-based surveys

Members of households that contain thesmartphone-based survey participants who failed to

complete paper/web-based survey

Smartphone-based survey targests (householdmembers)

2010 Population census (Kumamoto City)

Male Female

Conclusion

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

0-9 10-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89 90-99 100-

Prop

ortio

n

Age

2010 Population census (Kumamoto City)

Smartphone-based survey targests (householdmembers)

Members of households that contain thesmartphone-based survey participants whofailed to complete paper/web-based surveyParticipants in both smartphone-based andpaper/web-based surveys

Participation Rate: Male > Female

0

2

4

6

8

10

12

14

16

1 2 3 4 5 6 7 8 9 10+

Num

ber o

f par

ticip

ants

(per

son)

Number of trips reported during paper/web-based survey (trips/day)

Female

Male

Male Female Total Smartphone-based survey

Number of samples 28 22 50 Average trip rate: A 4.14 4.59 4.34

PT survey (Paper/Web-based survey)

Gross trip rate: B 2.98 2.61 2.78 Net trip rate 3.26 3.03 3.14

t-statistics for statistical test (difference between average A & B)

1.44 (No statistical

difference)

4.04*

3.14*

(*: Statistically significant difference at 1% level) 1

2

KUMAMON A local mascot of Kumamoto Prefecture, and now he is the most famous character in JAPAN. We used him in these apps, named “SUMAKUMA”.

Departure & Arrival Tap Departure Tap

Arrival Tap

Map Tap

End Tap

©2010    kumamoto  pref.  kumamon  #3901

ACKNOWLEDGEMENT •  Research collaborator: the Urban Planning Bureau, Kumamoto Prefecture, and the Urban Policy Bureau, Kumamoto City •  Graduate students of Kumamoto University: Mr. Kotaro Nohara, Mr. Keisuke Matsuda, and Mr. Shotaro Imura for their

development of the iPhone-version Apps and the data analysis •  Mr. Yoshihiro Sato, Kumamoto University for this poster design •  Trans Field Inc. for their assistance in the development and preparation of the Android-version App and server management •  Fukuyama Consultants Co., Ltd. for their support on several matters on the survey •  Research fund by “Committee on Advanced Road Technology” in the Road Bureau, Ministry of Land, Infrastructure,

Transport, and Tourism, Japan

•  Initial results of the first smartphone-based travel survey conducted jointly with a large-scale household travel survey in Kumamoto, Japan

•  Created smartphone-based travel surveys apps for Android & iPhone •  13,279 Invitation letters were mailed to households •  Ultimately, 97 joined the smartphone-based survey with zero

incentives •  Males were more likely to join the smartphone-based survey than

females •  Female and 20s were more likely to fail to complete the survey •  People employed in manufacturing, energy, medical and public-

related service occupations showed higher participation rates •  The average trip rate for smartphone-based was higher than that for

traditional household travel survey

Future works: -A discrete choice analysis of smartphone-based survey participation

•  The model could be explained by some demographics described here •  Sampling strategies and removal of sample selection bias

- Increase the sample size •  Our research team conducted another smartphone-based survey in

Kumamoto downtown area with rewards ($5) during Nov. &Dec. 2013 and collected 1,086 samples. Details will be reported soon.

Smartphone-based Travel Survey System

Survey System

Comparison of Travel Survey Method

Purely voluntary participant learned of the survey by reading a newspaper article(Oct. 9, 2012) or accessing the survey committee’s website

Gender

Age

Occupation Response Methods

Trip Rates

Comparison of trip rates among smartphone-based survey and paper/web-based survey participants

Android,  the  Android  logo,  Google  Play™,  the  Google  Play™  logo  are  registered  trademarks  or  trademarks  of  Google,  Inc.  Apple,  the  Apple  logo,  Mac  OS  are  trademarks  of  Apple  Inc.,  registered  in  the  U.S.  and  other  countries.      iPhone,  iPad  are  trademarks  of  Apple  Inc.  In  addiGon,  other  logos  and  names  may  be  trademarks/registered  trademarks  of    other  companies.    

•  Smartphone-based survey participants tend to recall every trip

•  An active person who travels frequently during the day is more likely to join a smartphone-based survey

Trip rate distributions of smartphone-based survey participants.

Possible  Reason

Higher trip rates by smartphone-based survey

ExisGng  study Consistent finding with existing GPS-based studies Bricka et al. (2009), Stopher et al. (2007), Bricka & Bhat (2006), Stopher & Shen (2011),Du & Aultman-Hall (2007)

0% 5% 10% 15% 20%

Agricultural, forestry & fisheriesConstruction

ManufacturingElectricity, gas & water supplyInformation & communication

Transportation & postalWholesale & retail trades

Finance & insuranceReal estate and goods rental & leasing

Professional & technical servicesAccomodations, eating & beverage services

Living-related & personal services and…Education, learning support

Medical, health care & welfareServices, not otherwise classified

GovernmentOthers

Proportion

2010 Population Census(Kumamoto City)

Participants in smartphone-based survey

-  Half of smartphone-based survey participants responded household survey by web

-  Significantly higher than other participants

People in manufacturing industries tend to be more familiar with high-tech devices such as smartphones. People employed by public utilities and government might be expected to participate in this public-services related survey.

High participation Rate: 30s & 40s 20s are relatively low besides their high smartphone ownership

Summary of Smartphone-based Survey Participants

Speculation: Some of female participants became anxious about privacy issues related to the survey and failed to complete the survey

20s participants often fails to complete (=paper-based & smartphone-based) the survey

Tracking & Acceleration Data

Analysis of tracking & acceleration data are ongoing along with auto-mode detection analysis

Kumamoto

Paper GPS + Web Smartphone

Respondent burden Large Small Small Survey operation cost Large Middle (distribution cost

of GPS instruments) Very Small

Initial & other cost Basic and traditional New but available Development of app

Accuracy & reliability of data Low High High

Duration of survey Typically autumn weekday

Multi-days: weekly, monthly, yearly

Multi-days weekly, monthly, yearly

Path/route of traveler Not Available (N.A) Available Available Bus stop, interchange, trip timing information

Low accuracy Low response Precisely Available Precisely available

Trip purpose Written Select (by web diary) Tap on smartphone or auto voice input

Travel mode Written (un-linked-trips are often neglected)

Select or auto detect by acceleration data

Select or auto detect by accelerometer or auto voice input

Selection bias Property of sample

Young people are reluctant to reply Monitored person only

Limited only to owner of smartphone but largely distributed

•  Person Trip (PT) Survey (Household Travel Survey) in Kumamoto Metropolitan Area (pop= 1 million) in Japan, in the Autumn 2012

•  Mail-out/mail-back paper-based self-reporting survey (opt. web-based) •  Final response rate 38.9% Several strategies to increase •  More than 120 thousand household were asked to join the base survey •  Out of them, 13,279 household were asked to join the smartphone-based survey as well

as paper/web-based household survey •  No rewards (incentive) because of governmental constraints

•  Current system just records the track, but modification of record by participants’ devices are now being developed

•  Most participants mention battery life problem

People familiar with the Internet are more likely to join

Approval of Privacy policy & use of terms

The response rate for 20s were lowest for the traditional paper-based survey. If smartphone-only-survey are developed, they will be the potential candidate “SUMA” means “SMArtphone”, and “KUMA” means

“KUMAmon” or “KUMAmoto”

72.7% 80.6% 81.1% 82.7% 89.9% 95.7% 97.9% 97.4% 95.8%

50.9%

27.3% 19.4% 18.9% 17.3% 10.1% 4.3% 2.1% 2.6% 4.2%

49.1%

0%10%20%30%40%50%60%70%80%90%

100%

Web-based

Paper-based

Age of head of household survey participant in total

②① ③ ④ ⑤ ⑥ ⑦ ⑧ ⑨ ⑩ ⑪

Time

Acc

eler

atio

n (m

/s2 )

A Smartphone-based Travel Survey Trial Conducted in Kumamoto, Japan: An Examination of Voluntary Participants’ Attributes

Takuya MARUYAMA*, Shoshi MIZOKAMI*, and Eiji HATO**

(*Kumamoto University; **the University of Tokyo, Japan)

•  Emergence of smartphone-based travel survey as a promising alternative to paper-based or web-based self-reporting survey

•  Challenges in this new survey method should be carefully examined

•  Sampling targets were limited in most of existing smartphone-based survey study

Introduction

Survey Overview & Method

Objective (1) Describe the smartphone-based travel survey trial in Kumamoto, Japan (2) Examine the attributes of the smartphone-based survey’s participants

We collected 97 samples out of 13,279 households with zero incentive. Data is free of incentive bias

iPhone Ver. Android Ver. Participants touch the tap when they start and end a trip Travelers’ location is collected every 100 m.

Participants note their travel mode in addition to the start and end of their trip. Location data are collected every 10 second. Acceleration data are also collected.

Results With/Without

invitation letter

With/Without paper/web-

based response Device

Survey Period

Total 1 Oct 10-11

2 Oct 16-18

3 Oct 23-25

Supplementary Nov 20-21

With invitation letter attached

With Android

20 18 8 46

iPhone

6 7 1 14 Subtotal

26 25 9 60

Without Android

7 8 1 16

iPhone

6 5 1 12 Subtotal

13 13 2 28

Subtotal

39 38 11 88

Purely voluntary

participants (without invitation

letter)

With Android 1 2

3

iPhone

1

1 Subtotal 1 3

4

Without Android 3

1

4

iPhone 1

1 Subtotal 4

1

5

Subtotal 5 3 1 9 Total 5 42 39 11 97

1

28 (56.0%)

33 (43.4%)

18,432 (49.5%)

344,291 (46.9%)

22 (44.0%)

43 (56.6%)

18,799 (50.5%)

390,183(53.1%)

0% 20% 40% 60% 80% 100%

Participants in both smartphone-based andpaper/web-based surveys

Members of households that contain thesmartphone-based survey participants who failed to

complete paper/web-based survey

Smartphone-based survey targests (householdmembers)

2010 Population census (Kumamoto City)

Male Female

Conclusion

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

0-9 10-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89 90-99 100-

Prop

ortio

n

Age

2010 Population census (Kumamoto City)

Smartphone-based survey targests (householdmembers)

Members of households that contain thesmartphone-based survey participants whofailed to complete paper/web-based surveyParticipants in both smartphone-based andpaper/web-based surveys

Participation Rate: Male > Female

0

2

4

6

8

10

12

14

16

1 2 3 4 5 6 7 8 9 10+

Num

ber o

f par

ticip

ants

(per

son)

Number of trips reported during paper/web-based survey (trips/day)

Female

Male

Male Female Total Smartphone-based survey

Number of samples 28 22 50 Average trip rate: A 4.14 4.59 4.34

PT survey (Paper/Web-based survey)

Gross trip rate: B 2.98 2.61 2.78 Net trip rate 3.26 3.03 3.14

t-statistics for statistical test (difference between average A & B)

1.44 (No statistical

difference)

4.04*

3.14*

(*: Statistically significant difference at 1% level) 1

2

KUMAMON A local mascot of Kumamoto Prefecture, and now he is the most famous character in JAPAN. We used him in these apps, named “SUMAKUMA”.

Departure & Arrival Tap Departure Tap

Arrival Tap

Map Tap

End Tap

©2010    kumamoto  pref.  kumamon  #3901

ACKNOWLEDGEMENT •  Research collaborator: the Urban Planning Bureau, Kumamoto Prefecture, and the Urban Policy Bureau, Kumamoto City •  Graduate students of Kumamoto University: Mr. Kotaro Nohara, Mr. Keisuke Matsuda, and Mr. Shotaro Imura for their

development of the iPhone-version Apps and the data analysis •  Mr. Yoshihiro Sato, Kumamoto University for this poster design •  Trans Field Inc. for their assistance in the development and preparation of the Android-version App and server management •  Fukuyama Consultants Co., Ltd. for their support on several matters on the survey •  Research fund by “Committee on Advanced Road Technology” in the Road Bureau, Ministry of Land, Infrastructure,

Transport, and Tourism, Japan

•  Initial results of the first smartphone-based travel survey conducted jointly with a large-scale household travel survey in Kumamoto, Japan

•  Created smartphone-based travel surveys apps for Android & iPhone •  13,279 Invitation letters were mailed to households •  Ultimately, 97 joined the smartphone-based survey with zero

incentives •  Males were more likely to join the smartphone-based survey than

females •  Female and 20s were more likely to fail to complete the survey •  People employed in manufacturing, energy, medical and public-

related service occupations showed higher participation rates •  The average trip rate for smartphone-based was higher than that for

traditional household travel survey

Future works: -A discrete choice analysis of smartphone-based survey participation

•  The model could be explained by some demographics described here •  Sampling strategies and removal of sample selection bias

- Increase the sample size •  Our research team conducted another smartphone-based survey in

Kumamoto downtown area with rewards ($5) during Nov. &Dec. 2013 and collected 1,086 samples. Details will be reported soon.

Smartphone-based Travel Survey System

Survey System

Comparison of Travel Survey Method

Purely voluntary participant learned of the survey by reading a newspaper article(Oct. 9, 2012) or accessing the survey committee’s website

Gender

Age

Occupation Response Methods

Trip Rates

Comparison of trip rates among smartphone-based survey and paper/web-based survey participants

Android,  the  Android  logo,  Google  Play™,  the  Google  Play™  logo  are  registered  trademarks  or  trademarks  of  Google,  Inc.  Apple,  the  Apple  logo,  Mac  OS  are  trademarks  of  Apple  Inc.,  registered  in  the  U.S.  and  other  countries.      iPhone,  iPad  are  trademarks  of  Apple  Inc.  In  addiGon,  other  logos  and  names  may  be  trademarks/registered  trademarks  of    other  companies.    

•  Smartphone-based survey participants tend to recall every trip

•  An active person who travels frequently during the day is more likely to join a smartphone-based survey

Trip rate distributions of smartphone-based survey participants.

Possible  Reason

Higher trip rates by smartphone-based survey

ExisGng  study Consistent finding with existing GPS-based studies Bricka et al. (2009), Stopher et al. (2007), Bricka & Bhat (2006), Stopher & Shen (2011),Du & Aultman-Hall (2007)

0% 5% 10% 15% 20%

Agricultural, forestry & fisheriesConstruction

ManufacturingElectricity, gas & water supplyInformation & communication

Transportation & postalWholesale & retail trades

Finance & insuranceReal estate and goods rental & leasing

Professional & technical servicesAccomodations, eating & beverage services

Living-related & personal services and…Education, learning support

Medical, health care & welfareServices, not otherwise classified

GovernmentOthers

Proportion

2010 Population Census(Kumamoto City)

Participants in smartphone-based survey

-  Half of smartphone-based survey participants responded household survey by web

-  Significantly higher than other participants

People in manufacturing industries tend to be more familiar with high-tech devices such as smartphones. People employed by public utilities and government might be expected to participate in this public-services related survey.

High participation Rate: 30s & 40s 20s are relatively low besides their high smartphone ownership

Summary of Smartphone-based Survey Participants

Speculation: Some of female participants became anxious about privacy issues related to the survey and failed to complete the survey

20s participants often fails to complete (=paper-based & smartphone-based) the survey

Tracking & Acceleration Data

Analysis of tracking & acceleration data are ongoing along with auto-mode detection analysis

Kumamoto

Paper GPS + Web Smartphone

Respondent burden Large Small Small Survey operation cost Large Middle (distribution cost

of GPS instruments) Very Small

Initial & other cost Basic and traditional New but available Development of app

Accuracy & reliability of data Low High High

Duration of survey Typically autumn weekday

Multi-days: weekly, monthly, yearly

Multi-days weekly, monthly, yearly

Path/route of traveler Not Available (N.A) Available Available Bus stop, interchange, trip timing information

Low accuracy Low response Precisely Available Precisely available

Trip purpose Written Select (by web diary) Tap on smartphone or auto voice input

Travel mode Written (un-linked-trips are often neglected)

Select or auto detect by acceleration data

Select or auto detect by accelerometer or auto voice input

Selection bias Property of sample

Young people are reluctant to reply Monitored person only

Limited only to owner of smartphone but largely distributed

•  Person Trip (PT) Survey (Household Travel Survey) in Kumamoto Metropolitan Area (pop= 1 million) in Japan, in the Autumn 2012

•  Mail-out/mail-back paper-based self-reporting survey (opt. web-based) •  Final response rate 38.9% Several strategies to increase •  More than 120 thousand household were asked to join the base survey •  Out of them, 13,279 household were asked to join the smartphone-based survey as well

as paper/web-based household survey •  No rewards (incentive) because of governmental constraints

•  Current system just records the track, but modification of record by participants’ devices are now being developed

•  Most participants mention battery life problem

People familiar with the Internet are more likely to join

Approval of Privacy policy & use of terms

The response rate for 20s were lowest for the traditional paper-based survey. If smartphone-only-survey are developed, they will be the potential candidate “SUMA” means “SMArtphone”, and “KUMA” means

“KUMAmon” or “KUMAmoto”

72.7% 80.6% 81.1% 82.7% 89.9% 95.7% 97.9% 97.4% 95.8%

50.9%

27.3% 19.4% 18.9% 17.3% 10.1% 4.3% 2.1% 2.6% 4.2%

49.1%

0%10%20%30%40%50%60%70%80%90%

100%

Web-based

Paper-based

Age of head of household survey participant in total

②① ③ ④ ⑤ ⑥ ⑦ ⑧ ⑨ ⑩ ⑪

Time

Acc

eler

atio

n (m

/s2 )

A Smartphone-based Travel Survey Trial Conducted in Kumamoto, Japan: An Examination of Voluntary Participants’ Attributes

Takuya MARUYAMA*, Shoshi MIZOKAMI*, and Eiji HATO**

(*Kumamoto University; **the University of Tokyo, Japan)