190 likes | 325 Views
Modal split Hana Brůhová-Foltýnová Kolin Institute of Technology. VOCA, Prague, November 16, 2012. Structure of the presentation. What is modal split What we must be careful about regarding modal split Publically accessible databases of modal split Research using modal split
E N D
Modal split Hana Brůhová-Foltýnová Kolin Institute of Technology VOCA, Prague, November 16, 2012
Structure of the presentation • What is modal split • What we must be careful about regarding modal split • Publically accessible databases of modal split • Research using modal split • Factors influencing modal split (literature overview) • Statistical analysis (own research) VOCA, Prague, November 16, 2012
What is modal split • Modal split (mode share, mode split) = the relative share of each mode of transport • Calculated for personal and freight transport • based on passenger-kilometres (p-km) for passenger transport and tonne-kilometres (t-km) for freight or goods transport or on the number of trips (typically in urban areas to „balance“ non-motorized traffic) • Usually defined for a specific geographic area and/or time period • A useful indicator and a basis for direct demand models (started 1960s) VOCA, Prague, November 16, 2012
What we must be careful about • Comparability of data • What is the unit/bases of the modal split • Unit (number of trips x pkm/vehicle-km x number of trip makers) • What trips are included (all trips x commuting) • Geographical area (the whole city x an intersection) • How the data are collected (census x own survey x estimated using models and older data) • Whose trips are included (the whole population x adult population) • Quality of surveys (data collection) - sample, questions (usual trips x previous day etc.) VOCA, Prague, November 16, 2012
Sources of data on modal split • EPOMM http://www.epomm.eu/tems/index.phtml • created with the support of Intelligent Energy Europe in the project EPOMM-PLUS • Each city across Europe can upload their own data and control the data themselves • cities in the EU with more than 100.000 inhabitants • Urban Audit database • statistics for 258 citiesacross 27 European countries. • almost 300 statistical indicators ondemography, society, the economy, the environment, transport, the information society, public health, and leisure • Wikipedia • Annual reports / special studies of cities VOCA, Prague, November 16, 2012
Modal split - summary • An easy understandable and often used indicator • Easy to compare cities / regions / countries / continents …. • Good for observing trends • Mode choice is a key concept of some traffic models • BUT • Comparability of data – what is the unit/basis and how data were collected and modal split calculated • Reliability - existence of different data about modal split from different sources VOCA, Prague, November 16, 2012
Factors influencing the choice of mode • Characteristics of the trips maker • Car availability/ownership • Possession of a driving licence • Household structure • Income • Decisions made elsewhere (need to use a car at work, take children to school, etc.) • Residential density • 2. Characteristics of the journey • The trip purpose • Time of the day • 3. Characteristics of the transport facility VOCA, Prague, November 16, 2012
Factors influencing the choice of mode • 3. Characteristics of the transport facility • Quantitative • Relative travel time • Relative monetary costs • Availability and cost of parking • Qualitative • Comfort and convenience • Reliability and regularity • Protection, security VOCA, Prague, November 16, 2012
Factors influencing cycling • Many studies identified several types of factors, such as: • physical aspects - hilliness, the size of the city • demographic factors (age, gender, ethnicity, education levels, ....) • individual attitudes and ecological beliefs (all at family / local community and work place) • perceived cycling safety • Transportation policy, such as: • discouraging usage of cars • existence and quality of cycling infrastructure VOCA, Prague, November 16, 2012
Statistical analysis: Data used • City-level data: • EPOMM data on modal split and data from open public sources on cycling infrastructure; N=35 • Data consistency among cities? • 2) Urban Audit data (2 waves: 1999-2002 and 2003-2006); N=69, resp. 59 • (panel data only 37) • Similar results by using the both data sets; in the paper, the UA is used (bigger sample and a higher consistency among data)
Statistical analysis: Methodology • Model: the logistic regression (aka multinomial logit model) • Where: stj is the share of mode j in city t;Xtare city level explanatory variables (mainly the cycling infrastructure); and ϒ are parameters to be estimated; • ∑jstj = 1, i.e., the shares of all modes sum to 100% • The logistic regression enables not only model the impact of cycling infrastructure on the cycling share, but also to identify from which mode people switch towards cycling • The main explanatory variable = the ratio of the length of cycling infrastructure to population • We consider 4 transport modes: cycling, walking, public transport, cars
Statistical analysis: Results • (1) the length of the cycling infrastructure is significantly and positively related to the cycling share • (2) the shares of cars and walking do not depend on the length of cycling infrastructure for the first wave of data collection, while the second wave data suggest that cycling and walking are substitutes, while cycling and car usage are complements • (3) the share of public transport significantly declines with the length of the cycling infrastructure at least for the data from the second wave of collection
Discussion • Similar results obtained with the two data sets • A novel approach – all transport modes are included (not only cycling) • Nelson and Allen (1997): 1 additional mile of cycling infrastructure per 100,000 residents = 0.069% increase in commuters cycling to work; our results: 0.144% to 0.269% increase in the share of cycling • Limitations: • correlation does not necessarily imply causation (the good infrastructure may be in those cities, where people like to bike, rather than vice versa); • cycling policies tend to be implemented as packages, therefore our results might identify the effect of a typical policy mix rather than the effect of just cycling infrastructure • if we want to address these limitations, we would need very detailed panel data
Thank you for your attention Contact: Ing. Mgr. Hana Brůhová-Foltýnová, PhD. Kolínský technologický institut, o.s. E-mail: bruhova@koltech.cz Tel.: 736 43 43 47 VOCA, Prague, November 16, 2012