Answer Young People Just Like Old People? Impacts and Consequences of Different Household Travel Survey Methods

Transcription

1 Answer Young People Just Like Old People? Impacts and Consequences of Different Household Travel Survey Methods Dr.-Ing. Martin Kagerbauer 7 March 2013 New Techniques and Technologies for Statistics (NTTS), Session 17C: Special session on COST/Shanti, Brussels INSTITUTE FOR TRANSPORT STUDIES KIT University of the State of Baden-Wuerttemberg and National Research Center of the Helmholtz Association

2 Agenda Background of the survey Different methods of the survey Comparison of the different survey methods Response rates, sample sizes, types of households, car-ownership, age, profession, season tickets for public transport Mobility key figures, modal split and trip purpose Conclusion and Outlook 2 Kagerbauer

3 Background of the survey Region of Stuttgart Conurbation of 179 cities and boroughs A total of about 2.7 million inhabitants Household Travel Survey 5,500 households Multi-day survey Duration one week Objectives of the Survey Update the mobility database Input for a transport demand model (microscopic and macroscopic) Information about variability and stability of travel behaviour 3 Kagerbauer

4 Background of the survey Data collected during the survey Socio-demographic data concerning the household and all persons living in the household Trip data with location of origin and destination (transferred into geocodes) Survey is based on German Mobility Panel (MOP) and combines other survey designs Quality assurance Sample size Efficiency Check against MOP Regional Data MiD,... Verification Sample Main Sample Methode for large sample size 4 Kagerbauer

5 Background of the survey Survey methods Verification Sample approx. 1,000 HH Sample size Main Sample approx. 4,500 HH PAPI method (paper-pencil-interview) following the design of the German Mobility Panel CATI-method (computer-assistedtelephone-interview) CAWI-method (computer-assistedweb-interview) A lot of questions Only questions required for modelling 5 Kagerbauer

6 Different methods of the survey PAPI method (Paper-Pencil-interview) Household and person questionnaire (e.g. number of persons/children living in the household, number of cars, information about the cars, parking situation, public transport situation, sex, age, profession, car and public transport availability etc.) Trip diary (start and end time of trip, modes used, trip purpose, duration, trip length) Questionnaire is sent back by mail and plausibility checked Design based on German Mobility Panel (MOP) Allows benchmarking without methodological differences Time-tested method and common practice in Germany The PAPI-sample serves as verification sample in the comparison of the different survey methods 6 Kagerbauer

7 Different methods of the survey CATI method (computer-assisted-telephone-interview) Participants are asked to report their trips by phone All participants are called two or three times during the survey period and asked to report their trips Trips are entered into the computer directly The data is immediately plausibility checked In case of mistakes in the data, interviewer can ask participants immediately to clarify Disadvantage: Participants can forget some trips in the haste of a telephone call Innovative method in multi-day surveys but common practice in cross-sectional surveys in Germany 7 Kagerbauer

8 Different methods of the survey CAWI method (computer-assisted-web-interview) Participants report their trips via internet in a software tool Data is checked one day at a time In case of mistakes, participants are contacted and asked to give missing information Participants can take part in the survey at any time they want Disadvantage: Participants do not answer very thoroughly Innovative method in multi-day surveys tested in cross-sectional surveys in Germany 8 Kagerbauer

9 Different methods of the survey Survey procedure Total sample size (5,500 Households) Random division of the total sample to two subsamples Households are randomly drawn (registration offices) Only mailing address provided Merge telephone numbers, if possible Verification Sample (1,000 Households) Main Sample (4,500 Households) Participant can choose between CATI and CAWI-methode Contacting the participants CATI-Sample CAWI-Sample First contact by mail, introducing the survey and asking to participate When phone number available: another contact by phone (this generally leads to higher response rates than only mail) Questionnaires Memory jogger and calls Web access 9 Kagerbauer

10 Comparison of the different survey methods Response rates Response rates Ratio of households successfully surveyed [%] (=useful questionnaires / sample size) Ratio of households willing to participate [%] (=households willing to participate / sample size) Ratio of households successfully surveyed [%] (=useful questionnaires / households willing to participate ) Stuttgart Region Survey 2009 CATI CAWI CATI+CAWI PAPI Not applicable due to the survey method Not applicable due to the survey method 8.1% 8.8% 16.7% 15.1% 64.1% 39.7% 51.8% 63.0% CATI+CAWI very similar to PAPI In CATI more of the households willing to participate actually take part in the survey then in CAWI Participants in CATI are more reliable 10 Kagerbauer

11 Comparison of the different survey methods Types of households Stuttgart Region Survey 2009 Types of households ALL CATI CAWI CATI+CAWI PAPI (CATI+CAWI+PAPI) Sample size [housholds] 2,215 2,250 4,465 1,096 5,561 Single household 7.1% 12.2% 9.7% 8.0% 9.4% Couple household without children 11.2% 22.0% 16.7% 20.0% 17.3% Family or single parent household with children < 19 years 19.6% 33.9% 26.8% 25.5% 26.5% Senior household 49.7% 17.5% 33.5% 28.2% 32.4% Multi-person-household without children < 19 years 12.5% 14.4% 13.4% 16.8% 14.1% No classification 0.0% 0.0% 0.0% 1.6% 0.3% Overall comparison between PAPI and CATI+CAWI is similar Share of senior households high in CATI and very low in CAWI. Share of single, couple and single parents households high in CAWI and very low in CATI Seniors are available by telephone and not so familiar with technology Single persons and couples without children are hard to reach by phone but more familiar with internet 11 Kagerbauer

12 Comparison of the different survey methods Age of the persons Age Stuttgart Region Survey 2009 CATI CAWI CATI+CAWI PAPI ALL (CATI+CAWI+PAPI) CATI: High share of elderly and smaller share of younger people (19 to 34 years) CAWI: Smaller share of elderly and high share of teens and middle aged people PAPI: Gap between people with 19 to 34 years (less) and 35 to 64 years (more) Results for CATI+CAWI compared with PAPI method are similar Stuttgart Region population parameter Sample size [persons] 5,166 6,149 11,315 2,852 14,167 2,679,000 0 to 18 years 20.3% 27.2% 24.0% 23.6% 23.9% 18.8% 19 to 34 years 8.7% 15.6% 12.5% 11.3% 12.2% 19.6% 35 to 64 years 38.3% 46.5% 42.7% 46.7% 43.5% 41.9% 65 years and older 32.7% 10.7% 20.8% 17.5% 20.1% 19.7% Not specified 0.0% 0.0% 0.0% 1.0% 0.2% 0.0% It is difficult to reach elderly people with the CAWI method and similarly difficult to reach the age group between19-34 for CATI 12 Kagerbauer

13 Comparison of the different survey methods Profession of the persons Stuttgart Region Survey 2009 Profession ALL CATI CAWI CATI+CAWI PAPI (CATI+CAWI+PAPI) Sample size [persons] 5,166 6,149 11,315 2,852 14,167 Fully working 22.2% 32.8% 28.0% 29.2% 28.2% Partly working 10.8% 13.6% 12.3% 14.1% 12.7% Not working/ jobless 6.2% 7.7% 7.0% 7.6% 7.1% In education 18.8% 23.9% 21.6% 23.9% 22.1% Retired person 35.9% 11.9% 22.9% 19.5% 22.2% Not specified 6.1% 10.1% 8.3% 5.8% 7.8% CATI: High ratio of retired persons and low ratio of fully working people CAWI: Low ratio of retired persons and high ratio of fully working people PAPI: Similar to CATI+CAWI Only CAWI and CATI represent all professions primarily the fully working and the retired people 13 Kagerbauer

14 Comparison of the different survey methods Mobility key figures Stuttgart Region Survey 2009 Mobility key figures ALL CATI CAWI CATI+CAWI PAPI (CATI+CAWI+PAPI) Sample size [person-days] 34,496 39,585 74,081 16,653 90,734 Trip makers [%] 84% 87% 86% 90% 87% Trips per person and day [no] Kilometres per person and day [km] Minutes per person and day [min] Average speed [km/h] Average trip length Lower share of trip makers and lower amount of kilometers per day in CATI due to higher ratio of elderly persons No significant differences between the methods for the trips per person Number of kilometers per day and the time spent for transport are higher in CAWI than in CATI (as a follow: different levels of speed and trip length) T-test: Kilometers and travel time differ significantly between the methods 14 Kagerbauer

15 Comparison of the different survey methods Modal Split Stuttgart Region Survey 2009 Modal split ALL CATI CAWI CATI+CAWI PAPI (CATI+CAWI+PAPI) Sample size [person-days] 34,496 39,585 74,081 16,653 90,734 Foot 27.6% 22.9% 25.0% 22.3% 24.5% Bike 4.3% 5.5% 4.9% 4.6% 4.9% Motorcycle and car as driver 42.6% 44.1% 43.5% 47.2% 44.1% Car as passenger 13.8% 13.4% 13.6% 13.2% 13.5% Public Transport (Bus, rapid transit) 10.5% 12.3% 11.4% 10.6% 11.3% Railway 1.1% 1.6% 1.4% 1.8% 1.5% Others 0.1% 0.2% 0.2% 0.3% 0.2% Similar results for all survey methods Small difference for foot : Share is higher for CATI method Methodical effect : In a telephone interview, the interviewer can immediately ask about forgotten trips, especially short trips made by foot 15 Kagerbauer

16 Comparison of the different survey methods Trip Purpose Stuttgart Region Survey 2009 Trip purpose ALL CATI CAWI CATI+CAWI PAPI (CATI+CAWI+PAPI) Sample size [person-days] 34,496 39,585 74,081 16,653 90,734 Work 7.7% 11.6% 9.8% 10.3% 9.9% Buisiness trips 1.2% 2.1% 1.7% 2.4% 1.8% Education 3.7% 5.1% 4.5% 3.9% 4.4% Leisure 18.5% 16.1% 17.2% 15.8% 16.9% Shopping 11.6% 9.5% 10.4% 9.9% 10.3% Visitation 12.5% 9.9% 11.1% 11.8% 11.2% Service 4.3% 4.8% 4.6% 4.6% 4.6% Others 0.2% 0.5% 0.4% 0.8% 0.5% Back home 40.3% 40.4% 40.4% 40.5% 40.4% CATI: Share of work and business is lower due to high share of retiree CAWI: Share of education is high due to high ratio of young persons Combination of CATI+CATI show similar results as PAPI method 16 Kagerbauer

17 Conclusions and Outlook PAPI method Works well and provides reliable results without unexpected non-response effects Significant differences between CATI and CAWI methods Younger people have strong preference to join CAWI survey Elderly or retired people clearly favor telephone interview CATI + CAWI method Combination of CATI and CAWI method provides similar results to PAPI method As a result, most key figures of household and person attributes as well as mobility behavior are affected for both CATI and CAWI method 17 Kagerbauer

18 Conclusions and Outlook CAWI method alone leads to massive overrepresentation of younger people CATI method alone leads to overrepresentation of elderly people Combination of CAWI method with more approved methods (especially CATI) is recommended But: Combination of different survey methods leads to higher costs and complexity of survey 18 Kagerbauer

19 Thank you for your attention! Answer Young People Just Like Old People? Impacts and Consequences of Different Household Travel Survey Methods Dr.-Ing. Martin Kagerbauer, Karlsruhe Institute of Technology (KIT), ( ) +49 (721) (phone) INSTITUTE FOR TRANSPORT STUDIES KIT University of the State of Baden-Wuerttemberg and National Research Center of the Helmholtz Association