Journal of Transport Geography 18 (2010) 75–84

Contents lists available at ScienceDirect

Journal of Transport Geography

journal homepage: www.elsevier.com/locate/jtrangeo

Interrelations between travel mode choice and trip distance: trends

in Germany 1976–2002
Joachim Scheiner *
TU Dortmund University of Technology, Faculty of Spatial Planning, Department of Transport Planning, 44227 Dortmund, Germany

a r t i c l e i n f o a b s t r a c t

Keywords: In recent decades, trends in travel behaviour have been characterised by increasing trip distances and a
Trip distance modal shift towards the private car. This paper reports findings from longitudinal analyses of the German
Travel mode choice nation-wide travel survey KONTIV for the period 1976–2002. It focuses on travel mode choice, subdi-
Motorisation vided by distance categories, and also takes car availability and city size into account. In addition, trends
Longitudinal analysis
in car availability itself are examined by city size categories. The results indicate that even within the
same distance categories car use has considerably increased. In some cases bicycle use has increased
as well. Gains in the use of the private car are mainly at the expense of trips on foot and by public trans-
port. Accordingly, the shift in modal split towards the car is not (only) caused by increasing trip distances
but took place even within distance classes. Once car availability is taken into account, the modal shifts
appear to be considerably weaker. This suggests that once car availability is held constant the decision
rationales of mode choice for a certain trip distance have remained relatively stable. The increase in
motorisation over the study period was considerably weaker in large cities than in small towns, although
the cities started from a lower level in the 1970s. Thus, the motorisation divide between cities on the one
hand, and suburban and rural areas on the other hand has become ever wider. For travel mode choice, the
picture is similar. What is more, the results suggest that even car owners are more inclined to walk a
given distance in the cities than in small towns, even more so if they live in a central urban area. The built
environment, thus, appears to have a strong impact on whether an available car is used or not.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction environment on travel mode choice is thus in this case indirect:

travel mode choice depends on trip distance, and trip distance de-
Time and again, transport researchers have highlighted a num- pends on access to facilities within a certain distance radius (i.e. on
ber of key attributes of the built environment likely to affect travel the least required distance to a certain activity). These interrela-
behaviour: density, land-use, and distance to the nearest centre tions have been found in numerous empirical studies (see for over-
(see overviews in Stead and Marshall, 2001; Cervero, 2006). What views and empirical studies Cervero, 2002; Schwanen et al., 2004;
is more, the connectivity of street networks has frequently been Guo and Chen, 2007 and other papers in the same issue).
seen as important, particularly in North America and Australia However, there is also a direct interrelation between the built
due to the widespread cul-de-sac structures in the suburbs (e.g. environment and travel mode choice. As well as NMT, public trans-
Crane, 2000; Badland and Schofield, 2005). The overall spatial port (PT) also accounts for a relatively high share of trips in dense,
determinants of travel have been summarised using keywords mixed-use structures for two reasons. Firstly, high population and/
such as density, diversity, design (Cervero, 2002). or activity density often goes along with restrictions for car use,
The built environment has a particularly strong impact on two such as lack of parking space, high traffic density, and low travel
elements of travel-demand: trip distances and travel mode choice. speed, and therefore reduces the comparative disadvantage of PT
A dense, mixed-use urban structure allows the population to make against the private vehicle (in some cases even producing the
short trips due to the proximity of housing to other activities. As a opposite effect). Secondly, the high demand density encourages
consequence of the short distances, a comparatively large propor- an attractive PT system. Nonetheless density as such has little influ-
tion of the trips may be undertaken by non-motorised transport ence on travel mode choice, particularly on NMT (Filion et al.,
modes (NMT, i.e. by bicycle or on foot). The impact of the built 2006; Forsyth et al., 2007), when not combined with mixed land-
use.
* Tel.: +49 231 755 4822; fax: +49 231 755 2269.
The interrelations between the built environment and trip dis-
E-mail address: [email protected] tances as well as between the built environment and travel mode

0966-6923/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jtrangeo.2009.01.001
76 J. Scheiner / Journal of Transport Geography 18 (2010) 75–84

choice have been investigated in detail in numerous studies. How- However, such trends leave a number of the interrelationships
ever, this is not true for the interrelations between trip distances between various aspects of travel behaviour open. For instance, re-
and mode choice. The lack of research is particularly striking with search has found that long-distances in peripheral locations tend
respect to long-term analysis that could allow conclusions on the to be at least partly compensated by lower trip frequencies (see
stability or instability of travel behaviour to be drawn. This is the Holz-Rau et al., 1999 for shopping trips), although some findings
focus of this paper. The following section provides a brief literature seem to contradict this (Naess, 2006). Furthermore, it is as yet un-
overview, followed by some hypotheses. These hypotheses are clear whether travel mode changes are actually a consequence of
examined for the period 1976–2002 on the basis of descriptive increasing distances. Based on a study of commuting in the region
analyses of the German national travel survey KONTIV. Section 3 of Bremen, Bahrenberg (1997) argues that car use has increased
introduces the data and methodology, and Section 4 presents the over the period 1970–1987 even within distance categories, rather
results. The paper concludes with an outlook. than as a consequence of distance increases. He concludes that the
shift towards the car was not caused by spatial changes but is
2. Trip distances, travel mode choice and the built environment: rather an expression of independent individual decisions.
the state of the research The interrelations between travel mode choice and trip dis-
tances form the focus of the following two sections. A distinction
2.1. General trends in travel behaviour is made between least required distance to the nearest facility
(‘nearest centre’) on the one hand, and realised distances on the
General trends in travel behaviour in recent decades may be other hand.
roughly summarised in terms of five key characteristics.
First, per capita trip frequencies have remained more or less 2.2. Travel mode choice and distance to the nearest facility
constant over time. Although in Germany the survey ‘Mobility in
Germany’ 2002 showed considerably higher trip figures than the In contrast to realised trip distances, minimum necessary dis-
previous KONTIV surveys (Kunert and Follmer, 2005; Holz-Rau tances to nearest facilities such as shopping centres, leisure oppor-
and Scheiner, 2006), this is a consequence of changes in the survey tunities or schools can be directly affected by spatial planning
methodology. concepts. There is likely to be a knock-on effect for realised trip dis-
Second, people’s travel time expenditure has generally been tances and travel mode choice. However, there is no common
recognised to be relatively invariant over time and space, even if understanding of the details of these interrelations, e.g. of distance
this is likely to be true only on an aggregate level (Mokhtarian thresholds.
and Chen, 2004; in the Dutch context a long-term increase has One reason for this may be that many studies work with rather
been detected by Van Wee et al., 2006). generalised distance categories. For instance, the German state
Third, the second half of the 20th century was characterised by North Rhine–Westphalia implemented a programme intended to
the triumphing of the private car (Fig. 1). Mass motorisation led to promote rail-oriented housing schemes. This programme was lar-
a considerable shift in travel mode choice towards the car at the gely based on the observation that car use among those living in
expense of NMT and, to a certain extent, also of PT. Today about a catchment area of one kilometre around a railway station was
75% of all travel distances in Germany are covered by car (BMVBS, slightly lower than outside this radius (ILS, 1999). However, the
2007), not unlike the situation in many other European countries. possible existence of a decline in car use within the 1 km-radius
What is more, travel by aeroplane has increased markedly, albeit has not been explored.
from a low level. Holz-Rau (1991) examined shopping trips in a Berlin neigh-
Fourth, as a consequence of the shift from slow to faster modes bourhood on a micro-spatial basis. His results show that car use
(particularly the car) achieved travel speeds have considerably for shopping increases rapidly among motorised households from
increased. a distance of 325 m or more to the nearest grocery store. For dis-
Fifth, in conjunction with faster modes and constant travel time tances exceeding 670 m motorised households hardly use any
expenditures people have enormously expanded their activity transport modes other than the car. Individuals without access to
spaces. Accordingly, per capita travel distances have increased. a car tend to switch to the bicycle when the distance exceeds
Due to the more or less constant trip frequencies this is also true 325 m.
for the distances per trip. The daily life-worlds of many individuals Holz-Rau et al. (1999) found considerably longer, but less fre-
have shifted from the local to the regional level. quent shopping trips in poorly served (mono-functional) neigh-

45
40
km / inhabitant*day

aeroplane
35
car /
30
motorcycle
train
25
local PT
20
bicycle
15
10 on foot

5
0
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

Fig. 1. Travel distance per capita and day in Germany by travel mode, 1950–2005. The first vertical line indicates German reunification, the second line a change in the model
estimation methodology in 1994. Source: author’s analysis based on BMVBS (2007), data before 1976 estimated using Apel (1995) as a basis.
 J. Scheiner / Journal of Transport Geography 18 (2010) 75–84 77

bourhoods, as compared to mixed-use neighbourhoods with good Mackett (2003) examines reasons for using the private car for
shopping facilities. The proportion of car trips among all shopping short trips up to one mile in length and finds the most frequently
trips is markedly higher in mono-functional neighbourhoods, cited reasons to be: goods transport (22%), escort (17%), too far
although in absolute terms the differences are less pronounced (11%), and convenience (10%). The latter reason was given more of-
due to the lower trip frequency. Owing to the longer distances in ten than average in rural areas and small towns. From this the
the mono-functional neighbourhoods, the car is in any case used hypothesis may be derived that short to medium distances are
much more for shopping. more frequently covered on foot in cities than in towns or in the
Results from existing studies are not consistent. For the US, countryside. This may be due to the lower motorisation level, the
Handy and Clifton (2001) found that proximity to shopping facili- higher generalised costs of car use (lack of parking space at the ori-
ties did not seem to reduce car travel. Yet, in another study dis- gin and/or destination, lower travel speed), or the more exciting
tance to the nearest facility turned out to be an important environment in cities.
impact factor influencing the frequency of shopping trips on foot To the best of my knowledge, there are hardly any spatially dif-
(Cao et al., 2006). ferentiated studies on this topic concerning trends over time. The
Not all trip purposes are equally suited for use in such studies, results found by Scheiner (2006b) point towards an increasing di-
because of the lack of clarity for some types of activity about the vide in travel mode choice between cities and smaller communi-
extent of appropriateness of opportunities for certain require- ties: in cities, car use increases more slowly and from a lower
ments. It is usual to assume that a shopping centre or a grocery level than in smaller communities.
store suits the needs of all. However, the same assumption may Recent debate about residential self-selection suggests that the
not be valid for specialised retail branches, and certainly cannot conditions of the built environment, which are reflected in dis-
be made with relation to the appropriateness of the nearest work- tances to certain destinations should not be regarded as a fixed
place or of a particular location for a stroll. pre-condition of life. Cao et al. (2006) show that short distances
to the next shop markedly increase the frequency of walking for
2.3. Travel mode choice and realised distance shopping purposes. At the same time, however, the subjective
importance individuals assign to the accessibility of shopping facil-
Transport planning commonly assumes that walking is the fast- ities increases the likelihood of walking as well, suggesting that
est travel mode for distances under one kilometre, the bicycle for people sort themselves into the environments they prefer. For
distances between one and 6 km, and the car for distances over shopping trips the effect of self-selection is stronger than for stroll-
6 km; access and egress time are taken into consideration (Zumkel- ing. Scheiner and Holz-Rau (2007) carried out a similar study on
ler and Nakott, 1988). the basis of structural equation modelling. The results confirm
The effect of realised trip distances on mode choice is, if at all, the important effect of residential self-selection on travel mode
mostly examined in rather coarsely meshed distance categories. choice, even when objective attributes of the neighbourhood are
The lowest category typically includes trips of up to about one to being controlled for.
2 km, or one mile (Kloas and Kunert, 1993; Bahrenberg, 1997; Sch- The decision in favour of a certain travel mode obviously does
lossberg et al., 2006; DfT, 2006). Working with these relatively not only depend on trip distance. Important factors affecting the
broad categories may mask considerable variation within the propensity to walk are individual motivation, available resources
categories and (in longitudinal studies) important shifts over time (transport means, financial resources, health), the attractiveness
(Scheiner, 2002). An exception is Vågane (2007) who provides of the route, and social roles and needs, which are reflected in
differentiated analyses for Norway. employment, gender and age, among other variables. For instance,
Distance thresholds vary between regions and countries adolescents cover considerable distances to school on foot (Sch-
(Badland and Schofield, 2005, p. 186). This may have cultural, eco- lossberg et al., 2006). The socio-demographic factors may be sum-
nomic, topographic or climatic reasons. In many highly motorised marised under the term life situation (Scheiner and Holz-Rau,
countries the car is the preferred mode of transport even for short 2007).1 The availability of a car plays a central role in this respect,
trips, while in developing countries very long-distances are because the speed that can be achieved by car is a pre-condition
covered on foot. Among the more developed countries there are for long trips under a restricted time budget. Car availability conse-
considerable differences as well. In the UK 76 percent of trips up quently turns out to be an important pre-decision for an individual’s
to one mile (1.6 km) are undertaken on foot (DfT, 2006, pp. 15– travel behaviour, which has a notable impact on his or her activity
16). In Germany the equivalent figure is only 60% for trips up to spaces, trip distances and travel mode choice (Simma and Axhausen,
1 km in length (calculated from DIW/INFAS, 2003), and in Norway 2001; Scheiner and Holz-Rau, 2007).
it is 53% (Vågane, 2007). Besides substantial differences, methodo- In addition, trip purpose has an important influence on the pro-
logical differences may also play a role here. pensity to walk. This is self-evident with respect to trips with an
There are also marked international differences in bicycle use. intrinsic motivation to walk, such as strolling or hiking. The
While the proportion of short trips undertaken on foot is lower requirement of carrying (shopping) goods impacts or even deter-
in Germany than in the UK, this is largely compensated by the bicy- mines the chance of walking. In addition, the high economic cost
cle. Generally, the bicycle is widely used in the Netherlands and in of travel time (job trips, business trips) may limit the acceptability
Denmark, in Germany less so, in the UK, France and the US even of slow travel.
less so (EU, 2000; Giuliano and Dargay, 2006; Gatersleben and It is important to note that there is no clear causal relationship
Appleton, 2007). Among other factors this has to do with the bicy- between trip length and travel mode choice from a theoretical
cle being seen as a serious transport mode. There are also clear in- point of view. It seems plausible that increasing trip lengths may
ter-urban differences within individual countries which seem to cause a shift towards the car. However, the causality might also
suggest the existence of ‘bicycle cultures’, but may also indicate be the other way round. Given a travel time budget which is by
different socio-demographic compositions. For instance, in the
German university towns of Freiburg, Munster and Erlangen the
1
bicycle plays a much more significant role in daily travel than in In transport studies these factors are usually denominated as socio-demograph-
ics, though this is just a formal term that does not say anything about the reasons why
most other German cities (BMVBW, 2002). However, I am not the underlying variables should influence travel behaviour, whereas life situation
aware of any comparisons that take trip length categories into explicitly points to an individual’s personal circumstances (e.g. social roles, social
consideration. contact) relevant for his or her travel.
78 J. Scheiner / Journal of Transport Geography 18 (2010) 75–84

and large stable over time, the increase in car use may allow for the respective basic samples without regional supplements
longer trips. The study by Ye et al. (2007) may serve to support (www.clearingstelle-verkehr.de).
the former direction of causality. They model the interrelation be- The continual improvement of the KONTIV design since 1976
tween trip chain complexity and transport mode, and conclude causes problems when comparing the four surveys (Table 1). For
that models in which chain complexity affects mode choice per- instance, in 1976 the respondents had to complete trip diaries
form better than models which assume reverse causality. This is for more than one day. This led to a lower number of trips being
true both for work trips and non-work trips. Although trip chain recorded on the second and third days due to the respondents’
complexity and trip length are certainly two different things, this increasing ‘fatigue’. From this survey, only the first random day is
result suggests that mode choice might be an outcome of what available for analysis. In the following surveys, only one random
somebody has to do and where he or she has to do it. day was recorded for each respondent. In the 1989 survey, the col-
This interpretation is further supported by the findings of Lan- lection of the completed questionnaires by messengers led to the
zendorf (2001), p. 205ff on the sequence of decisions about activi- under-representation of highly mobile individuals/households.
ties, activity places and travel modes in leisure travel. His findings This effect was further intensified due to the substitution of these
indicate that in the overwhelming majority of cases people first de- households with others. In addition, it seems likely that in many
cide on the destination, before they decide on the travel mode. Ta- cases the trip diaries were completed by a representative of the
ken together, these findings might be interpreted as an indication household when the messenger arrived. This led to even more
that people decide on a certain destination (and thus, implicitly, on unrecorded trips (Kloas and Kunert, 1994). In the 2002 survey,
a certain trip length) and the mode choice decision tends to be ‘at the telephone method was used for the first time and led to dis-
the end of the pipeline’, even if there are certainly trips for which it tinctly higher trip counts, compared to the former KONTIV surveys,
is the other way round, e.g. trips to the countryside without a pre- due to the direct enquiry approach. Households whose telephone
determined destination. number could not be traced were contacted by post and asked
either to supply their telephone number or to complete a self-
2.4. Hypotheses administered questionnaire. More than 80% of those in this group
who participated in the survey supplied their telephone number
To sum up, the interrelation between realised or necessary trip (Kunert and Follmer, 2005).2 A special questionnaire module for
lengths on the one hand and travel mode choice on the other hand business trips increased the trips counts even further.
has not yet been conclusively determined. In addition, there is a Furthermore, the samples of the surveys are not immediately
considerable lack of research concerning trends over time in this comparable (see Kunert et al., 2002; Scheiner, 2006a). In the
area. The following hypotheses may be derived from the literature 1976, 1982 and 1989 surveys the basic population was the ‘Ger-
review: man-speaking’ residential population (the criterion ‘German-
speaking’ was handled rather arbitrarily), and the lower age limit
 Travel mode choice corresponds closely to travel distances. In was set at six years in 1989, but at ten in 1976 and 1982. In
this paper, it is treated as dependent on travel distance. This 2002 the whole residential population including foreigners was
hypothesis may sound not very spectacular, but serves to intro- considered for the first time, and the survey was extended to East
duce the following hypotheses. Germany. Moreover, in 1989, and partially in 1982, a random route
 Even for a given distance, travel mode choice depends on the method was used which took non-registered residents into ac-
transport means available to an individual (besides other count, in contrast to the other surveys.
resources and social roles). As a consequence, we may expect Due to these problems analyses here are limited to consider-
an increase in car use over time even for short trips due to the ation of German individuals (1976 to 1989: total sample, 2002:
increasing motorisation. Within groups defined by a certain persons with German nationality) aged 10 and older. Business trips
level of car availability, however, travel mode use for a given trip are excluded. The analysis of the 2002 survey is limited to the old
distance should by and large remain constant over time. Should Länder (former West Germany). The resulting net samples
this be proved empirically, it would indicate a stable rationale (counted in persons as well as trips) are given in Table 1. All dis-
underlying mode choice. tance estimates are self-reported by the respondents.
 In urban contexts the costs of car use relative to non-motorised
modes are higher (or, to put it the other way round, the benefit 3.2. Analysis
of the car is lower) than in suburban and rural areas. As a conse-
quence, short trips are more often made using non-motorised The findings reported in this paper are based on descriptive
modes in urban contexts than elsewhere, even within groups analyses of the four KONTIV surveys, complemented by cross-sec-
defined by a certain level of car availability. tional spatial comparisons. Travel behaviour indicators studied in-
clude modal split subdivided by distance categories, city size
categories and car availability. In addition, car availability itself is
3. Methodology examined by city size categories. An initial spatial classification
distinguished three region types developed by the Bundesamt für
3.1. Data used Bauwesen und Raumordnung (agglomerations, urbanised regions,
rural regions) over and above city size categories. This was based
Long-term trends in travel behaviour of the German population on the assumption that the increase in car travel may have been
on an individual basis can best be analysed by using KONTIV data particularly steep in suburban communities, compared to rural
(‘continued survey of travel behaviour’). KONTIV is a repeated communities. As the results did not show any systematic variation
semi-official nation-wide survey undertaken on behalf of the Fed- between the region types, the spatial differentiation here is limited
eral Transport Ministry. It is based on the random day principle. to city size categories.
The analyses in this paper use the four KONTIV surveys available
to date. They were carried out in the years 1976, 1982, 1989 and
2
2002. The last survey was renamed as ‘Mobility in Germany’ Although this still does not solve the problem that households with ’secret’
telephone numbers are likely to be a selective sample and may be characterised by
(MID) after a comprehensive methodological revision. The data active lifestyles and thus high mobility, the survey results do not suggest that this
are provided by the Clearingstelle Verkehr in Berlin. They include leads to underestimation of trips.
 J. Scheiner / Journal of Transport Geography 18 (2010) 75–84 79

Table 1
Methodological comparison of the four KONTIV surveys.

KONTIV 1976 KONTIV 1982 KONTIV 1989 KONTIV/MID 2002

Survey institute Socialdata Socialdata (Sozialforschung Emnid DIW, infas-Institut für angewandte
(Sozialforschung Brög) Brög) Sozialwissenschaft
Sampling procedure Address books 1/3 address books, 2/3 Random route Community register
random route
Population German speaking residential population aged. . . Total residential
10 years and older 10 years and older 6 years and older Population
Survey methodology Self-administered postal self-administered postal self-administered, by messenger with CATI + postal survey
survey survey completion support
No. of random days 2–3 1 1 1
Return rate 72% 66% 64% 42%
Sample of analysis 41,318 38,411 40,194 25,730
(individuals)*
sample of analysis 74,010 72,056 81,763 53,392
(trips)**

Source: author’s composition based on Kloas and Kunert (1994), Kunert et al. (2002), Kunert and Follmer (2005).
*
After filtering individuals under 10 years of age, foreigners, East Germans.
**
After filtering complex trip chains with more than two trips.

Examination of modal split by distance categories faces the trip frequencies in cities and small towns differently. There is,
problem that travel mode choice does not necessarily depend on however, no reason for such an assumption. Comparisons of trends
the length of a simple trip, but rather on the length of the whole thus seem reasonable regardless of methodological differences be-
trip chain. For instance, the car may be used for a short trip when tween the surveys, and may lead to instructive results.
the destination is only an intermediate stop in a chain. In order to
account for the effects of complex trip chains on travel mode
choice, the analyses were limited to ‘simple’ home-based trip 4. Results
chains with two trips (outward and return trip). This means that
68% of all reported trips are considered, taking all surveys together. 4.1. Travel mode choice in distance categories 1976–2002
What is more, it should be noted that car availability was re-
corded in the first three surveys in terms of individual car owner- In the KONTIV data there are some irregularities over time in
ship; in 2002 respondents were asked about individual car modal split within the distance categories. These are at least partly
availability with four possible answer categories (at any time, due to methodological changes between the surveys. For instance,
occasionally, by way of exception, never). In this paper respon- in the 1989 survey the proportions of trips undertaken on foot are
dents who claimed to have a car available at any time are treated higher in some distance categories than one would expect, while
as car owners. This conforms to the previous surveys. the corresponding car shares are lower (Table 2). The bicycle
Because of the limited comparability of the surveys, the time shares show irregularities as well. This is probably due to low-
series should not be over-interpreted as overall trends. Thus, the mobility and neighbourhood-oriented individuals being overrepre-
interpretation mainly refers to comparisons of trends, e.g. between sented in the 1989 survey (see Section 3.1).
different city size categories, rather than marginal trends in total. Focussing on the main tendencies makes clear, however, that
Such relative interpretations only become problematic when the tendency to walk short distances has declined markedly over
changes in the survey methodology affect the units to be compared time. To a limited degree this is compensated for by the bicycle,
in different ways. To give an example: a comparison of trends for which in Germany has gained in importance over recent decades:
cities and small towns is problematic only if the change from the in the 1970s it was generally regarded as a travel mode for chil-
written to the telephone-based survey design affected reported dren, adolescents and the poor. Nonetheless the bicycle did not

Table 2
Modal split by trip distance, 1976–2002.

km 1976 1982 1989 2002

On foot Bicycle PT* Car** On foot Bicycle PT* Car** On foot Bicycle PT* Car** On foot Bicycle PT* Car**
60.2 96 3 0 1 92 5 0 3 95 3 0 2 94 5 0 1
0.2–0.4 90 7 0 3 84 10 0 5 89 7 0 3 81 11 0 7
0.4–0.6 81 12 0 7 76 15 1 8 76 12 0 11 64 19 0 17
0.6–0.8 73 14 1 13 66 19 0 14 74 14 2 10 56 21 1 21
0.8–1.0 64 15 2 19 53 23 1 22 58 20 1 20 38 19 1 40
1.0–1.5 51 19 3 26 44 24 2 29 48 24 3 24 25 19 3 53
1.5–2.0 39 17 8 36 30 23 6 41 32 22 5 41 18 17 5 60
2–3 24 15 14 46 20 20 13 46 19 20 10 50 10 14 7 68
3–5 10 11 26 53 10 12 21 57 8 14 17 61 4 9 10 77
5–7 3 7 30 59 3 7 25 64 2 7 21 69 1 6 11 81
7–10 1 4 28 66 2 5 26 67 1 5 19 74 1 4 12 82
10–20 1 2 29 68 0 3 23 73 1 3 16 80 0 2 10 87
>20 0 0 25 74 0 1 23 75 0 1 12 86 1 1 13 85
Total 37 10 14 39 29 13 13 44 29 13 10 48 21 10 7 62

Source: author’s analysis. Data: KONTIV 1976, 1982, 1989, 2002.

Complex trip chains with more than two trips excluded. The values do not always sum up to 100 due to the exclusion of other transport modes (mostly 0–1%).
*
PT: public transport (including long-distance train and coach).
**
Car including motorcycle.
80 J. Scheiner / Journal of Transport Geography 18 (2010) 75–84

gain in total (over all distance categories), because its gains on surveys in recording car availability, only two ‘extreme groups’
short trips were countered by the shift in distance distribution to- are considered here: individuals without a car in their households
wards longer trips. (termed: car-less) and individuals having access to a car at any
The main winner of the decline in walking, however, is the pri- time (termed: car owners).
vate car. The car shares increase in all distance categories, most Turning our attention to car owners, there are some shifts in
pronouncedly among short trips below 2 km, albeit from a low le- modal split among short trips below 2 km, namely from walking
vel. This increase is mainly at the expense of walking, while for (and, to a lesser extent, from driving) towards the bicycle. This
longer trips it is at the expense of PT. PT declines sharply in the might well be understood as a mixture of increasing time rational-
medium distance categories between 3 km and 20 km, i.e. in urban ity on the one hand, and ‘cultural’ value change (increasing envi-
and regional transport (its decline on long-distance trips is smaller; ronmental awareness, health awareness, appreciation of the
this is not explicitly shown in Table 2). bicycle) on the other hand. For instance, the increasing apprecia-
The picture changes as soon as car availability is taken into ac- tion of the bicycle in Germany since the 1970s is supported by Pu-
count (Table 4). Due to methodological differences between the cher and Buehler (2008), and the Eurostat survey provides
evidence for the long-term increase in environmental awareness
over the past decades (Aidt, 2005, see also Kuckartz et al., 2007
Table 3
Car availability by gender and age group, 1976–2002. for more recent trends in Germany).
However, taking into account that the modal changes found
Gender Age group
here are mainly at the expense of trips on foot, time rationality
Men Women All 18–24 25–64 65+ All seems to be dominant. This is generally supported by the observa-
Car fully available *
tion of an increasing acceleration of society (Virilio, 2005), and
1976 65.8 14.2 37.8 34.1 45.2 13.3 37.8 more specifically by recent evidence on the ‘densification’ of time
1982 79.8 32.2 56.9 40.3 66.0 30.4 56.9
use by means of simultaneous activities or multitasking, particu-
1989 76.3 30.6 52.9 48.0 61.2 25.1 52.9
2002 80.1 60.3 69.6 57.1 76.4 50.7 69.6 larly by making use of information and communication technolo-
gies during travel (Lyons and Urry, 2005; Kenyon and Lyons,
Share of the group among those with full car availability
1976 79.5 20.5 100.0 9.8 83.4 6.8 100.0 2007), although there is still a substantial lack of long-term studies.
1982 72.9 27.1 100.0 8.3 82.6 9.1 100.0 However, the cultural value change that might hide behind the
1989 70.3 29.7 100.0 9.7 81.2 9.1 100.0 figures should not be underestimated. While in 1976 only one
2002 54.1 45.9 100.0 6.4 78.5 15.1 100.0 mode of travel other than the car was common among car owners
*
Car fully available: 1976–1989: individual car owner; 2002: car available for – i.e. their own feet – some of them at least occasionally use the
driving at any time. Source: author’s analysis. Data: KONTIV 1976, 1982, 1989, bicycle for short trips in 2002. This may well be interpreted as an
2002. expression of a certain flexibilisation of travel mode choice.

Table 4
Modal split by trip distance and car availability, 1976–2002.

km 1976 1982 1989 2002

* ** * **
On foot Bicycle PT *
Car **
On foot Bicycle PT Car On foot Bicycle PT Car On foot Bicycle PT* Car**
***
Car fully available
60.2 94 2 0 4 91 4 0 6 92 2 1 6 95 3 0 1
0.2–0.4 84 5 0 11 80 8 0 11 84 7 1 9 78 12 0 10
0.4–0.6 70 5 0 25 70 9 0 20 68 9 0 23 62 16 0 22
0.6–0.8 54 7 0 38 59 12 0 27 64 14 1 22 50 20 0 30
0.8–1.0 42 6 0 52 41 14 0 43 46 12 1 41 33 15 0 51
1.0–1.5 31 6 1 62 34 13 1 52 39 15 1 44 20 16 0 64
1.5–2.0 23 5 1 72 21 9 2 67 22 11 1 66 14 13 2 70
2–3 15 4 2 79 16 7 3 73 14 11 3 71 8 10 3 79
3–5 7 3 3 87 8 5 4 82 7 6 4 84 3 7 3 87
5–7 2 3 5 89 3 5 3 89 2 3 6 88 1 4 4 90
7–10 1 1 4 94 1 3 5 90 1 4 4 91 1 3 5 90
10–20 0 1 6 93 0 2 6 91 0 2 4 94 0 1 4 94
>20 0 0 11 89 0 0 13 86 0 0 7 92 1 1 8 90
Total 19 3 4 74 19 6 4 70 18 6 3 72 17 7 3 73
No car in household
60.2 97 2 0 1 96 3 1 1 99 1 0 0 96 3 1 0
0.2–0.4 93 6 0 1 92 6 1 2 97 3 0 0 92 6 0 2
0.4–0.6 87 11 0 1 88 9 2 1 92 6 0 2 76 19 1 4
0.6–0.8 85 11 3 1 80 15 0 4 86 11 1 1 73 16 3 2
0.8–1.0 81 13 5 1 80 11 6 3 78 16 2 4 58 24 8 9
1.0–1.5 71 18 6 5 69 14 9 8 71 16 6 7 45 23 11 21
1.5–2.0 60 15 17 8 51 23 21 5 48 26 13 12 38 24 23 15
2–3 40 17 30 14 36 21 32 10 32 27 27 13 22 31 31 14
3–5 18 13 55 14 18 15 57 10 16 21 46 17 11 16 48 23
5–7 5 12 66 17 5 10 69 17 5 11 59 24 3 15 59 21
7–10 4 6 69 20 4 5 65 24 2 8 65 24 2 6 63 29
10–20 0 3 70 26 1 3 72 24 1 7 57 30 0 13 45 40
>20 0 1 67 32 0 1 62 36 0 4 42 54 1 0 54 46
Total 57 11 23 9 51 12 28 9 55 14 19 11 44 16 24 15

Source: author’s analysis. Data: KONTIV 1976, 1982, 1989, 2002.

*
PT: public transport (including long-distance train and coach).
**
Car including motorcycle.
***
Car fully available: 1976–1989: individual car owner; 2002: car available for driving at any time. Further remarks see Table 2.
 J. Scheiner / Journal of Transport Geography 18 (2010) 75–84 81

Table 5
Modal split by trip distance and city size category, 2002 (car owners).

km <50,000 inh. 50–100,000 inh. 100–500,000 inh. >500,000 inh.

* ** * ** * **
On foot Bicycle PT Car On foot Bicycle PT Car On foot Bicycle PT Car On foot Bicycle PT* Car**
60.2 94 5 0 1 97 3 0 0 96 2 0 2 100 0 0 0
0.2–0.4 71 16 0 12 82 8 0 10 85 5 0 10 93 5 0 1
0.4–0.6 57 17 0 27 68 9 0 22 61 21 1 17 81 13 0 6
0.6–0.8 43 20 0 37 42 20 0 39 62 20 1 17 70 15 0 14
0.8–1.0 30 13 0 56 29 18 0 53 42 17 1 40 47 14 0 37
1.0–1.5 18 15 0 67 11 16 0 72 25 18 2 55 37 20 1 42
1.5–2.0 11 11 1 76 19 12 0 69 15 19 4 62 22 17 8 52
2–3 9 8 1 82 5 11 2 82 9 14 7 70 9 12 7 71
3–5 3 5 0 91 2 7 2 89 3 9 8 80 4 8 12 76
5–7 1 4 0 95 1 1 4 94 1 7 7 86 4 6 15 74
7–10 1 2 2 95 0 6 2 92 1 3 7 88 2 3 23 71
10–20 0 1 2 96 0 3 1 96 0 3 5 92 0 1 15 84
>20 1 1 9 89 0 0 5 95 0 0 7 92 0 2 12 86
Total 15 7 2 76 14 8 2 76 18 10 5 67 24 8 10 58

inh.: Inhabitants. For further remarks see Table 2. Source: author’s analysis. Data: KONTIV 2002.

Besides time rationality and value change, a change in the social trips longer than about two to three kilometres does bicycle use
structure of car owners may be reflected here, e.g., the increasing decrease markedly. In this distance category, PT use increases con-
shares of women and retirees owning a car. The share of women siderably and quickly reaches its maximum, which is held even for
among all car owners more than doubled from 20% to 46% over long-distance trips. Individuals without a car use the bicycle much
the study period, and the share of those aged 65+ increased from more often than car owners even for relatively long-distances.
7% to 15% (Table 3). As a consequence, the activity spectra of car
owners may have changed over the study period (increase in shop- 4.2. Comparing city size categories
ping and leisure trips). This again may have an impact on mode
decisions. What is more, the improvements in bicycle infrastruc- It was hypothesised above that short to medium distances are
ture (mainly in the 1990s) may be reflected in the increase in bicy- more likely to be seen as acceptable walking distances in urban
cle use in some distance categories. contexts than in other spatial settings. The KONTIV data allow
However, it is striking that in many of the distance categories for a comparison between city size categories. When making this
the car proportions in 2002 are very similar to those in 1976. comparison car availability has to be controlled due to the lower
This suggests that for a given distance and a given availability motorisation level in large cities compared to small towns. The
of transport means the decision rationale of mode choice has comparison is therefore limited to individuals with access to a
hardly changed over time. Given certain conditions, the behav- car at any time (car owners, Table 5). This includes 69% of the adult
ioural outcome appears to be fairly stable. It has to be noted, population who are of particular interest, as it may be more rea-
however, that private vehicles may only be characterised as sonable to assume a free mode choice decision for these individu-
pre-conditions of behaviour from a short-term perspective. In als than for others.
the long run the possession of a car is in itself subject to certain The results confirm the hypothesis. Short trips are much more
decision rationales. often covered by NMT in large cities than in smaller towns. Up to
Among the car-less, car use increased considerably from extre- a distance of 1.5 km the NMT figures are no less than about 20%
mely low values in 1976. Taking the lack of direct access to a car in points higher in the largest cities than in towns with less than
the household into account, this might be due to a change in exter- 50,000 inhabitants. Only for the shortest trips (<200 m) do small
nal circumstances. With increasing motorisation, even individuals town residents use NMT almost as often as city dwellers, although
in car-less households increasingly have a chance to participate in they show a stronger tendency towards the bicycle than towards
motorisation and use a car from outside their own households, e.g., walking. For trips between 200 and 400 m in length the difference
cars owned by their adult children, parents, organised car-sharing, is 11% points, and it increases to 20% points from 400 m. Distances
etc. At the same time, a shift in mode choice in favour of the bicycle over 2 km are seldom walked, regardless of the spatial setting.
and at the expense of walking can be observed for this group in However, medium distance trips between one and seven kilome-
many distance categories. Nevertheless, the shifts within the cate- tres are undertaken more often by bicycle in urban environments
gories are considerably smaller for the car-less than for the total than in small towns. At the same time, the share of PT is markedly
population. Again this suggests a certain stability in the decision higher in cities than in small towns.
rationales of mode choice. The spatial environment thus indeed appears to have a certain
Of relevance to the issue of using spatial planning concepts to influence on the use of the car: city dwellers tend to leave their
affect travel mode choice is the question of the distance thresholds car at home for short trips more often than small town residents,
at which mode choice changes notably (Tables 2 and 4). even if they have it readily available. There is no conclusive answer
The analyses confirm the importance of a highly differentiated as to the driving forces behind this. One might assume higher gen-
subdivision of distances (Tables 2 and 4). The share of walking de- eralised costs of car use in urban environments due to traffic den-
creases at a trip distance of not more than 400 m (in small towns sity and lack of parking space. Urban environments might be more
even at 200 m, see below). A further decrease is at a distance of exciting or otherwise attractive for walking.3 In either case, both
800 m. For car owners the thresholds are lower than for those
without a car in the household. However, the distance thresholds
3
are less obvious than expected. At least it is more appropriate for trip chaining and therefore for walking, due to
the high density and variety of land-use. For instance, among car owners the share of
The share of the bicycle notably increases at distances of no trips that are part of complex trip chains is 41% in cities with more than 500,000
more than 200–400 m. It reaches its maximum at 600–800 m inhabitants, while in communities with less than 100,000 inhabitants it is only 37%
and decreases slightly in the longer distance categories. Only for (all surveys taken together).
82 J. Scheiner / Journal of Transport Geography 18 (2010) 75–84

interpretations suggest that within cities the NMT shares should be Table 7
higher in inner city neighbourhoods than in the outskirts. Evidence Travel mode choice by city size category, 1976–2002.

for this can be found by a, albeit rough, spatial subdivision of the City size (1000 inh.) NMT PT* car**
KONTIV data within cities (Table 6). The analysis distinguishes be- <2 1976 33.8 9.2 56.9
tween areas where detached and semi-detached houses dominate 1982 30.2 8.6 61.2
and those where larger buildings dominate. The latter type essen- 1989 37.4 4.8 57.8
tially represents centrally located, dense, mixed-use neighbour- 2002 25.2 3.2 71.7
2–5 1976 38.8 8.3 52.8
hoods, while the former by and large represents less dense, less 1982 35.1 9.1 55.8
mixed-use peripheral areas. The results show that in the peripheral 1989 37.2 5.5 57.3
areas of the large cities the car and the bicycle are used more often 2002 23.5 3.9 72.5
for short trips, whereas in central areas people are more inclined to 5–20 1976 42.9 6.9 50.2
1982 36.7 7.6 55.8
walk. The spatial differences are substantial. For longer trips, PT is
1989 37.0 5.3 57.6
used more often in central areas. In total, car use among car owners 2002 27.4 3.8 68.8
is markedly lower in the central areas of the cities than in periph- 20–100 1976 41.8 8.8 49.4
eral neighbourhoods. 1982 37.4 8.5 54.1
The following section addresses the question as to whether 1989 41.8 6.4 51.8
2002 30.1 4.8 65.1
there are trends over time in the spatial differentiation of motori-
100–500 1976 38.0 13.7 48.3
sation and travel mode choice. 1982 38.5 13.0 48.5
1989 41.4 12.0 46.7
4.3. Comparing city size categories: trends over time 2002 34.2 9.5 56.3
500+ 1976 38.9 19.8 41.3
1982 35.5 20.5 44.0
In this section the focus is on spatial comparisons of trends in 1989 36.9 19.0 44.1
travel mode choice and car availability. Car availability refers to 2002 36.1 19.1 44.8
car ownership of individuals as well as population shares according Total 1976 40.1 11.2 48.7
to the motorisation of the households they live in. 1982 36.5 11.1 52.4
1989 39.0 8.9 52.1
It should be noted that the analyses may again include method-
2002 30.2 7.2 62.5
ological flaws. In the 1989 data the NMT shares are higher than ex-
pected, while the motorisation level is lower (Tables 7 and 8). For further remarks see Table 2. Source: author’s analysis. Data: KONTIV 1976, 1982,
1989, 2002.
Again this is likely to be caused by the under-representation of
highly mobile individuals/households in this survey. In the 1982
data the population share living in households without a car is Table 8
unexpectedly low (Table 8). Individual car ownership and population shares by number of cars in the household,
by city size category, 1976–2002.
However, the spatial comparison is instructive in any case. As
expected, the PT share has declined remarkably over the study per- City size Individual car No car in One car in 2+ Cars in
iod (Table 7). The same is true for NMT. Both these trends are con- (1000 inh.) ownership* household household household
(Population shares)
firmed for all city size categories except for the largest cities with
more than 500,000 inhabitants. In these cities the PT share remains <2 1976 38.4 20.6 58.4 21.0
1982 58.0 7.1 54.8 38.2
on a constant level, while the NMT share only marginally declines.
1989 60.2 10.7 48.4 40.9
In medium size cities of between 100,000 and 500,000 inhabitants 2002 81.0 5.4 42.7 51.9
NMT faces only minor losses as well. 2–5 1976 37.9 23.8 54.6 21.5
Motorisation shows a corresponding spatial picture (Table 8). 1982 54.7 7.6 56.5 35.9
Car availability clearly increased in all city size categories over 1989 54.8 16.0 46.1 37.9
2002 78.9 5.4 43.9 50.7
5–20 1976 36.9 26.6 56.2 17.2
1982 56.8 9.6 55.8 34.6
Table 6 1989 54.2 18.4 49.8 31.8
Modal split by trip distance and location within the city, 2002 (car owners, cities 2002 76.6 8.7 45.7 45.6
>500,000 inhabitants). 20–100 1976 37.3 28.8 55.6 15.6
1982 56.3 12.1 58.3 29.6
km Central areas Peripheral areas 1989 52.8 19.9 54.3 25.8
2002 72.9 12.0 51.7 36.3
On foot Bicycle PT* Car** On foot Bicycle PT* Car**
100–500 1976 35.7 33.7 52.4 13.9
60.2 100 0 0 0 96 0 0 4 1982 54.5 17.9 57.8 24.2
0.2–0.4 96 3 0 1 85 11 0 4 1989 48.3 28.0 50.9 21.1
0.4–0.6 85 9 0 6 73 20 0 6 2002 65.1 18.7 52.7 28.6
0.6–0.8 77 11 0 11 52 24 0 24 500+ 1976 33.1 39.8 49.0 11.2
0.8–1.0 61 10 0 29 31 18 1 47 1982 49.9 23.7 57.5 18.8
1.0–1.5 39 16 2 43 35 25 0 40 1989 46.4 31.5 50.1 18.5
1.5–2.0 22 15 12 51 23 20 2 55 2002 57.7 28.8 52.4 18.8
2–3 12 9 9 69 5 16 5 74 Total 1976 36.3 30.1 54.0 15.9
3–5 5 8 17 71 2 9 4 85 1982 55.0 13.5 57.1 29.4
5–7 2 7 16 75 8 5 13 73 1989 51.7 22.2 50.8 27.0
7–10 4 2 28 67 0 6 15 79 2002 71.3 13.8 49.2 36.9
10–20 0 1 16 83 0 1 12 87
*
>20 0 1 13 86 0 3 11 86 The column shows individual car ownership, but for 2002 individual car avail-
Total 28 6 12 55 18 11 6 64 ability at any time due to a different survey methodology. Source: author’s analysis.
Data: KONTIV 1976, 1982, 1989, 2002.
Due to a lack of data on small-scale location within individual cities, the neigh-
bourhoods are classified according to dominant house types. Neighbourhoods
where detached and/or semi-detached houses dominate are classified as ‘periphe-
the study period. However, the increase was least distinct in large
ral’. Neighbourhoods where larger buildings dominate are classified as ‘central’. For cities. Roughly speaking, the population share with individual car
further remarks see Table 2. Source: author’s analysis. Data: KONTIV 2002. ownership doubled from 1976 to 2002 in all city size categories
 J. Scheiner / Journal of Transport Geography 18 (2010) 75–84 83

up to 100,000 inhabitants. E.g., in small communities with less of PT for longer trips. According to this, the shift in modal split to-
than 2000 inhabitants it rose from 38% to 81%; in cities from wards the car is not (only) caused by increasing trip distances and,
100,000 to 500,000 inhabitants the increase is slightly lower; in thus, expanding activity spaces, but took place even within dis-
cities with more than 500,000 inhabitants even lower (‘only’ from tance classes. Once car availability is taken into account, these
33% to 58%). modal shifts appear to be considerably weaker. This suggests that
The widening spatial gap in motorisation is particularly striking once car availability is held constant the decision rationales of
with respect to households with more than one car. In communi- mode choice for a certain trip distance have remained relatively
ties with less than 2000 inhabitants, 21% of the population older stable, while the change in decision rationales mainly resulted in
than 10 years of age lived in households with more than one car increasing car ownership.
in 1976. By the year 2002 this proportion had increased by a factor Spatial comparisons show that the increase in motorisation
of 2.5–52%. In cities with more than 500,000 inhabitants the corre- over the study period was considerably weaker in large cities
sponding share increased from 11% to 19% in the same period. This than in small towns and the countryside, although the cities
corresponds to a factor of 1.7. This is clearly a marked increase. started from a lower level in the 1970s. Thus, the motorisation di-
However, given that the population in the large cities started this vide between cities on the one hand, and suburban and rural
period from a markedly lower motorisation level than the small areas on the other hand has become ever wider. For travel mode
town and rural population, one may well speak of a widening choice, the picture is similar. While car use in small towns steeply
gap. E.g., the proportion of people living in non-motorised house- increased at the expense of PT and NMT, modal split remained
holds was 39.8% in the largest cities in 1976, while in the smallest rather stable in the cities. This may be traced back to the built
communities it was only 20.6%. environment, or to the residential self-selection of households
This raises the question as to whether travelling by automobile without a car or with only one car who move to (or stay in)
would have increased (to the extent it has) if urban structure had the cities.
developed in favour of the large cities instead of the small (subur- What is more, the results suggest that even car owners are more
ban) communities. Basically: can suburbanisation be blamed for inclined to walk a given distance in the cities than in small towns,
the increase in car travel? This interpretation is possible, and, con- even more so if they live in a central urban area. For trips over two
sidering the effects of the built environment on motorisation and kilometres the use of PT or the bicycle is more prevalent in the cit-
travel mode choice found in numerous studies (e.g. a recent study ies than in smaller towns.
by Scheiner and Holz-Rau, 2007), it is plausible and likely to have The built environment thus indeed appears to have an impact
a kernel of truth. However, the debate on residential self-selection on whether an available automobile is used or not. Without being
noted in Section 2 has to be accounted for as well. There is clear able to factor out the key determinants for this, the reasons may be
evidence that car availability is not only an outcome of house- found in the transport system as well as in urban structure. Firstly,
holds’ residential location decisions, but also a pre-condition for lack of parking space and high car traffic density may be decisive,
these decisions. Highly motorised households tend to move to i.e. higher generalised costs of car travelling. Secondly, urban envi-
peripheral locations (and on the regional level this means, by ronments may be more attractive or exciting for walking, or more
and large, to small suburban communities), while households appropriate for coupling activities on foot.
with no or only one vehicle tend to move to the cities (Scheiner, Given the lower motorisation combined with the lower car use
2005, 2006a). of motorised individuals in the cities, one may conclude in any case
Thus, the trends found here may reflect several mechanisms, that the dense, mixed land-use structure of the cities – particularly
namely the effect of urban structure (large cities tend to curb in the centrally located neighbourhoods – appears to be a key fac-
motorisation), the self-selective in-migration of households with- tor for a relatively low level of car use. This is even more so the
out a car or with only one car, and the self-selective stay of the case, as the well-known shorter travel distances of city dwellers
same type of households in the cities. This type of residential loca- have not been considered here.
tion behaviour may reflect preferences as well as (non) available This has important, albeit not new, consequences for policy. De-
resources. The different effects that may be hidden here cannot spite urban sprawl and increasing spatial dispersion, most cities in
be separated using the data analysed. However, one point may a European context still have a lively centre, accommodating an ur-
be noted: ban population who lead their daily lives with relatively short
As car travelling increasingly becomes a societal norm and also trips, low motorisation levels and high levels of walking. Although
a necessity (because of the spatial and individual separation of ur- travel behaviour has been shown to be at least partly an effect of
ban functions such as housing, working, shopping and so on), the residential self-selection, most related studies find that the built
big cities become (with some slight exaggeration) the last areas environment has significant effects as well (Cao et al., 2006; Schei-
where one can still live without a car and where car-less house- ner and Holz-Rau, 2007). This study confirms that spatial planning
holds are gathering, regardless of whether their car-less life is vol- may indeed affect travel. But even if the specific travel behaviour of
untarily chosen or caused by a lack of resources. urban – as compared to suburban or rural – populations is mainly
an effect of residential self-selection, there is still a ‘built environ-
ment effect’ in terms of the provision of options for residential
5. Outlook choice. There is no residential self-selection as long as there are
no options to be selected. Thus, even in this case it is still worth-
This paper reports findings from longitudinal analyses of the while making cities more healthy, attractive and liveable for those
German nation-wide travel survey KONTIV for the period 1976– who prefer to live there, thereby providing options for living with-
2002. It focuses on travel mode choice, subdivided by distance cat- out a car or at least for using it less. This is particularly true as sub-
egories, car availability, and city size categories. In addition, trends urbia might face increasing problems in the future due to
in car availability itself are examined by city size categories for the demographic ageing and rising transport prices (Hesse and Schei-
said study period. ner, 2007).
The results indicate that even within distance categories car use Finally, the limitations of the dataset and the methodology of
has considerably increased. The proportion of bicycle use has par- analysis should be kept in mind. The four surveys differ in sam-
tially increased as well. The gains of the private car are mainly at pling procedure, survey methodology and content of the question-
the expense of trips on foot for shorter trips, and at the expense naires. Because of these limitations, the time series should not be
84 J. Scheiner / Journal of Transport Geography 18 (2010) 75–84

over-interpreted as overall trends. Thus, the interpretation mainly Handy, S.L., Clifton, K.J., 2001. Local shopping as a strategy for reducing automobile
travel. Transportation 28 (4), 317–346.
referred to comparisons of trends, e.g. between different city size
Hesse, M., Scheiner, J., 2007. Suburban areas – problem neighbourhoods of the
categories, rather than marginal trends in total. Improved long- future? Ger. J. Urban Studies 46 (2), 35–48.
term monitoring of travel behaviour in terms of adequate, stable Holz-Rau, C., Scheiner, J., 2006. Die KONTIVs im Zeitvergleich. Möglichkeiten und
survey methodologies is clearly desirable for Germany. Schwierigkeiten beim Vergleich der Erhebungswellen. Internationales
Verkehrswesen 58 (11), 519–525.
With respect to the methods of analysis applied, the findings re- Holz-Rau, C., Rau, P., Scheiner, J., Trubbach, K., Dörkes, C., Fromberg, A., Gwiasda, P.,
ported in this paper are based on descriptive statistics. Clearly, a Krüger, S., 1999. Nutzungsmischung und Stadt der kurzen Wege: Werden die
multivariate analysis could lead to a more thorough understanding Vorzüge einer baulichen Mischung im Alltag genutzt? Werkstatt Praxis (7/
1999). Bundesamt für Bauwesen und Raumordnung, Bonn.
of the relationship between trip distance and mode choice as well Holz-Rau, H.-C., 1991. Verkehrsverhalten beim Einkauf. Internationales
as possible changes over time in this relationship. Such an analysis Verkehrswesen 43 (7–8), 300–305.
should account for a range of other variables beside the ones con- ILS (Institut für Landes- und Stadtentwicklungsforschung Nordrhein–Westfalen),
2002. Baulandentwick lung an der Schiene. ILS, Dortmund.
sidered here, such as socio-demographics, transport prices (public Kenyon, S., Lyons, G., 2007. Introducing multitasking to the study of travel and ICT:
transport fares, fuel), or congestion levels. The changes in all of examining its extent and assessing its potential importance. Transport. Res. A
these in recent decades may have affected trends in travel dis- 41 (2), 161–175.
Kloas, J., Kunert, U., 1993. Vergleichende Auswertungen von Haushaltsbefragungen
tances and mode choice. However, this would significantly extend zum Personennahverkehr (KONTIV 1976, 1982, 1989). Forschungsprojekt im
the scope of the research project this paper is based on. A multivar- Auftrag des Bundesministers für Verkehr (FE-Nr. 90361/92). DIW, Berlin.
iate longitudinal analysis would, however, be a worthy subject of Kloas, J., Kunert, U., 1994. Über die Schwierigkeit, Verkehrsverhalten zu messen.
Verkehr Technik 47 (3–5), 91–100 and 187–197.
further research and a future paper.
Kuckartz, U., Rheingans-Heintze, A., Rädiker, S., 2007. Tendenzen der Umwelt- und
Risikowahrnehmung in einer Zeit des Wertepluralismus. Umweltbundesamt,
Dessau.
Acknowledgements Kunert, U., Follmer, R., 2005. Methodological advances in national travel surveys:
mobility in Germany 2002. Transport. Rev. 25 (4), 415–431.
The research in this paper was funded by the German Federal Kunert, U., Kloas, J., Kuhfeld, H., 2002. Design characteristics of national travel
surveys. An international comparison for ten countries. Presentation at the
Ministry for Education and Research (BMBF) within the framework 2002 Annual Meeting of the Transportation Research Board.
of the programme ‘Research for the Reduction of Land Consump- Lanzendorf, M., 2001. Freizeitmobilität. Unterwegs in Sachen sozial-ökologischer
tion and for Sustainable Land Management (REFINA)’. It is part of Mobilitätsforschung. Materialien zur Fremdenverkehrsgeographie (56).
University, Trier.
the project ‘Integrated residential location information as a contri-
Lyons, G., Urry, J., 2005. Travel time use in the information age. Transport. Res. A 39
bution to reduce land consumption’ (see www.raumplanung.uni- (2), 257–276.
dortmund.de/vpl/dienst/eng/content/forsch/forschpro/refina.html Mackett, R.L., 2003. Why do people use their cars for short trips? Transportation 30
(3), 329–349.
for further information). I would also like to thank two anonymous
Mokhtarian, P.L., Chen, C., 2004. TTB or not TTB, that is the question: a review and
reviewers for their helpful comments. analysis of the empirical literature on travel time (and money) budgets.
Transport. Res. A 38 (9–10), 643–675.
Naess, P., 2006. Accessibility, activity participation and location of activities:
References exploring the links between residential location and travel behaviour. Urban
Studies 43 (3), 627–652.
Aidt, T.S., 2005. The rise of environmentalism, pollution taxes and intra-industry Pucher, J., Buehler, R., 2008. Making cycling irresistible: lessons from the
trade. Econ. Gov. 6 (1), 1–12. Netherlands, Denmark and Germany. Transport. Rev. 28 (4), 495–528.
Apel, D., 1995. Möglichkeiten zur Steuerung des Flächenverbrauchs und der Scheiner, J., 2002. Die Angst der Geographie vor dem Raum. Anmerkungen zu einer
Verkehrsentwicklung. Zwischenbericht. Difu-Materialien (1/95). Deutsches verkehrswissenschaftlich-geographischen Diskussion und zur Rolle des Raumes
Institut für Urbanistik, Berlin. für den Verkehr. Geographische Revue 4 (1), 19–44.
Badland, H., Schofield, G., 2005. Transport, urban design, and physical activity: an Scheiner, J., 2005. Auswirkungen der Stadt- und Umlandwanderung auf
evidence-based update. Transport. Res. D 10 (3), 177–196. Motorisierung und Verkehrsmittelnutzung: ein dynamisches Modell des
Bahrenberg, G., 1997. Zum raumfetischismus in der jüngeren verkehrspolitischen Verkehrsverhaltens. Verkehrsforschung Online 1 (1), 1–17.
diskussion. In: Eisel, U., Schultz, H.-D. (Eds.), Geographisches Denken. Urbs et Scheiner, J., 2006a. Housing mobility and travel behaviour: a process-oriented
Regio (65). University, Kassel, pp. 345–371. approach to spatial mobility. Evidence from a new research field in Germany. J.
BMVBS (Bundesministerium für Verkehr, Bau und Stadtentwicklung) (Ed.), Verkehr Transport. Geogr. 14 (4), 287–298.
in Zahlen 2007/2008. Deutscher Verkehrs-Verlag, Hamburg, 2007. Scheiner, J., 2006b. Does individualisation of travel behaviour exist? Determinants
BMVBW (Federal Ministry of Transport, Building and Housing) (Ed.), National and determination of travel participation and mode choice in West Germany,
Cycling Plan 2002–2012: Ride Your Bike! BMVBW, Berlin, 2002. 1976–2002. Die Erde 137 (4), 355–377.
Cao, X., Handy, S.L., Mokhtarian, P.L., 2006. The influences of the built environment Scheiner, J., Holz-Rau, C., 2007. Travel mode choice. affected by objective or
and residential self-selection on pedestrian behaviour in Austin, Texas. subjective determinants? Transportation 34 (4), 487–511.
Transportation 33 (1), 1–20. Schlossberg, M., Greene, J., Phillips, P.P., Johnson, B., Parker, B., 2006. School trips:
Cervero, R., 2002. Built environments and mode choice: toward a normative effects of urban form and distance on travel mode. J. Am. Plann. Assoc. 72 (3),
framework. Transport. Res. D 7 (4), 265–284. 337–346.
Cervero, R., 2006. Alternative approaches to modelling the travel-demand impacts Schwanen, T., Dijst, M., Dieleman, F.M., 2004. Policies for urban form and their
of smarth growth. J. Am. Plann. Assoc. 72 (3), 285–295. impact on travel: the Netherlands experience. Urban Studies 41 (3), 579–603.
Crane, R., 2000. The influence of urban form on travel: an interpretive review. J. Simma, A., Axhausen, K.W., 2001. Structures of commitment in mode use: a
Plan. Lit. 15 (1), 3–23. comparison of Switzerland, Germany and Great Britain. Transport Pol. 8 (4),
DfT (Department for Transport) (Ed.), Transport statistics bulletin. National Travel 279–288.
Survey 2005. DfT, London, 2006. Stead, D., Marshall, S., 2001. The relationships between Urban form and travel
DIW/INFAS. Mobilität in Tabellen. Tabellenprogramm zur Erhebung Mobilität in patterns. An international review and evaluation. Eur. J. Transport.
Deutschland 2002. DIW, Berlin, 2003. Infrastructure Res. 1 (2), 113–141.
EU (Ed.), 2000. Transport in Figures 2000. European Commission, Luxembourg. Vågane, L., 2007. Short car trips in Norway: is there a potential for modal shift?
Filion, P., McSpurren, K., Appleby, B., 2006. Wasted density? The impact of Toronto’s Paper presented at the European Transport Conference, Leiden, Netherlands
residential-density-distribution policies on public-transit use and walking. (17–19 October).
Environ. Plan. A 38, 1367–1392. Van Wee, B., Rietveld, P., Meurs, H., 2006. Is average daily travel time expenditure
Forsyth, A., Oakes, J.M., Schmitz, K.H., Hearst, M., 2007. Does residential density constant? In search of explanations for an increase in average travel time. J.
increase walking and other physical activity? Urban Studies 44 (4), 679–697. Transport. Geogr. 14 (2), 109–122.
Gatersleben, B., Appleton, K.M., 2007. Contemplating cycling to work: attitudes and Virilio, P., 2005. Negative Horizon: An Essay in Dromoscopy. Continuum, London.
perceptions in different stages of change. Transport. Res. A 41, 302–312. Ye, X., Pendyala, R.M., Gottardi, G., 2007. An exploration of the relationship between
Giuliano, G., Dargay, J., 2006. Car ownership, travel and land use: a comparison of mode choice and complexity of trip chaining patterns. Transport. Res. B 41, 96–
the US and Great Britain. Transport. Res. A 40 (2), 106–124. 113.
Guo, J.Y., Chen, C., 2007. The built environment and travel behaviour: making the Zumkeller, D., Nakott, J., 1988. Neues Leben für die Städte: Grünes Licht fürs
connection. Transportation 34 (5), 529–533. Fahrrad. Bild der Wissenschaft 5, 104–113.