You are on page 1of 8

G Model

AAP-4383; No. of Pages 8 ARTICLE IN PRESS


Accident Analysis and Prevention xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Accident Analysis and Prevention


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

An experiment on rider stability while mounting: Comparing


middle-aged and elderly cyclists on pedelecs and conventional
bicycles
D.A.M. Twisk a,∗ , S. Platteel a , G.R. Lovegrove b
a
SWOV Institute for Road Safety Research, PO Box 93113, 2509 AC The Hague, The Netherlands
b
Engineering, Management & Education, The University of British Columbia, 1137 Alumni Ave, Kelowna BC V1 V 1V7, Canada

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

Article history: Pedelecs, popular among elderly cyclists, are associated with a higher injury risk than conventional bicy-
Received 30 December 2015 cles. About 17% of these injuries are due to falls while (dis)mounting. Using instrumented bicycles, this
Received in revised form study aimed to identify factors contributing to the stability of self-chosen mounting methods in four
14 November 2016
user groups: 30–45 versus 65+ years of age and males versus females. Mounting stability on pedelecs
Accepted 9 January 2017
was compared with that on conventional bicycles, in controlled experimental setting (task in a fenced
Available online xxx
off parking lot) but also in real traffic conditions (traffic light turns green). Two mounting phases were
differentiated: phase 1 as the transition from ‘earth bound’ to ‘balance’ and phase 2 as the acceleration
Keywords:
Pedelec
to achieve harmonized cycling. Stability was operationalised in terms of the duration of these phases:
Elderly cyclists the shorter their duration, the higher the stability. Pedelecs were shown to be less stable in phase 1 than
Single-bicycle accidents conventional bicycles, irrespective of user group. For all user groups, only in phase 2 the advantages of
Stability electrical support kicked in. Results obtained in traffic conditions confirmed the patterns obtained in the
Muscle strength controlled setting, with as only difference a lower speed in traffic conditions, which held for both mount-
Mounting behaviour ing phases and bicycle types. Also measures of physical limitations due to low muscle strength were
shown only to be compensated for by pedal support in phase 2 and not in phase 1. Further, mounting
characteristics affected pedelec stability in phase 1 and not in phase 2. Higher stability was associated
with a) starting while seated and b) using the pedal to push off. Although, these mounting characteris-
tics were confounded with age, gender, and muscle strength, the pattern of results still suggest certain
mounting techniques to be more beneficial for pedelecs. The results further illustrate the importance
of a deeper understanding of the interactions of bicycle types and user groups on critical manoeuvres
and their potential contribution to the optimisation of pedelec design and the training of safe mounting
techniques.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction further specified in following sections (See 2.2) to comply with EU


regulations (Hendriksen et al., 2008; Van Boggelen et al., 2013).
In the Netherlands, the group of elderly cyclists (aged 65+) is With rising age, specific tasks in traffic become increasingly diffi-
growing exponentially. Since life-expectancy continues to increase, cult (e.g., Brouwer and Davidse, 2002). These physical limitations
and the first Dutch baby-boomers have already hit 65, prognoses may be one of the contributing factors for the rising numbers of seri-
are that the proportion of elderly in the population will rise to ous injuries, especially among elderly cyclists (Twisk et al., 2015),
24% in 2030 (Hoeymans et al., 2014). This age group also produces while the use of pedelecs might expose them to even higher injury
the largest number of customers for electrically assisted bicycles risks. Based on a case control study of cyclists seeking medical treat-
with pedal support up to 25 km/h, hereafter called pedelecs and ment in hospital emergency rooms, Schepers et al. (2014) found
injury risks of pedelecs to be higher than that of conventional
bicycles, even after controlling for differences in age, gender and
mileage. The high injury risks of elderly cyclists are mainly associ-
∗ Corresponding author.
ated with single bicycle crashes, accidents not involving other road
E-mail addresses: Divera.Twisk@SWOV.nl (D.A.M. Twisk),
Gord.Lovegrove@ubc.ca (G.R. Lovegrove). users (Weijermars et al., 2015, 2016). For elderly, 20–25% of injuries

http://dx.doi.org/10.1016/j.aap.2017.01.004
0001-4575/© 2017 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Twisk, D.A.M., et al., An experiment on rider stability while mounting: Comparing middle-aged and
elderly cyclists on pedelecs and conventional bicycles. Accid. Anal. Prev. (2017), http://dx.doi.org/10.1016/j.aap.2017.01.004
G Model
AAP-4383; No. of Pages 8 ARTICLE IN PRESS
2 D.A.M. Twisk et al. / Accident Analysis and Prevention xxx (2017) xxx–xxx

were reported to have occurred while mounting or dismounting In addition to age, gender, and muscle strength, the way in which
the bicycle. For younger age groups this injury cause was rarely cyclists mount their bicycle may also affect their stability. Earlier
mentioned (Ormel et al., 2008). Also Schepers et al. (2014) found studies have identified two dominant mounting methods whose
relatively a high proportion of injuries on pedelecs being due to use differs by age group and gender (Dubbeldam et al., 2016). So
falls while mounting or dismounting (17%) and only 8% on conven- far, little is known about the impact of these different methods on
tional bicycles. Similar patterns were reported for injury accidents stability in relation to age, gender and bicycle type.
among senior cyclists in Sweden (Rolfsman et al., 2012). A deeper understanding of these issues is deemed necessary
Not only accident studies show mounting to be a critical phase when considering interventions related to road infrastructure,
with a high risk of falling. Also the relationship between speed bicycle design, and training. For road infrastructure it is relevant
and stability points into that direction. Up to a speed of approx- for allowing sufficient time and space for elderly cyclists to com-
imately 12 km/h bicycles are not self-stabilizing, requiring rider plete essential manoeuvres such as crossing streets and negotiating
actions to keep balance by steering into the fall. After reaching that light controlled intersections. Thus, information from this type of
critical speed, little rider actions are required to maintain balance study should be fed into micro simulation models, and used in road
(Kooijman et al., 2011; Meijaard et al., 2007; Schwab et al., 2012). infrastructure design. For bicycle design, the results may be used
Further, Dubbeldam’s study (2016) showed mounting to be more to set out a strategy to optimize the electrical support system for
safety critical than dismounting. That still more injuries occur while different groups of cyclists or even for each individual. For training,
dismounting (Kruijer et al., 2012) may be due to other interfering the results can be used to teach specific strategies depending on a
factors such as emergency braking or other reasons for imbalance person’s age, or even to train from the onset of cycling, strategies
requiring an unplanned dismount (Boele et al., in press). Also lab- that are beneficial for a life time. To contribute to these understand-
oratory studies comparing the response of elderly and younger ings the following research questions were addressed in this study,
cyclist to a sudden disturbance showed these situations to be more thereby comparing conventional bicycles and pedelecs, males and
demanding for elderly cyclists to remain stable (Bulsink et al., In females, and middle aged and elderly cyclists:
press). Q1: To what extent is mounting stability associated with rider and
Studies so far have mainly looked at conventional bicycles and bicycle characteristics?
not at pedelecs. If stability in the mounting phase on conven- Q2: To what extent does stability differ in a natural setting from a
tional bicycles is lower In the elderly than in younger cyclists, controlled setting?
pedelecs may be beneficial for elderly cyclists, as was suggested by Q3: Does muscle strength play a role in achieving stability?
Dubbeldam et al. (2016). Note that the higher risks as identified by Q4: To what extent has mounting method an additional effect on
Schepers et al. (2014) may not be due to the higher risks of pedelecs stability?
but to selective recruitment. That is, elderly who are more vulner- The research questions 1, 3 and 4 lend themselves to be tested in
able and in poor physical condition may prefer to use pedelecs controlled experimental settings. Research question 2 needs to be
rather than conventional bicycles, while elderly in good physical answered by comparing performance in controlled conditions with
conditions may still prefer to use conventional bicycles. Further, that in natural conditions, in this case starting from a traffic light.
although the slightly higher speeds of pedelecs may increase crash To overcome the earlier mentioned limitations due to selective
rates and injuries, the benefits may come from being able to fit into recruitment, each participant rode both bicycle types in counter
the ‘normal’ speed patterns of other cyclists (Vlakveld et al., 2015), balanced order.
and from being more stable in the mounting phase, since the pedal
support may help them to accelerate more quickly. However, one
problem often mentioned by users is the higher weight of pedelecs
(Franke-Bartholdt et al.„ 2014). Pedelecs are both 50% (9 kg) heav- 2. Method
ier than conventional bicycles (Van Boggelen et al., 2013), as well as
top-heavy because of battery placement (usually on the rear rack). 2.1. Participants
The higher weight requires a higher physical effort to acquire stable
riding speeds, whereas the higher centre of gravity may destabilize Sixty-one adults (30 women, 31 men) were recruited through
the rider-bicycle combination in the mounting phase. The present letters sent to approximately one thousand addresses in the area
study addresses the question whether the pedelec’s pedal support of The Hague. Inclusion criteria were: either between 30 and 45
in combination with higher weight and centre of gravity have a net years of age (middle adulthood) or aged 65 years or above (elderly),
positive effect on acceleration speed in the mounting phase, and cycling regularly, being in good health, and being able to ride a
whether that differs by user group, such as elderly and middle-aged conventional bicycle. Experience with any type of pedelec was not
cyclists and males and females. An additional feature that requires required. Participants received a gift voucher of 25 euros for their
further investigation is that of the generalizability of the results participation. Data from two participants were excluded from data
obtained in the controlled conditions. To explore this, the present analysis because of data collection failure, while data from two oth-
study compares mounting behaviour in controlled conditions with ers were excluded because of not meeting the inclusion criteria.
similar manoeuvres in real traffic. The physical characteristics of the middle aged group (n = 28), 39%
Positive effects of pedelecs may be expected especially for per- male, mean age 38, minimum age 30, maximum 45) and the elderly
sons with lower levels of muscle strength and physical fitness. Since (n = 29, 54% male, mean age 70, minimum 65, maximum 79) were in
acceleration is an important feature for bicycle stability, and know- line with expectations. Men compared to women were significantly
ing that acceleration requires physical effort, muscle strength and taller (M = 180 cm versus M = 166 cm; t(54) = −9.746; p < 0.001),
physical fitness might be influencing factors. In general, elderly heavier (M = 82.6 versus M = 67.5; t(53) = −5.37; p < 0.001), and
have lower muscle strength and are less fit than younger age stronger (M = 84.7 versus M = 56.7; t(54) = −5.99; p < 0.001). Men
groups. Moreover stability and other functions in older age are also and women did not differ significantly with respect to their BMI
associated with muscle strength (Schaap et al., 2013; Stel et al., (t(54) = −1.285; p = 0.204). Age groups only differed significantly on
2003). Still, little is known about the actual relationship of mus- hand grip strength, with elderly being less strong (M = 60.6) than
cle strength with stability. Therefore the present study assesses middle-aged participants (M = 78.8; F(1,54) = 11.06; p = 0.002). Fur-
the ‘fitness’ of its participants and relates that to their stability on ther, elderly cycled more frequently than middle-aged participants,
conventional bicycles and pedelecs. but the total distance per week did not differ.

Please cite this article in press as: Twisk, D.A.M., et al., An experiment on rider stability while mounting: Comparing middle-aged and
elderly cyclists on pedelecs and conventional bicycles. Accid. Anal. Prev. (2017), http://dx.doi.org/10.1016/j.aap.2017.01.004
G Model
AAP-4383; No. of Pages 8 ARTICLE IN PRESS
D.A.M. Twisk et al. / Accident Analysis and Prevention xxx (2017) xxx–xxx 3

2.2.2. Hand grip strength and questionnaires


Prior to the cycling tasks, participants performed a hand grip
strength test. Participants were asked to squeeze a hand grip
dynamo meter twice for each hand as hard as they could for 2 s. The
average score of these four trials was used in the analyses. Other
information, for instance about body height, weight, and cycling
experience, was obtained by questionnaire.

2.3. Route and task

The experiment took place in the Dutch city of Leidschendam


in summer 2013. Participants cycled a route in real traffic and
performed several tasks on a parking lot (PL), once on the conven-
tional bicycle (C) and once on the pedelec (P). The order of bicycle
type was randomized across participants. The route had a length
of approximately 3.5 km and consisted mostly of separate bicycle
paths and roads in residential areas. See for more detail about the
route Vlakveld et al. (2015). After each trip participants carried out
four tasks in an empty outdoor parking lot, one of which was a
mounting/acceleration task. For the current study, only the data
from this mounting/acceleration task was used.
For the comparison of the information obtained at the parking
lot, also the mounting behaviour at a traffic light (TL) was analysed.
For the mounting/acceleration task in the parking lot, participants
were instructed as follows: ‘On my signal (whistle), mount and
accelerate as fast as possible to reach a speed of 17 km/h, then brake
Fig. 1. The Pedelec, along with the instruments that the participants carried. to stand still (quickly but safely) ending with one or both feet on the
ground’. If this task was not executed according to instructions, par-
ticipants were asked to repeat the task, with a maximum of three
attempts. Only data from the last attempt were included in the anal-
yses. No instructions were given for the starting task at the traffic
2.2. Materials light. When mounting or dismounting, participants were advised
not to use the “Cowboy” method (swing their leg over the rear rack),
2.2.1. Bicycles and instrumentation because of the presence of the container holding the laptop on the
Two identical instrumented bicycles were used. A conventional bicycle rack.
bicycle, type Batavus Socorro, 2012 step through frame model, fit-
ted with 21 gears and a weight of 16 kg. A pedelec, type Batavus 2.4. Variables
Socorro Easy, 2012 step through frame model, with electrical pedal
support up to 25 km/h, a rear axis mounted motor, a battery above 2.4.1. Stability in two task phases
the rear wheel, four power settings, 21 gears, and a weight of To facilitate data analysis of the stability constructs, three mark-
27,4 kg. The pedelec’s pedal support was regulated by a torque ers were placed on the observed video footage, to mark two phases.
sensor, supplying a level of power in relation to the power sup- Phase 1, starting the moment a participant started moving in reac-
plied by the cyclist and was in accordance with EU legislation tion to the whistle or green light and ending the moment the
for vehicles category L1e-A (Bike Europe november, 2014). Sup- speedometer recorded a speed equal to 6 km/h. This phase corre-
port is given immediately, but if too little power is exerted by sponds with Dubbeldam’s observation that at the speed of 5 km/hr
the rider than pedal support will be relatively weak as well. This cyclists have completed mounting (2016). Phase 2 started after
specific pedelec was used in the study, since two independent con- phase 1, ending the moment in time the speedometer recorded a
sumer tests reported its pedal support system being superior in speed equal to 10 km/h. This corresponds with Dubbeldam’s phase
its class, highly intuitive and creating a natural cycling experience of ‘harmonized’ cycling and ends just before the bicycle becomes
(Consumentenbond, 2016; Fietsersbond, 2014). Further, its design self-stabilizing at a speed of about 12 km/h (Schwab et al., 2012). In
was identical to the conventional bicycle and in terms of stability this study ‘stability’ was defined as the time – in seconds- needed
the position of motor and battery the best option available in 2014. to reach these target speeds. The longer times were taken to be
Of the pedelec’s four support settings – no support, low support, associated with lower levels of stability.
normal support or high support – only normal support was used
in the experiment. On the pedelec, participants were instructed 2.4.2. Task conditions
not to change power settings or gears. On the conventional bike, There were two task contexts: an experimental controlled con-
participants were allowed to change gears. text in which participants were instructed to mount at the parking
In cooperation with Delft University of Technology, both bicy- lot in comparison to a natural unobtrusive mounting behaviour at
cles were equipped with a speedometer, GPS and rotation sensor, a traffic light. These mounting procedures were observed for each
tri-axis accelerometer, steering angle sensor (potentiometer), and participant on both bicycle types, namely a conventional bicycle
a camera facing the cyclist (see Fig. 1). To record data, a small and a pedelec. This resulted in four experimental conditions. All
laptop was placed in a case on the rear bicycle rack. On the partici- participants took part in each experimental condition.
pant’s bicycle helmet a Go Pro camera was attached, to record video
footage from the cyclist’s point of view. In the current study, only 2.4.3. Classification of mounting methods
the video footage and speedometer data were used for analyses. From a sample of the video footage, five general mounting meth-
For more detailed information see Vlakveld et al. (2015). ods (MM) were identified that were defined by three mounting

Please cite this article in press as: Twisk, D.A.M., et al., An experiment on rider stability while mounting: Comparing middle-aged and
elderly cyclists on pedelecs and conventional bicycles. Accid. Anal. Prev. (2017), http://dx.doi.org/10.1016/j.aap.2017.01.004
G Model
AAP-4383; No. of Pages 8 ARTICLE IN PRESS
4 D.A.M. Twisk et al. / Accident Analysis and Prevention xxx (2017) xxx–xxx

Table 1
Mounting Methods (MM) & Associated Characteristics (MC).

Method (MM)

1. Stand Pedal 2. Sit Pedal 3. Stand Steps 4. Sit Steps 5. Cross


(A) Body Position stand sit stand sit stand
Characteristics (MC)
(B) Start pedal pedal steps steps pedal/steps
(C)Through Frame before before before before after

characteristics (MC): (A) body position, (B) start, and (C) through 3. Results
frame. The cyclists’ body position was either ‘stand’ (on the ground)
or ‘sit’ (on the saddle). Then the cyclist would start to gain speed by 3.1. Stability, age, gender and bicycle type
using the pedal or by using steps. ‘Pedal’ means that the pedal crank
is placed in a 45 degrees up from horizontal position, and the par- For the controlled condition, mean scores and standard devia-
ticipant gains speed by pushing the pedal down. Participants used tions for stability are presented in Table 2. Note that the standard
‘steps’ if they gained speed by pushing off from the ground (once deviations in the elderly are larger than in the middle-aged group,
or several times) with the foot not placed on the pedal. Thirdly, the especially in the elderly female group. A repeated measures ANOVA
cyclist would move one leg to the opposite side of the bike, through was carried out on the controlled condition with bicycle type as
the frame, either before or after initiating movement. Using these a within-subjects factor, with age group and gender as between-
three factors, we were able to uniquely label each event as one of subjects factors, and with stability in phases 1 and 2 as the
the five mounting methods (see Table 1). dependent variable. The results show for phase 1 a main effect of
Bicycle type F (1,52) = 7.26, p = 0.01, eta = 0.123, of the pedelec being
significantly slower. Interactions of bicycle type by age groups, or
by gender were all not significant, suggesting no specific effects for
these groups. Further as expected there was a main effect of age,
2.5. Research questions and statistical analyses elderly being slower than the middle aged group and males being
faster than females. Also in phase 2 the main effect of bicycle type
To test the relationships of individual rider characteristics on sta- was significant, but now pedelecs were faster than conventional
bility in relation to bicycle type and mounting phase, a repeated bicycles F (1,52) = 17.00, p = 0.00. Similar to phase 1, there was a
measures ANOVA was carried out with bicycle type as within sub- main effect of age F (1,52) = 7.57 p = 0.008 with a lower stability
ject variable, and age and gender as grouping variable. Mauchli’s for elderly, and a main effect of gender with a lower stability for
test of the spherity was not applied as this study design only females (F (1,52) = 10.22 p = 0.002). None of the interactions with
included two conditions and two groups. Corrections were applied bicycle type were significant indicating that there were no differ-
where necessary based on Levene’s Test for Equality of Variances. ences in the effect of the bicycle types dependent on age or gender.
Violations of the normality assumption were identified, but ignored
since ANOVA is generally considered a robust test in cases in which
there are about equal numbers of participants in each experimental 3.2. Natural conditions
group (Field, 2009), as is the case in the current study.
To test the additional impact of task context, repeated measures To assess whether patterns in the natural condition differed
ANOVA were carried out with ‘context’ as an additional within sub- from patterns obtained from the task in the parking lot, an ANOVA
ject variable, mainly to explore the presence of a main effect of with repeated measures on bicycle type and context was carried
context, and the presence of additional significant interactions of out, with age and gender as grouping variable. The results showed
context with age group, gender, and bicycle type. a significant main effect of context F(1,37) = 32,29, p = 0.00 in phase
To assess the additional impact of mounting methods and char- 1, with lower stability at the traffic light than in the parking lot.
acteristics, hierarchical linear regressions were carried out entering None of the interactions were significant. Also in phase 2, only con-
variables as predictors for stability, in the following order: 1) actual text was significant (F (1,37) = 19,624 p = 0.000, eta = 0.347) with
age and gender, 2) body position, start, and through frame result- lower stability at the traffic light than in the parking lot. No inter-
ing in four separate analyses (mounting phase X bicycle type). Hand actions were found of context with bicycle type, gender and age.
grip was excluded because of its high correlation with gender and This implies that although being generally less stable in real traf-
age. fic, the behaviour patterns in real traffic by age group, gender and

Table 2
Mean and Standard deviations of stability by gender, age, bicycle type and mounting phase (parking lot).

Conventional Pedelec Differb


a a
Mean (sec) SD Mean (sec) SD

Middle-aged 1.58 0.22 1.88 0.47 −0.30


Female
Elderly 2.14 0.71 2.50 0.79 −0.36
Phase 1
Middle-aged 1.42 0.40 1.45 0.20 −0.03
Male
Elderly 1.72 0.46 1.91 0.56 −0.19

Middle-aged 1.26 0.29 1.11 0.65 0.15


Female
Elderly 1.83 0.79 1.27 0.96 0.56
Phase 2
Middle-aged 0.91 0.24 0.67 0.31 0.24
Male
Elderly 1.32 0.45 0.93 0.28 0.39
a
Lower values indicate a higher stability.
b
Negative values indicate a lower stability on the pedelec than on the conventional bicycle.

Please cite this article in press as: Twisk, D.A.M., et al., An experiment on rider stability while mounting: Comparing middle-aged and
elderly cyclists on pedelecs and conventional bicycles. Accid. Anal. Prev. (2017), http://dx.doi.org/10.1016/j.aap.2017.01.004
G Model
AAP-4383; No. of Pages 8 ARTICLE IN PRESS
D.A.M. Twisk et al. / Accident Analysis and Prevention xxx (2017) xxx–xxx 5

Table 3
Correlations of hand grip strength with age, gender, and stability.

2 3 4 5 6 7 8 9 10 11

1.Age 0.13 −0.44** 0.40** 0.42** 0.41** 0.43** 0.48** 0.23 0.41** 0.18
2. Gender 1 0.63** −0.07 −0.30* −0.22 −0.32* −0.19 −0.10 −0.34* −0.28*
3. Handgrip 1 −0.25 −0.41** −0.37** −0.48** −0.40** −0.15 −0.50** −0.22
4. Conventional at light Phase 1 1 0.39** 0.36* 0.40** 0.50** 0.22 0.39** 0.26
5. Conventional at light Phase 2 1 0.53** 0.61** 0.69** 0.40** 0.63** 0.45**
6. Conventional in Parking lot Phase 1 1 0.21 0.31* 0.18 0.49** 0.11
7. Conventional in Parking lot Phase 2 1 0.58** 0.19 0.52** 0.50**
8. Pedelec at light Phase 1 1 0.25 0.64** 0.25
9. Pedelec at light Phase 2 1 0.16 0.14
10. Pedelec in Parking lot Phase 1 1 0.16
11. Pedelec in Parking lot Phase 2 1

Figures in bold are not statistically significant.


*
Correlation is significant at the 0.05 level (2-tailed).
**
Correlation is significant at the 0.01 level (2-tailed).

bicycle type did not significantly differ from those in the parking not reach 100 percent, this means that our analysis of differences
lot. as a result of bike type may in part be confounded by differences in
chosen mounting method.
3.3. The association of hand grip strength with age, gender, and As shown in previous analyses, also mounting methods and
stability characteristics were confounded with age and gender. Further,
ANOVAs showed main effects for age and gender on accelera-
To assess the impact of hand grip strength, Table 3 presents tion in both phases. To assess the additional impact of mounting
the zero-order correlations of the predictors age, gender and hand- method on acceleration, a hierarchical regression was carried out
grip with the dependent variable stability (variables 6 through 11). with age, gender and three mounting characteristics as predictors −
Note that handgrip strength is highly correlated with age and gen- all with two categories. To avoid collinearity, muscle strength was
der. Furthermore, hand grip strength has an inverse relationship not entered in these analyses. Stability in phases I and 2 were the
with stability, suggesting that the stronger participants are, the dependent variables. The analyses were further carried out sepa-
higher their balance is. Irrespective of context, for pedelecs, muscle rately for the two bicycle types. Age and gender were entered first,
strength loses its impact in phase 2. This provides further support followed by the three mounting characteristics. All correlations
for the hypothesis that for participants with low muscle strength, between the predictors were below 0.80. As shown in Table 6 the
pedelecs may be beneficial, but only in the second phase of mount- variance inflated factor (VIF) was below 10 (see also Field, p. 224),
ing (i.e. that pedelecs may be worse for them in the first phase, as average VIF were not substantially greater than 1, and tolerance
discussed in Section 4 below). was above 0.1.
The Durbin-Watson statistic showed that the assumption of
independent errors is tenable, that it is not less than 1 and not
3.4. Additional impact of mounting methods
greater than 3, with the best being close to 2. The explained variance
(adjusted R 2 ) varies by bicycle type and mounting phase, being
Table 4 presents an overview of the percentage of participants
the highest for the conventional bicycle phase 2 (0.51), meaning
within each age and gender group for the five mounting method
that 51% of the variance in ‘stability’ is shared with the four pre-
and the three mounting characteristics. The results confirm ear-
dictors, and the lowest for the pedelec phase 2 for which it almost
lier findings showing mounting methods and characteristics to be
does not matter who mounted it and in what manner. Also, the
confounded with age and gender. Note further that elderly males
Standardized Beta of mounting characteristics shows an almost
are more similar to middle-aged males than elderly females are to
inverse relationship with bicycle type and mounting phase. Mount-
middle-aged females. In fact, middle-aged females are more sim-
ing characteristics have almost no relationship with stability for a
ilar to middle-aged man, than to elderly females. Elderly women
conventional bicycle in phase 1 and for a pedelec in phase 2. Con-
stand out. Only members of this group uses the cross style, and, a
versely, mounting characteristics matter for conventional bicycles
relatively high proportion of them (46%) do so.
but only in phase 2, but for pedelecs only in phase 1. In phase 1 the
Table 5 shows mounting behaviour not to differ between bicycle
stability of pedelecs is higher when the bicycle is used by a younger
types, since similarity is high >60%. However, since similarity did

Table 4
Percentage of mounting method and characteristics on conventional bicycle in controlled conditions.

Middle-aged Elderly

Female Male Female Male

Method% 1.Stand pedal 52.9 9.1 23.1 13.3


2.Sit pedal 29.4 90.9 7.7 80.0
3.Stand steps 17.6 23.1 6.7
4. Sit steps 46.2
5. Cross

Stand 70.6 9.1 92.3 20


Body Position
Sit 33.3 90.9 7.7 80
Pedal 82.4 100 38.5 93
Characteristics % Start
Steps 17.6 61.5 6.7
Before 100 100 53.8 100
Through Frame
After 46.2

Please cite this article in press as: Twisk, D.A.M., et al., An experiment on rider stability while mounting: Comparing middle-aged and
elderly cyclists on pedelecs and conventional bicycles. Accid. Anal. Prev. (2017), http://dx.doi.org/10.1016/j.aap.2017.01.004
G Model
AAP-4383; No. of Pages 8 ARTICLE IN PRESS
6 D.A.M. Twisk et al. / Accident Analysis and Prevention xxx (2017) xxx–xxx

Table 5
Correspondence in mounting methods in percentages between pedelec and conventional bicycle in controlled conditions.

Pedelec

Stand Pedal Sit Pedal Stand Steps Sit Steps Cross

Stand Pedal 60.0 13.3 13.3 0.0 13.3


Sit Pedal 7.1 85.7 3.6 3.6 0.0
Conventional Stand Steps 14.3 0.0 85.7 0.0 0.0
Sit Steps 0.0 0.0 0.0 0.0 0.0
Cross 16.7 0.0 0.0 0.0 83.3

person, and in terms of mounting characteristic if a person is seated, stability on pedelecs in phase 1 and higher stability in phase 2. This
uses the pedal to push off, or has moved through the frame after indicates that the results obtained in the controlled conditions may
the start of the mounting maneuver. be generalized to real traffic situations. Note that, these negative
effects on stability for pedelecs in phase 1 may even be greater in
4. Discussion real traffic when cyclists are not pushed to go as fast as they can.
As hypothesized muscle strength was shown to play a role in
The study addressed factors affecting the mounting stability stability. On a conventional bicycle, low muscle strength was asso-
of pedelecs in comparison to conventional bicycles for four user ciated with lower stability. This shortcoming was more or less
groups: elderly versus middle-aged cyclists, and males versus eliminated when a pedelec was used, but again only in phase 2.
females. Based on previous studies (Dubbeldam et al., 2016; In phase 1, stability on a pedelec was even lower for persons with
Kooijman et al., 2011; Straathof, 2014), two phases were identi- low muscle strength; thus increasing fall risk for this user group
fied in the mounting manoeuvre. Phase 1 involving the transference rather than decreasing it.
from ground support to balance, which starts at the first movement The questions of whether mounting characteristics differed
until a speed of 6 km/h has been reached. Phase 2 involving the by bicycle type and whether specific characteristics were more
transition from balance to harmonized cycling, which is the period beneficial than others for balance, first of all showed that the
between 6 km/h and 10 km/h. It is assumed the longer these phases use of mounting methods was confounded with age and gender.
last, the lower the stability is. At speeds higher than 10 km/h, the Specifically, elderly women differed from the other groups. Their
bicycle becomes self-stabilising, requiring little rider effort to keep mounting method – using the outside foot to push down the near-
balance (Schwab et al., 2012), and the mounting manoeuvres have est pedal, and to accelerate by pushing form the ground – was rarely
been completed (Dubbeldam et al., 2016). None of these previous used in any of the other user groups. As suggested by Dubbeldam
studies however have compared mounting method and stability et al. (2016), physical characteristics related to age and gender may
on pedelecs with those on conventional bicycles, nor studied these result in specific mounting techniques to be preferred over others.
factors in natural traffic conditions In addition to these earlier mounting studies (Dubbeldam et al.,
The present study showed the pedelec to be less stable in phase 1 2016; Straathof, 2014), the present study found that participants
than the conventional bicycle. This was the case for both age groups used almost identical methods whether mounting a pedelec or a
and for males and females alike. In phase 2, however, the pedelec conventional bicycle. Thus personal habits rather than bicycle type
had a better stability than the conventional bicycle, again irrespec- may determine their use. Since the pedelec’s pedal support system
tive of rider age or gender. The comparison with stability in real responds to the force exerted on the pedals, mounting strategies
traffic conditions showed that speeds were generally lower in real that succeed in creating a strong force are advantageous for sta-
traffic than in controlled conditions. In the latter, cyclists were stim- bility overcoming the disadvantages of the bicycle’s higher weight.
ulated to go as fast as they could, while in the former they could This may also explain why persons with high muscle strength have
set their own pace. Still, the general patterns as observed in the an advantage and also mounting strategies that exert force by using
controlled conditions were still present in real traffic: Thus a lower one’s body weight – e.g. pushing the pedal down by standing on it

Table 6
The additional effect of mounting characteristic (excluding hand grip strength).

Criterion Statistically significant Predictorsa Adjusted R2 Durbin-Watson Standardized Beta Tolerance VIF

Total 0.21 1.97


Conventional phase 1 Age 0.44 0.98 1.02
Gender (female = 1, male = 2) −0.28 0.98 1.02

Total 0.51 2.16


Age 0.29 0.84 1.19
Conventional phase 2 Gender (female = 1, male = 2) n.s.
Body position (stand = 1, sit = 2) −0.32 0.50 2.0
Start (pedal = 1, steps = 2) 0.39 0.64 1.6

Total 0.46 1.89


Age 0.43 0.88 1.14
Gender (female = 1, male = 2) n.s.
Pedelec phase 1
Body position (stand = 1, sit = 2) −0.24 0.46 2.15
Start (pedal = 1, steps = 2) 0.37 0.70 1.43
Through frame (1 = after, 2 = before) 0.25 0.75 1.33

Total 0.08 1.50


Pedelec phase 2 Gender (female = 1, male = 2) −0.37 0.84 1.19
Start (pedal = 1, steps = 2) n.s. 0.84 1.19
a
Predictors entered in the model were in the order: age, gender, body position, start, and through frame. Only statistically significant predictors and Beta’s (p > 0.05) are
presented.

Please cite this article in press as: Twisk, D.A.M., et al., An experiment on rider stability while mounting: Comparing middle-aged and
elderly cyclists on pedelecs and conventional bicycles. Accid. Anal. Prev. (2017), http://dx.doi.org/10.1016/j.aap.2017.01.004
G Model
AAP-4383; No. of Pages 8 ARTICLE IN PRESS
D.A.M. Twisk et al. / Accident Analysis and Prevention xxx (2017) xxx–xxx 7

with one’s full weight, while it is in its top position and moving indicator, or that it should be supplemented by indicators such as
through the frame after mounting – have a better balance in phase steering angle and steering reversal rates. Further, task conditions
1 on a pedelec. These strategies make better use of the characteris- should be elaborated upon to include dismounting and other safety
tics of the pedelec. The fact that participants used the same strategy critical conditions. In-depth accident analyses should be used to
further suggests that mounting is habitual behaviour, coming from identify these safety critical situations (Boele et al., in press). At
hours of deliberate practice (Ericsson et al., 2007), and resulting in present, the range of testing conditions will be severely restricted
automatic routines, not requiring any attention and being suscepti- for ethical reasons, since in all these conditions cyclists need to be
ble to errors (Shriffrin and Schneider, 1977). Once automated such protected against injuries due to actual falls and collisions. So far,
routines can only be modified with great effort (Rasmussen, 1985). such safe testing conditions – similar to driving simulators – are
not yet available for cycling.
4.1. Implications of the study
5. Conclusion
To reap the benefits from pedelecs and to reduce the risk of
falling, the study has provided some deeper understanding of the Pedelecs were shown to be less stable in the early stage of
interaction of different user groups with this technically advanced mounting (phase 1) than conventional bicycles, irrespective of
bicycle type. Further study is needed to test the generalisability user group. For all user groups, only in phase 2 the advantages of
of the findings. Still even at this stage of development, the results electrical support kicked in. Results obtained in traffic conditions
of the present study have implications for the design of pedelecs, confirmed that the patterns observed in the controlled setting also
the lay-out of areas where cyclists frequently mount or dismount, held for both mounting phases and bicycle types, with the only dif-
the training of bicycle competencies, and for the tests of the man- ference being a lower observed speed in traffic conditions. Also,
machine interphase. measures of physical limitations due to low muscle strength were
For pedelec design the most important finding is that of poor shown only to be compensated for by pedal support in phase 2
balance in the first phase of mounting, up to a speed of 6 km/h. This and not in phase 1. Mounting characteristics affected pedelec sta-
calls for developing pedelecs that are at least equally easy to mount bility in phase 1. Higher stability was associated with a) starting
as conventional bicycles, or even better that provide support in such while seated and b) using the pedal to push off. No such benefits
a manner that it improves balance in the first stage of mounting, were observed for phase 2. Although these mounting characteris-
especially that of elderly and cyclists with low muscle strength. The tics were confounded with age, gender, and muscle strength, the
‘Sophie’ bicycle (Dubbeldam et al., In press) (see) is a first attempt pattern of results still suggest certain mounting techniques to be
towards the development of such a pedelec. more beneficial for pedelecs. These results further illustrate the
For infrastructure, the findings provide input for guidelines con- importance of a deeper understanding of the interactions of bicycle
cerning locations where cyclists frequently mount and dismount, types and user groups on critical manoeuvres and their potential
such as at intersections and crossings. The instability in the different contribution to the optimisation of pedelec design and the training
phases of mounting – irrespective of bicycle types – demands suffi- strategies for safe mounting techniques.
cient path width and a level surface free of obstacles. The observed
differences in user groups and bicycle types are relevant for use Acknowledgements
in design tools such as micro-simulations models (Morsink et al.,
2016). The incorporation of this information in such design tools This study was facilitated by the Dutch Ministry of Infrastruc-
ensures that future road designs are accommodating the needs ture and Environment. The authors would like to thank Dr. Arend
of the growing population of elderly bicyclists and the increasing Schwab from Delft University of Technology for designing and
popularity of pedelecs. building the instrumented bicycles.
For training, the results of this study suggest that some mount-
ing characteristics are more beneficial for stability than others, and References
that cyclists do not modify their mounting method depending on
the type of bicycle. Note though that due to technical limitations, Bike Europe. november, 2014. All you need to know on. . . EU regulations for:
e-bikes pedelecs and speedpedelecs. Retrieved 19-10-2016, 2016, from http://
the conclusions on mounting methods of elderly male cyclists may
www.raivereniging.nl/ecm/?id=workspace://SpacesStore/4ee4a9c0-3b5a-
be the least valid in this study, since the ‘cowboy’ mounting method 48f7-9346-ce670f4a0e43.
could not be used, and is therefore absent in this study. Straathof Boele, M., van Duijvenvoorde, K., Doumen, M.J.A., Louwerse, R., Duivenvoorden,
(2014) however observed that 15% of participants used the cow- C.W.A.E., Davidse, R., in press. Crashes involving cyclists aged 50 and over in
the Netherlands: An in-depth study. Accident Analysis & Prevention: Special
boy method, all of whom where elderly men. The differences in Issue Cycling Safety.
mounting characteristics need further exploration, for instance by Brouwer, W.H., Davidse, R.J., 2002. Oudere verkeersdeelnemers. In: Schroots, J.J.F.
analysing mounting techniques in Naturalistic cycling studies. If (Ed.), Handboek psychologie van de volwassen ontwikkeling en veroudering.
Van Gorcum, Assen, pp. 505–531.
future studies confirm that ‘mounting method’ is a habit rather than Bulsink, V.E., Kiewiet, H., van de Belt, D., Maarten, B., Bart, K., in press. Cycling
a routine that can easily modified, then it needs to be considered to strategies of young and older cyclists. Human Movement Science.
train routines to young cyclists that are beneficial for a life time and Consumentenbond, 2016. Consumer Test Batavus Socorro Consumentenbond,
Retrieved 19/10, 2016.
that are safe in a wide range of conditions. However, if mounting Dubbeldam, R., Baten, C.T.M., Straathof, P.T.C., Buurke, J.H., Rietman, J.S., 2016. The
characteristics are easily modified at any age, then training might different ways to get on and off a bicycle for young and old. Saf. Sci. Spec. Issue
be used to assist cyclists in using the most optimal mounting tech- Artic.: Cycl. Saf.
Dubbeldam, R., Baten, C., Buurke, J.H., Rietman, J.S., in press. SOFIE, a bicycle that
niques for different bicycle types, such as using the pedal when
supports older cyclists? Accident Analysis and Prevention SI:Cycling Safety.
mounting a pedelec. Ericsson, A.K., Prietula, M.J., Cokely, E.T., 2007. The Making of an Expert. Harvard
For usability testing, our method is one of the first to evaluate Business Review Article, July.
Field, A., 2009. Discovering Statistics Using SPSS, 3rd ed. Sage publications Ltd,
the rider – cyclist performance for different task conditions, bicycle
London, UK.
types and user groups. Note though that the definition of ‘stability’ Fietsersbond, 2014 12-08-2014. Batavus soccoro easy. Retrieved 19- 10-2016.
originated from previous studies about rider and vehicle character- Franke-Bartholdt, L., Petzoldt, T., Schwanitz, S., Kühn, M., Gehlert, T., 2014. Pedelec
istics (Dubbeldam et al., 2016; Kooijman et al., 2011; Schwab et al., Naturalistic Cycling Study, GDV. CDV.
Hendriksen, I., Engbers, L., Schrijver, J., Gijlswijk, R.v., Weltevreden, J.B., Wilting,
2012), but only used bicycle speed as an indicator of stability. Fur- J.B., 2008. Elektrisch Fietsen: Marktonderzoek en verkenning
ther research should validate whether this is a sufficiently sensitive toekomstmogelijkheden. TNO Kwaliteit van leven, Leiden.

Please cite this article in press as: Twisk, D.A.M., et al., An experiment on rider stability while mounting: Comparing middle-aged and
elderly cyclists on pedelecs and conventional bicycles. Accid. Anal. Prev. (2017), http://dx.doi.org/10.1016/j.aap.2017.01.004
G Model
AAP-4383; No. of Pages 8 ARTICLE IN PRESS
8 D.A.M. Twisk et al. / Accident Analysis and Prevention xxx (2017) xxx–xxx

Hoeymans, N., Van Loon, A.J.M., van den Berg, M., Harbers, M.M., Hilderink, H.B.M., Schepers, J.P., Fishman, E., den Hertog, P., Klein Wolt, K., Schwab, A.L., 2014. The
van Oers, J.A.M., Schoemaker, C.G., 2014. Een gezonder Nederland. safety of electrically assisted bicycles compared to classic bicycles. Accid. Anal.
Kernboodschappen van de Volksgezondheid Toekomst Verkenning 2014. Prev. 73 (0), 174–180.
Rijksinstituut voor Volkdsgezondheid en Milieu, Bilthoven. Schwab, A.L., Meijaard, J.P., Kooijman, J.D.G., 2012. Lateral dynamics of a bicycle
Kooijman, J.D.G., Meijaard, J.P., Papadopoulos, J.M., Ruina, A., Schwab, A.L., 2011. A with a passive rider model: stability and controllability. Veh. Syst. Dyn. 50, 8.
bicycle can be self-stable without gyroscopic or caster effects, paper and Shriffrin, R.M., Schneider, W., 1977. Controlled and automatic information
supporting material. Science 332 (6027), 339–342 http://bicycle.tudelft.nl/ processing II. Perceptual learning, automatic attending and a general theory.
stablebicycle/. Psychol. Rev. 84 (127), 127–190.
Kruijer, H., den Hertog, P., Klein Wolt, K., Panneman, M., Sprik, E., 2012. Stel, V.S., Smit, J.H., Pluijm, S.M.F., Lips, P., 2003. Balance and mobility performance
Fietsongevallen in Nederland: een LIS vervolgonderzoek naar ongevallen met as treatable risk factors for recurrent falling in older persons. J. Clin. Epidemiol.
gewone en elektrische fietsen. VeiligheidNL, Amsterdam. 56 (7), 659–668.
Meijaard, J.P., Papadopoulos, J.M., Ruina, A., Schwab, A.L., 2007. Linearized Straathof, P., 2014. Step by Step Analysis of Bicycle Mounting and Dismounting:
dynamics equations for the balance and steer of a bicycle: a benchmark and Strategies and Kinematics: Investigating the Differences in Bicycle Mounting
review. Proc. R. Soc. A 463, 1955–1982. and Dismounting Strategies and Kinematics Between Young and Old Cyclists.
Morsink, P., Toes, E., van Overdijk, R., Dijkstra, A., 2016. Fietsmodellering: lopende http://essay.utwente.nl/65966/.
ontwikkelingen en toekomstige vragen (modeling bicycle characteristics: new Twisk, D.A.M., Davidse, R.J., Schepers, P., 2015. Challenges in reducing bicycle
developments and future issues). Paper Presented at the Nationaal casualties with high volume cycle use: lessons from The Netherlands. In:
Verkeerskundecongres. Gerike, R., Parkin, J. (Eds.), Cycling Futures: From Research into Practice
Ormel, W., Klein Wolt, K., den Hertog, P., 2008. Enkelvoudige fietsongevallen: Een Ashgate.
LIS-vervolgonderzoek Stichting Consument en Veiligheid Amsterdam. Van Boggelen, O., Van Oijen, J., Lankhuijzen, R., 2013. Feiten Over de Elektrische
Rasmussen, J., 1985. Trends in human reliability analysis. Ergonomics 28 (8), Fiets. Fietsberaad, Utrecht.
1185–1195. Vlakveld, W.P., Twisk, D., Christoph, M., Boele, M., Sikkema, R., Remy, R., et al., 2015.
Rolfsman, E., Bylund, P.O., Saveman, B.I., 2012. Single injury incidents among Speed choice and mental workload of elderly cyclists on e-bikes in simple and
pedestrians and biscyclists in Northern Sweden: safety and preventive issues. complex traffic situations: a field experiment. Accid. Anal. Prev. 74 (0), 97–106.
Saf. Sci. Monit. 16 (1). Weijermars, W., Bos, N., Stipdonk, H.L., 2015. Serious road injuries in the
Schaap, L.A., Koster, A., Visser, M., 2013. Adiposity, muscle mass, and muscle Netherlands dissected. Traffic Inj. Prev. 17 (1), 73–79.
strength in relation to functional decline in older persons. Epidemiol. Rev. 35 Weijermars, W., Bos, N., Stipdonk, H.L., 2016. Health burden of serious road
(1), 51–65. injuries in the Netherlands. Traffic Inj. Prev. 17 (8), 863–869.

Please cite this article in press as: Twisk, D.A.M., et al., An experiment on rider stability while mounting: Comparing middle-aged and
elderly cyclists on pedelecs and conventional bicycles. Accid. Anal. Prev. (2017), http://dx.doi.org/10.1016/j.aap.2017.01.004

You might also like