ArticlePDF Available

“I Don't Care About Social Control”: Why the Eye Image Effect Does Not Reduce Illegal Pedestrian Behavior in France

Authors:
“I don’t care about social control”: Why the eye image
eect does not reduce illegal pedestrian behavior in France


a Université de Strasbourg, CNRS, IPHC UMR 7178, Strasbourg, France
b Institut Universitaire de France, Paris, France
c Anthropo-Lab, ETHICS EA7446, Lille Catholic University, Lille, France
Corresponding author: Marie Pelé, Lille Catholic University, marie.pele@univ-catholille.fr
Abstract
!!!"#!!!$!!#
#!!!%!$""!
&'#  !&! ' !  !   %'# 
"!!$ ( !  %  !  &' !
#)!# "# !! %% ! *!! 
 !!  !#$ + ,-! !.  ,'-! !. !#!
% ! ' ! ! " !  "  '
 !  '   ' !#  " /'!$
0 !   ! '  % 1234
"!! #   # !#  !!$ 56 
!'6!7%#!$(7!#!
   7 ""! 7 % ! ! !!$ + %
'!!#!!#87  % 9#!!#!$
:'  !# % ,-! !.    '#   
"! !!#$ ( !! ! !'   ,/'.  !#
!#8!"!!&9&#!'!!#
;
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
1
#$ 5# "!% !! "!!  
!!%! !  /"!$ Although previous studies have
shown the effectiveness of nudges with eyes signals, these images do not appear to have a desired
beneficial effect on pedestrian crossing behavior.
Keywords<+##!!#!'#!&&#
1
27
28
29
30
31
2
1. Introduction
=1>?>3@A% '-! ""'!!%??A
"!!!)B+!1>;2*$B!
 ! !  #  !" % "!!$
C# !# % !  !  ## !
"!!!##%!!!!$
:' ! %6!##  "
%!! # /! ' "!!  !$ 0 #
D"!)EF>>>>>!*!!!%"!
G!  %!  #    ! )C! 
! !HIC+0J1>;2*$ !!
%;?A%!;FA%G!1>;@)C+01>;2*"!#
F4> "! %! ) ! "! ! +  !
H 1>;2*  2@>? #  !! G! )K!! 
H1>;2*$(!! ' #! 1>>>!
'!#!1>;>)C+01>;2*$
D!!#!"!%
    "!  1;A %  !! )C+0 1>;2*$ 0
"!!  !' !& !<  !! !  !!#
&#!!"!#)!<##!*
 !! ""   !'&      
!""!%"!%!)L#$1>;F*$ !!
& !&!    %    # ! 8! ! !
!#   ! "!   "  $ !
""!%!!&!'"%!
M
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
3
$D"%!!!
%N!!) $1>;4*$0!!%!%%
"!!!## "! #"!
!&&#)O#$1>;;*$ (!"!!"!! 
#! !  !&  !G##  !&!   
!#   ##  !% $ (! !  
!!!'/!
"!!!#!)5PQ!"9+1>>@*$
( "" % !&  "!! ' "  "!! % 
$ 0  !"8 ! % !#6 !!#! )"! %
!R! !# R !!'&* 7 "8! % "!!
!!##!8) $1>;2 $1>;2*$
D!  ! !!  7 "!! ' !!   #
 !'!$ (!'!&
!% %  !! 
!&#%"$ (  "!! !! 
! ' !N  ! %   %  
!%!! #"  )+! $1>>@*$ L!!
 !#)$#$"!!!*"
  %  % % "# ' !!#
!!!#  !% !!#! )=#  $ 1>>?*$ 0" 
!'"8!! #6%"!! '
%    % !!#$  "!! !! G!  
##%#!G!)"$1>;M 
 $ 1>;2*$ ( !  N #! % !  " N #
#!  "! # "" % # !!#!  "!
F
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
4
!!'&! )"  $ 1>;M*$ (  "!! ' !! , 
.%9#'!""#&!!&""
 " !!  !$ 0  ! "   ' 
"!!!"!!' S !!$(  !'
!%")("!$;2@?*$(
%&!"6 !  & !
 $ (! "!! ' "! ! "6    
#"!!#!!G!N"%
! )=#  $ 1>>@  ! "! ! + 
!H1>;2*$(!!!%"!!
S'%!!##%!!#'
! # !!$K#"!! '!
"'D)'!!#"!#*
!"!&%"!#$
D %!/ !&! "! !
'#&$ D " & !&!' 
7!)" $ 1>;M  $1>;4!
1>>2(P51>;;GT$1>;F*$(!7%#!
'' #!!< #! ""$ U#!
);?9M>!*&!&!!!'!!#
#!%57H$)1>;4*$C""
!'!&!%!
#!)V#$1>;2*$5#G/
8#!$
 %!  ! / "!   !& &#$
!!'!!#7"!!!!#
?
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
5
#"!$0#"%"!!'#N#!
'#!<!%'!) $1>;2 
$ 1>;2*$ 0!   8!    "  !! %
6#  !$ (  %    ' % 
#%!!#$0!%'!%'
 8! "! ' !! ' #  % 
" ' # "!  !$ (! ! ' 7! "
% % ! % )D  $ 1>;>*$ :'  
%' !&!  !! #!   ' !!#  #"!$
0 ! ' !!  #"!   !    !!
 !  ")   $ 1>;2! 1>>2 V# $
1>;2*$
!"!"
!&!  7# % !!$D"!!
#!!"D)1$;A%!!#!
!#!""F;$2AD*) $1>;21>;4
 $ 1>;M*$ (! 7   "   %  "!
"!!   !!   &!  "!!! % ! 
D"!!)  $1>;2*$0 "!!
!  %!   D ' 6!  
! )= P L 1>;M*$ K  Q"  !
"!!#!!"5D)=P
L 1>;M*$ (%    7!   N %
"  7 ! )0  $ 1>1> "  $
1>;M*!!%"!"!%"!%!
"#$
3
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
6
D !" ! %! %!
%"!!$D"!#%  !  "
!$ 0!#   % !  N ! 
!#!!#8#'"!!&
&#!#86)$;22> $1>;4O#
$1>;;V#$1>;21>;4*$D!#%"!
%!!""!-!!&$D"
 % # !!#! ! "!  '   % 
"!#)W#PC-1>>M $1>;2O#$
1>;;*$  #! ' # ! #  # !!#!
!'#!)57H$1>;4*$(!!
""!""!#
% "!! !   7! )5' P = 1>>;*$
! ! ! !  # "!!  " !%
!$Q#!!"%#!!
!!7#$D"
9!"$O
!   ' '# "!   # %  !#6
!!'&'!""!!!%"!!
&!)34$4A!"'!!#!!??$;A'
!!#*)5#$1>;?*$%"!&!
!!31$2A%!!""?>$;A%"!
&"!"!! )5#  $1>;3*$(!!" & 
% %!G! !# %
"!! ! "$ :' ! 7 ! # !  !#!
4
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
7
"#-! &!! !" ! 
")&$1>11*$
Q#!"
 & !G!  #$ 0 ! ' =!  $ )1>>3*
  "  !  7 % !  '  !#
" $ 0  %  +'! B! 
 !! '! "  "  " % &! %  !$ (
"'!"#!%7!/'!"%!
 ! % 7!  "!   $ ( !G! " 1$43
! #'#'!!%/'!$O
"#  7 % !9  /' #! ' ! % 
#!  !!  % "! 7! #  !$ (
"!%#!87!#!
%)Q!9! $1>;;*!# '!  ! !")DP
=#X1>;1*##!") '$1>;1*#%
%#B!) #"!1>;F*#
 ! )  $1>;4*$  ! nudges,
!! % !##!#  #  ' !# !
 )( P ! 1>>2*$ 0 % %  #   !
"%!'!$K#
! !"      ' %  !9
&#"!!%!G!%!
$ (!  !  8    R 
$
(!%#!!%%!!"
!!'!!!!#$0!%
@
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
8
!#!&%!"%   7    %
#"!!!#!!#6!!#!$(%!!!
!7%!#)#!#*!!#!
"!!#!$('!#!'!"<,-!!.
#,'-!!$.O"!#!'!!!#
!!#! !" # "!! '  ! '
#"$(%'#"!!'"!"$K#
'!#!!!!#'#
N#!!!&&#)$$'##
'!!#%*$
2. Materials and methods
1$;$0#!
K" !'!  ! !# %! 
!$(!" !)$1>11*S!""! 
!)D&('*!!!!"%M>"!%
!' 7 !)# %*$S!!
'!&%"!!,L%!'&#!
#Y.  ,O !# ! #" % !!#  
"!#Y. "!'!&!',!.,.,$.
=!!!!#!'!%"!
!$( 8!!'!'' #"!!)D#;*
!!'!'"!!)D#;*$(!
#!'!/'!)D#;*$K#!!!
2
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
9
% # )D#;*   9 ! ! )Q P  1>>4
UX&!1>>2*$
Figure 1: Images selected during the preliminary study (Sueur et al., 2022). The two-test photo
represent eyes sign (a) and (b), and the control photo flowers (c).
2.2. Set-up for the experiment
These images were printed on traffic signs (50 × 20 cm) and hung just above the pedestrian signal
box (Figure 2a).
The study comprises of four experimental conditions:
- Two control conditions: No sign and “flowers” sign.
- Two test conditions: “Children’s eyes” sign and “woman’s eyes” sign.
All observations were filmed by a Canon EOS60D camera on a tripod at about 1.70 meters from the
ground. The observer was positioned offset from the pedestrian crossing, to be hidden from
pedestrians and thus avoid impacting their behavior (Figure1*$
;>
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
10
Figure 2: Photo showing the flower sign at an observation site. The signs should be installed next to
the pedestrian light, ideally above but not below (a). Diagram of the observer's and camera's angle
of view on the targeted pedestrians (b).
1$M$C!%"!!
( ! %!!   "!! % !#  % 1@13F2
!  1>;2$ ( !! '    !! %
!# # !! )' ' 8   S"
!"!"!%#!*$5"!%"!!!
!   ! #! ' ! $ 0! # 
'!!!!&!'!
!"'!#$
( !#6 "! !!#! ' !     %
!#<Z !+=B!)(;*$(%
!!!'!#'-0%
# KS    )0K* %  !# Q9
")!#$"1>;?*$(!!!'!F>>
!'%!!!!$O
"'!"!'!""!!!#
!!$
Table 1: Information on observation sites.
Site Quai de Paris Nuée Bleue Université
Geographical coordinates
48.584250, 7.741250
48.584472, 7.748806
48.577667, 7.764472
;;
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
11
Number of lanes One lane
(one way)
Two lanes Four lanes
(with one for the bus)
Tramway Yes Yes No
Cycle path Yes No Yes
Average number of cars per hour (one
way)
49.9 110.9 136.3
Average pedestrian frequency per hour 109.3 154.2 148.3
(!!&"'"!$(8!'!%;M"
;?C1>;2%'!#$(!
'! % ;3   11>1>  !   !!
)'!!#!*    '/'!
!#$(!!&"%D'@$$
;1"$$)# !!*' "'! 
)$$! "!%!'* ' '!! %
"!!)$#$'&!*$
( ! % ! " !  " "  '
6 %  % M3 ! % #!$ (   S
!%%!"
!!96'!!$
C"!!!!#%!'
!!'!$L!!'!
 %"!! !!## N #!
%!  % "!! !#
!"#)K;24F*$(!'!'89
)K!*$
(%   !! ) !* ! '  %
"!%!"#)K;24F*$("!! !
%!!'!'!!#'"!#
;1
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
12
'!)$$!'!!#!''#
!!##*$ !!#!!###!
'!!!!7%#!
 $   '!  & '  ! % 
"!!'!")$$'"!&&"&#*
'     "!!$ (! %  M>4; "!!
!!!#!!;>F)M$FA*'$
1$F$Q!"
+ %# % '!  %  "!!$ (  !
!%"!!'
% !!#!$( =!9 % '!% 
"5L #)5L *'!!"$
( " '! ""   C+0  0K$ L# #!
"!!   ""   %   ! % 
S!$(  ! !!  % %
!$
1$?$!
 ! '  %  "!<  !
$
2.5.1. Behavioral variables
9 O#<'#)!!*%"!%!!#
!$(!!!#%"!
'"%'!"!'&%"
'"!%!  %!'&#$O# !
;M
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
13
S  6 %  "! !!! ' !""#   !'&
)D#M*$
9 "<each pedestrian has been coded by 1 if they have a waiting time higher than zero and 0 if
they cross directly (waiting time = 0).
9 !!-#6!<%&!%"!
%  "! #  !    %   % 
"!'!#!$
9 :!<!"!"%$(
"!     & "  "  !!# 
!!#)$1>1>*$Pedestrians have been coded by 1
if they express a behavior of hesitation and 0 if they do not.
Figure 3: Diagram of the time variables recorded and calculated for each pedestrian.
2.5.2. Environmental variables
9 Q")!#*< !#/'!!#'-! !!# 
-!!!#$
9 C!!<Z !B!+=$
;F
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
14
9 !!##<!"!%"!#
%!!#$Each pedestrian has been coded by 1 if they cross at a green light and zero if they
cross at a red light.
9 !&&#<!&&#)!!*!  '  "!-!
"%8!)!&*%
"!!!#)D#M*$
9 L!!<!"! %!! ! !"!
 "!  7# '  ! % &#  
"&#$
2.5.3. Social variables
9 5"<!"!%"!!!#"%
%%"!!'#!$K
"! #!  #" %   !" %  !
"! !! F$;> !! % Z  ! ?$13 !! %
+=4$>>!!%B!$(!!!"
#"!!!#%!)#'
1>!!#!"!*$0!!"!!/
%!!%$(%
#"!%'"!!$
9 <   %  "!   !!# #
"!!$
9 C%"<%"'N#!
  "  %  "!$ (  ' !! 8!
)!!*!&!&;!#!!&1
;?
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
15
$K8#!"  $)1>;4*%!#%
!$
9 K"<' "!! "%"!
)[U!-[+-*$ K "!!"%  &#'&#
!!'"!$
2.5.4. Individual variables
9 <)*%)D*$
-K#<!;>9!!!$D"'"!
""!F?years old is put in the 40 categories. The possible categories range from 0
for children, aged 0 to 9, to 80 for those aged 80 and above.
These variables are extracted from the videos, thanks to the free software BORIS (Behavioral
Observation Research Interactive Software v 7.9.24), which allows for temporal accuracy to the
nearest centi-second.
2.6. Inter-observer agreement
The video analyses were made over two successive periods by two different observers (LL and MJ).
The same video was thus observed several times to adjust the inter- and intra-observer results. The
agreement was checked 3 times: before, during, and after the end of observations for each variable.
An agreement rate (Elie & Colombet, 2011) of at least 80% validated the observations for the
variables of sex, hesitation, distractors, group, children, order of departure and accompanied. The
age estimate was validated using Kendall’s coefficient (W> 0.8; p-value <0.05) (Abdi, 2007). For
the different durations, a maximum difference of 500 centi-seconds was tolerated. Indeed, the
frame-by-frame analysis on the BORIS software could create mismatches in the order of a centi-
;3
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
16
second, making 100% reproducibility difficult for the time variables (i.e., waiting time and risk
taking).
The observation method and video quality did not allow us to obtain a sufficient agreement
percentage for the pedestrian gaze variable (pedestrians looking at the signal). The visibility of the
faces on the videos was not precise enough to correctly discern eye movements and directions. The
inter- and intra-observer agreement for this variable often appeared insufficient and varied between
60% and 80%. The pedestrians’ gaze data was not considered reliable enough and hence was not
used in this study.
2.7. Statistical analysis
Generalized linear mixed models (GLMM) were performed to measure the impact of different
independent explanatory variables on response variables. Different statistical families were used
depending on the variables to be explained, to respect the application of the models (normality and
homoscedasticity), as follows: quasi-Poisson, gamma, or binomial. Some data were modified to fit
in with the adapted model (Table 2). Site was included as a random effect. Several models were
tested to investigate the variables of interest in this study. The first variable to be explained is the
red-light crossing rate (PR ratio), which is the number of pedestrians who arrived at and crossed on
the red light, out of the total number of pedestrians who arrived at the red light (crossing on red and
green light). The explanatory variables are the experimental condition, number of passing cars, and
presence or absence of the tramway (Table 2). The crossing light was also tested as a response
variable for each pedestrian, to observe the potential effects of variables at the individual level. The
data was sorted to separate pedestrians who waited and those who did not, to study waiting times.
We wanted to observe the impact of the different variables on whether pedestrians waiting. Risk
taking and Hesitation are also variables of interest, and will be studied for pedestrians who crossed
on a red light only. The explanatory variables are the experimental condition, age, sex,
accompanied, distractors, group, order of departure, and children (Table 2). The group/order of
;4
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
17
departure and sex/age interactions were also tested and presented where they had a significant
effect, through the statistical model. Pairwise t-tests with Benjamini Hochberg correction or a
pairwise Wilcoxon test was also used to compare the different modalities when there was a
significant effect of the variables. All analyses were carried out in R Studio (© 2009-2021 RStudio,
PBC, version 1.4.1717). We set the statistical significance at α = 0.05.
Table 2. Laws and variables used in each model.
Model Family used Response variables Explanatory variables
(1) Quasi-Poisson Ratio PR Sign + Tram + Car
(2) Binomial Crossing light
(green or red)
Sign + Age + Sex + Accompanied + Distractors +
Group + Order of departure + Children
(3) Binomial Stop
(absence/presence of waiting time)
Sign + Age + Sex + Accompanied + Distractors +
Group + Order of departure + Children
(4) Gamma Waiting time
(pedestrians who have waited only)
Sign + Age + Sex + Accompanied + Distractors +
Group + Order of departure + Children
(5) Quasi-Poisson Risk taking (pedestrians who
crossed at red light only)
Sign + Age + Sex + Accompanied + Distractors +
Group + Order of departure + children +
Interaction Group: Order
(6) Binomial Hesitation (absence/presence) Sign + Age + Sex + Accompanied + Distractors +
Group + Order of departure + Children
3. Results
In total, 2967 pedestrians (1673 women and 1294 men) were observed arriving at the red light
(Table A.1); 2335 crossed at the red light (79%), and 632 crossed at the green light (21.3%). There
were 102 hesitations observed (3% of pedestrians) and 67 crossings conducted with at least one
child under 10 years old.
3.1. Rate of pedestrians crossing at a red light
Over all the hours of observations, the red-light crossing rate (PR ratio) for all sites combined is
0.76 ± 0.14 for the condition without a sign, 0.78 ± 0.11 for the flowers sign, 0.81 ± 0.16 for the
child’s eyes sign, and 0.80 ± 0.13 for the woman’s eyes sign. The first GLM does not show an
;@
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
18
effect of signs on the ratio PR (Table A.2). There is no significant difference in the red-light
crossing rate according to experimental conditions (child’s eyes sign: estimate = 0.014, t-value =
0.420, p = 0.678; woman’s eyes sign: estimate = 0.007, t-value = 0.201, p = 0.842, Table A.2).
However, there is a significant effect from the presence of the tramway (positive; p < 0.001) and
number of cars on the red-light crossing rate (negative; p < 0.001) (Table A.2).
3.2. Effect of variables on the color of the pedestrian light when crossing
GLM 2 shows a significant effect of the child eyes sign on the crossing light compared to no sign
(estimate = -0.311, z-value = -2.396, p = 0.017, Table A.3). Of pedestrians, 81% crossed at a red
light in the presence of the child’s eyes sign, 80% with the woman’s eyes sign, 78% with the
flowers sign, and 76% with no sign. However, the pairwise comparisons t-test did not show any
difference between the color of the light crossed in the presence of the child’s eyes sign and no sign
conditions (pairwise t-test, p = 0.11).
The percentage of crossings at the red or green light seems to be impacted by age (estimate = -
0.009, z-value = -2.627, p < 0.001, Table A.3). Of children 25% under the age of 10 (i.e.,
accompanied by their parents) crossed at red lights (pairwise t-test p < 0.05). People in their 40s
cross at red lights (86%) significantly more than 20- and 30-year-old people (77% and 76%,
respectively; pairwise t-test p < 0.05). Approximately 81% of adults from 50 years old crossed at a
red light..
There is an even stronger effect on gender (estimate = -0.474, z-value = -4.833, p < 0.001, Table
A.3). Of men, 84% crossed at a red light; this was only 75% for women. There is a significant
difference in the color of the light when crossing, depending on whether the person is accompanied
(estimate = -0.452, z-value = -3.444, p < 0.001, Table A.3). Of accompanied people, 83% crossed at
a red light, against 78% of single people. The distractors have a negative effect on crossing at the
red light (estimate = 0.164, z-value = 2.111, p = 0.035, Table A.3). The more distractions
pedestrians accumulated, the lower the percentage of reds (80% crosses to the red light without
;2
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
19
distractors, 77% with 1 distractor, 69% with 2 distractors, pairwise t-test p < 0.05). The GLM 2 also
shows a significant effect of group size (estimate = 0.200, z-value = 10.280, p < 0.001, Table A.3)
and order of departure (estimate = -0.313, z-value = -8.752, p < 0.001, Table A.3). The percentage
of pedestrians who cross at the red signal decreases with the size of the group and increases with the
departure order as a mimetic effect. Finally, the presence of at least one child has an impact on the
percentage of red crossings (estimate = 1.066, z-value = 5.168, p < 0.001, Table A.3). When at least
one child is present, only 58% of pedestrians cross on red versus 79% without children.
3.3. Waiting behavior before crossing
No effect is apparent on the presence of the signs. It seems that the visual nudge does not lead to
more waiting behavior at the pedestrian crossing (child’s eyes sign: estimate = -0.093, z-value = -
0.890, p = 0.373 ; woman’s eyes sign: estimate = -0.132, z-value = 1.235, p = 0.217 , Table A.4).
However, there is a significant difference in waiting behavior according to gender (estimate = -
0.475, z-value = -6.145, p < 0.001, Table A.4). Of women, 58% waited at red pedestrian lights
compared to 45% of men. To be accompanied reduces the stopping behavior of pedestrians
(estimate = -0.280, z-value = -2.818, p = 0.005, Table A.4). Of accompanied pedestrians, 49% stop
at traffic lights compared to 53% of unaccompanied pedestrians. There is also an age effect
(estimate < 0.001, z-value = -2.642, p = 0.008, Table A.4) with children under 10 years old always
stopping at traffic lights. Of them, 100% stopped, against 44%, 34% 44%, 48%, 47%, for age
groups of 30, 40, 50, 60 and 70, respectively (pairwise t-test p < 0.05). The 40-year-old category
seems to be the one that stops the least (34%) and most significantly compared to the under 40s
(pairwise t-test p < 0.05). Furthermore, there seems to be a certain trend: pedestrians’ waiting time
decreases until they are 40 years old, and then stops more often until they are 80 years and over. As
with waiting time, distractors encourage stopping at traffic lights (estimate = 0.305, z-value = 4.469,
p < 0.001, Table A.4). The more distractions pedestrians accumulate, the more they will stop (49%
stop with no distractors, 56% with one distractor, 73% with 2 distractors, pairwise t-test p < 0.05).
1>
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
20
When we examine the effect of group size, the percentage of stopping at a pedestrian light increases
with group size, until a certain point where the effect starts to stagnate (estimate = 0.199, z-value =
9.679, p < 0.001, Table A.4). A single pedestrian waits only 39% of the time, whereas one in a
group of 7-8 pedestrians will wait 80-100% of the time (pairwise t-test p < 0.05). GLM 3 also
shows an effect of the departure order (estimate = -1.145, z-value = -6.761, p < 0.001, Table A.4),
with the percentage of waiting pedestrians stagnating at around 50% and with the starting order
until the 15th, from which the percentage of waiting pedestrians decreases to 19% (pairwise t-test p
< 0.05). Lastly, an effect of the presence of children on the pavement can be observed here
(estimate = 0.470, z-value = 2.360, p = 0.018, Table A.4). A pedestrian stops 61% of the time when
a child is present, compared to 52% of the time when a child is not present.
3.4. Waiting time before crossing
Of the 2967 crossings analyzed, 1556 pedestrians (52.4%) waited before crossing (waiting time >
0). The following analysis is conducted on this dataset: The average waiting time was 15.71 ± 13.79
seconds for a maximum observation of 81.670 seconds and a minimum of 0.170 seconds.
No effect of signs on waiting time is shown on the fourth GLM (Table A.5). There is no significant
difference of waiting time according to the test conditions for woman’s eyes and child’s eyes signs,
compared to the control conditions (child’s eyes sign: estimate = 0.006, t-value = 1.683, p = 0.093;
woman’s eyes sign: estimate = -0.001, t-value = -0.142, p = 0.887, Table A.5).
Other variables appear significant in the model. There is an age effect (estimate < 0.001, t-value =
2.168, p = 0.030, Table A.5) with young people aged 10 to 30 waiting longer at the red lights than
adults aged 40 to 60 (pairwise t-test, p < 0.05). Pedestrians with distractors wait longer (estimate = -
0.011, t-value = -5.377, p < 0.001, Table A.5) and the pairwise t-test shows that the more distractors
the subject accumulates, the longer the waiting time may be (p-value < 0.05). The size of the group
can also impact the waiting time (estimate = -0.002, t-value = -3.404, p < 0.001, Table A.5). A
lonely pedestrian will wait 13 seconds on average, but a pedestrian in the presence of other
1;
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
21
pedestrians will wait 3 to 12 seconds more, which is evident because more pedestrians
automatically reduced the probability that one goes faster (Pelé et al., 2019b). Groups of seven
pedestrians seem to wait the longest (average waiting time of 23 seconds, pairwise t-test, p < 0.05).
3.5. Risk taking after crossing
Risk taking was studied only on pedestrians who crossed at a red pedestrian light (i.e., 2335
pedestrians). Risk-taking behavior is studied based on the risk time (i.e., the time between the
pedestrian’s departure and arrival of the first car after them). The average risk time was 27.46 ±
20.7 seconds for a maximum observation of 80.781 seconds and a minimum of 0.153 seconds.
Table A.6 shows that the control sign for flowers influences risk taking (estimate = -0.056, t-value =
-2.201, p-value = 0.028, Table A.6 ). It appears that the presence of this device leads to a shorter
risk time and thus, higher risk taking (Figure A.1). Pedestrians leave less time before the next car in
the presence of the control flowers sign, compared to the situation with the woman’s eye sign
(pairwise comparison t-test, p-value = 0.008).
There is also a strong age effect (estimate = 0.002, t-value = 3.655, p < 0.001, Table A.6) and
gender effect (estimate = -0.052, t-value = -2.989, p = 0.003, Table A.6). Youth and young adults
(10-30 years) have a risk time of 26.392 seconds and 25.286 seconds, respectively, as opposed to
adults in their 30s and 50s (average of 29.888 seconds and 33.389 seconds, respectively). The
difference in risk time is particularly large between 20-year-olds and 70-year-olds (about a 9 second
difference). Risk-taking is significantly lower for women than for men (Table A.6). On average,
women leave about 28.558 seconds between crossing and the next car, compared to 26.194 seconds
for men. The order seems to influence the risk time (estimate = 0.014, t-value = 2.539, p = 0.011,
Table A.6). This effect appears only as a weak tendency for the first to leave with the comparison
pairwise t-test (p = 0.051). The first to leave would take more risks than the second. Moreover,
there is a strong effect of the interaction of group size and starting order (estimate = -0.003, t-value
= -3.542, p < 0.001, Table A.6).
11
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
22
3.6. Hesitation behavior
GLM 6 tests the different variables on hesitation. It was conducted on pedestrians who crossed at a
red pedestrian light (i.e., 2335 pedestrians and 87 hesitations).
Of hesitations, 2% were observed in the presence of the women’s eyes sign. There is significantly
less hesitation in the presence of this image (estimate = -9.930, z-value = -2.470, p = 0.013, Table
A.7) compared to other conditions: 4% without the sign, 5% for the flowers sign, and 4% for the
child’s eyes sign. Nevertheless, the comparison pairwise t-test shows no significant difference (p >
0.05).
Pedestrians who were accompanied hesitated more than pedestrians who were alone (6% vs. 3%,
respectively; estimate = 0.732, z-value = 2.928, p = 0.003, Table A.7). Other variables do not
influence hesitation behavior (Table A.7).
1M
515
516
517
518
519
520
521
522
523
524
525
526
527
23
4. Discussion
The aim of this study was to test the impact of visual nudges within a pedestrian context. These
images could have an impact on pedestrian behavior by encouraging safety during dangerous
crossings. Contrary to initial predictions, the presence of the eye images did not have the expected
social control effect on the behavior of pedestrians crossing at red lights. The eyes signs had no
significant effect, or even an opposite effect to the desired one, on the different variables of interest.
Neither of the two eyes signs influenced the rate of red-light crossings (PR ratio), risk taking, or
whether pedestrians stopped. However, the child's eyes sign tended to reduce the waiting time at the
pedestrian crossing. In addition, the control flowers sign seemed to have significantly increased
pedestrian risk taking when crossing at the red light.
Other variables influenced these parameters. Age is one of the factors that emerged repeatedly
across the models. Children under 10 years of age, who are therefore accompanied by an adult,
crossed less often at a red light and always stopped before crossing. The results also show that
young adults tend to cross less often at a red light and have longer waiting times than adults aged 40
and above. The 40+ age group stops the least often at a red light. However, young people between
10-30 years old take more risks than people aged 30-50. Gender is also a determining factor. In
general, men show riskier behavior than women. They have a higher rate of crossing at a red light,
stop less often at a red light, and leave less time between themselves and cars. In addition to
individual characteristics, the environment is a determining factor. Being accompanied also plays a
role in these different parameters. A pedestrian accompanied by a familiar person crosses at a red
traffic light more often than if they had been alone. They will also stop less often, but will be more
likely to show hesitation. The use of a phone, headset, or both, and other distractors tends to
decrease the rate of red light running. Their presence increases the number of stops at lights and
lengthens waiting time. Moreover, these effects increase with the accumulation of distractors. The
1F
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
24
presence of a child on the pavement will also reduce the rate of illegal crossing and increase stops at
pedestrian crossings. In addition, a group effect emerged from this study. The rate of crossing on a
red light decreased with group size, unlike the percentage of stopping at the pedestrian crossing and
the waiting time before crossing, which increased with group size. In the same way, there is an
effect of the starting order. The rate of crossing on a red light tends to increase when after the first
pedestrian starts, and the last pedestrian to cross seems to stop less often. It is also worth noting that
there is an interaction between group size and starting order. It seems that the last pedestrian to
cross among a large group takes the greatest risks. Finally, the configuration of the road and signage
plays an important role. Risk behavior depends on traffic and traffic light cycles.
1?
554
555
556
557
558
559
560
561
562
563
25
4.1. Classic individual, social, and environmental variables influencing the variables of
interest
The results of this study are consistent with the literature and confirm that gender influences
crossing behavior. Men took more risks in relation to cars than women. Indeed, men are often
shown to exhibit more dangerous behaviors than women (Di Stasi et al., 2014; Pelé et al., 2017;
Tom & Granié, 2011; Vujanić et al., 2014; Wang et al., 2011). Age also seems to be a major factor
in pedestrian risk behavior according to our results. This variable appears in several of our models;
three groups can be dissociated: children under 10 years old who are accompanied, young adults
between 10-30 years old, and adults over 30 years old. The presence of a child can bring out social
norms and all types of pedestrians behave better; there are more crossings on green, and more
pedestrians stopping at red lights (Granié, 2010; Smetana, 1997). Adults, and particularly the 40-
year-old age group, show riskier behavior than young people. However, their risk taking in front of
the car seems to be lower than that of young people. Thus, it seems that the type of risk taking
depends on age. Youngs adults seem to be more aware of their risk-taking than older adults who
cross directly and put themselves in danger, often unintentionally (Giuffrè et al., 2017; Holland &
Hill, 2007, Daamen & Hoogendoorn, 2003; Zhang et al., 2019). There is also a distraction effect. In
this study, pedestrians with a distraction have safer crossing behaviors, in contrast to Nasar et al.
(2008), who show that using a phone while crossing the street leads to reduced attention and more
dangerous crossing behaviors. Thus, one strategy that may have been implemented by the
pedestrians is to use their phones while stopped and then gather information before crossing. In this
study, accompanied pedestrians showed riskier behavior than solo pedestrians. These results are
contrary to those found by Pelé et al. (2017), who suggest that crossing with a familiar person leads
to greater compliance. Being accompanied may distract the pedestrian. It was observed that it is
often the companion who holds back the other person when crossing, leading to hesitation. Indeed,
pedestrians show a strong mimicry in following social information when crossing the road,
13
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
26
sometimes inadvertently crossing against the red light (Faria, 2010; Pelé et al., 2017; Jay et al.,
2020).
Social factors also emerged in our analyses. Groups of pedestrians waited longer before crossing
than single pedestrians. It seems that the larger the group size, the more pedestrians stop and tend to
cross at green lights. Pedestrians conform to group pressure and tend to adopt the behaviors of
others to maintain their social credibility (Osman, 1982). Furthermore, pedestrians who cross first in
a group show the lowest risk-taking behavior. This is because the first to cross does so intentionally,
unlike followers, who may follow other pedestrians without having bothered to gather information
about their surroundings themselves. This may also explain the fact that the last to cross are the
ones who stop the least at the crossing. This mimicry may lead to greater risk taking, but, as the
number of pedestrians crossing increases, the risk to pedestrians decreases (Leden, 2002) because a
group is more visible than a single individual. However, the risk-time model shows that the first to
leave has higher risk-taking behavior. This result is opposed to the previous results but can be
explained by the fact that here, the pedestrian crossing was numbered in relation to the traffic light
cycle, and not in relation to the presence of a group. Thus, many of the pedestrians observed to be
numbered first were alone, which may represent a bias. The interaction between group size and
starting order is thus more relevant to interpret.
The design of the site itself is also important. The number of lanes is different at the sites studied.
The more lanes there are, the more cars there are (Pelé et al., 2017). Increasing the number of cars
tends to lead to a decrease in the red crossing rate. However, in some studies, traffic density does
not affect the infringement rate (Dommes et al., 2015). This heterogeneity is related to the fact that
pedestrian behavior is influenced by many car-related environmental parameters, such as the speed
of these vehicles (Zhang et al., 2017) or the number of parked vehicles (Tezcan et al., 2019). The
presence of a tram line also has a direct impact on the red crossing rate. It can cause traffic light
14
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
27
cycles to be irregular and influence pedestrian violation behavior (Keegan & O'Mahony, 2003;
Wang et al., 2011). To minimize the site effect, observations at many sites could be the solution.
4.2. Absence and unintended effect of nudges
No change in the pedestrian red light crossing rate was found between our four experimental
conditions. The rate of crossing on red for pedestrians arriving at the red light is 78.7%. Similar
rates were found in Strasbourg in the study by Pelé et al. (2017), with 76.3% crossing at a red light.
The rate of illegal crossings is still significantly high in France compared to those observed in other
countries. Of pedestrians, 5.7% crossed on red in Nagoya, Japan (Pelé et al., 2017) and this was
13.5% in a study conducted in Israel (Rosenbloom, 2009). Furthermore, the results on the rate of
pedestrians crossing on a red light, out of the total number of pedestrians arriving indifferently at
the red or green light, was 48.3%, as confirmed in Pelé et al. (2017). Higher percentages of
violations were found in the study of Sueur et al. (2013), with 67% crossing on a red pedestrian
light in Strasbourg when crossing alone. Once again, much lower percentages were found at
Japanese sites, with 2.1% and 6.9% of red-light crossings, respectively, in studies by Pelé et al.
(2017, all crossings)) and Sueur et al. (2013, crossing alone). Of red-light crossings, 40.4% were
found in the study by Rosenbloom (2009) in Israel.
The child's eye sign tends to reduce waiting time at the pedestrian crossing. This is contradictory to
the expected predictions, as this experimental sign should instead lead to safer behavior. This result
is even more surprising, as it has been shown that the presence of children promotes good behavior.
The presence of a simple image of neutral child's eyes does not influence the social norm of
pedestrians. The flowers control sign also had a surprising effect. It seemed to decrease the time
from crossing to the next car, compared to the other conditions, and thus increase risk taking. These
results may suggest that pedestrians did not interpret the nudges in the desired way and that the
1@
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
28
presence of a sign next to the traffic light disturbed them. It can be hypothesized that the pedestrians
were surprised or even distracted by the signs, which provoked a reaction similar to fear. Indeed,
the use of fear appeals in road safety campaigns is not conclusive, according to the literature review
by Wundersitz and Hutchinson (2011). Some studies have even shown that fear appeals can be
counterproductive and lead to defensive and maladaptive behaviors (Wundersitz & Hutchinson
2011; Elliott 2003).
However, this type of visual nudge (eyes sign) has been shown to lead to behavioral changes in
several studies (Bateson et al., 2006; Ernest-Jones et al., 2011; Francey & Bergmüller, 2012;
Panagopoulos, 2014; Powell et al., 2012; Sénémeaud et al., 2017). Several hypotheses can explain
the differences in results to our study.
First, the context is a parameter to consider. In many studies, eye nudges have been tested in an
academic setting (Bateson et al., 2006; Ernest-Jones et al., 2011; Sénémeaud et al., 2017); thus, it is
done in a professional, everyday setting. In other studies, such as Powell et al. (2012), it is done in a
shop and nudges created an increase in generosity. Similarly, the study by Francey and Bergmüller
(2012) occurred near a bus stop. However, in this study, the effects of the images were nuanced, as
they did not lead to a change in the rate of good behavior (i.e., sorting waste), but to an increase in
the time allowed to perform this behavior. Researchers who conducted a meta-analysis of studies on
eye nudges (Northover et al., 2017) highlighted the fact that the effect of these devices varied
greatly, depending on the social context. For example, these experiments work better in quiet
overcrowded places.
In addition, the abovementioned studies have been conducted in various countries, in the United
Kingdom (Bateson et al., 2006; Ernest-Jones et al., 2011; Powell et al., 2012), United States
(Panagopoulos, 2014), France (Sénémeaud et al., 2017), and Switzerland (Francey & Bergmüller,
12
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
29
2012). Our study took place on a French population sample. . Shiwakoti et al. (2019) have shown
that the strategy to adopt may depend on the culture of the target population. For example, Hoekstra
and Wegman (2011) report that in some countries, such as the Netherlands, there is a long tradition
of humor-based road safety campaigns. In contrast, in countries such as Australia, New Zealand, the
United States, and Great Britain, it is common to display explicit images of accidents, victims,
injuries, blood, and grief of road victims (Hoekstra and Wegman 2011). A study to select the most
relevant type of message in France might be interesting. The French are rather individualistic and
have little respect for pedestrian rules. Especially since the amount for a fine of crossing at a red
pedestrian light is very low (only 4 euros). However, nudges, by definition, are devices that
encourage people without constraining them. This experiment could show different results for
another population, one that is more respectful of pedestrian signals and more concerned about the
eyes of others, such as the Japanese (Sueur et al., 2013). The Japanese show vastly different
pedestrian behavior from the French, with only 2.1% of Japanese pedestrians crossing illegally,
compared to 41.9% of French pedestrians in the study by Pelé et al. (2017). Thus, testing eye
nudges in a pedestrian context, but in a country with a different culture, could be an interesting
experiment to conduct.
Finally, the meta-analysis by Northover et al. (2017) highlights the wide range of behaviors targeted
by the different gaze impact studies. The behaviors targeted by eye nudges are generally altruistic,
such as donations of money (Bateson et al., 2006; Powell et al., 2012) or blood (Sénémeaud et al.,
2017), or good deeds (Ernest-Jones et al., 2011; Francey & Bergmüller, 2012). These studies did
not target risky behaviors, such as those observed in this study.
The diversity of social contexts, populations, distance between the device and subject, and
behaviors targeted are all parameters whose weight in the results is difficult to measure. Numerous
studies have found effects to these nudges, representing artificial eyes, which would act as the
M>
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
30
presence of a real person observing someone else. However, there are also studies where the
presence of these eyes did not lead to changes in subjects' behaviors, either in frequency or
sequence. This was the case of 18 studies reported by Northover et al. (2017). This same study
found no evidence of any impact of the artificial eyes on subjects' generosity behaviors, and
suggests caution in interpretation.
M;
694
695
696
697
698
699
700
31
5. Conclusion
Visual nudge in the form of eyes signs are aimed at pedestrians who intentionally cross at a red light
(i.e., pedestrians who know the color of the light). They did not have the desired effect and tended
to increase risk taking rather than encourage safe behaviors. This suggests that the visual nudge
technique for pedestrians can have an impact, but that the images chosen here may not have been
appropriate. Testing other images according to the context and culture of the target population could
lead to changes in behavior for these pedestrians and hence better results. In addition, the target
behaviors can be subtle, in the order of a hundredth of a second, and are part of a complex pattern of
decision-making in a dangerous situation. It might be interesting to test the visibility of the device
and set up different signs with different levels of visibility (e.g., flashing). Furthermore, testing
audible devices and warning pedestrians of the color of the light when they are about to cross
illegally could be an interesting experiment. This system, which is more difficult to circumvent and
capable of acting on more pedestrians (e.g., pedestrians who cross unintentionally by following
previous pedestrians) could complete our results.
Funding: This work was supported by the ONISR, Observatoire national interministériel de la
sécurité routière.
Data statement: Data is available upon request.
Declaration of interests: None
Declaration of informed consent: No identifying information was collected from the pedestrians.
The data is anonymous, the focal pedestrians were only coded by their location, day, and time of
crossing the street. The Bas-Rhin Prefecture was informed about the experiment and the General
Data Protection Regulation (GDPR) was respected. The experiment was approved by the ONISR
and SIRAC. During recordings, pedestrians had the opportunity to be informed about the study if
they requested it. They could also have access to the contact information of the authors.
M1
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
32
Bibliography
Abdi, H. (2007). The Kendall rank correlation coefficient. Encyclopedia of Measurement and
Statistics. Sage, Thousand Oaks, CA, 508-510.
Altmann, J. (1974). Observational study of behavior: Sampling methods. Behaviour, 49(3-4), 227-
266.
Association prévention routière. https://www.preventionroutiere.asso.fr/2019/12/22/statistiques-
daccidents/
Barmeyer, C., & Davoine, E. (2013). Intercultural difficulties in French-German research and
development projects. In Andrea Zenker, Elisabeth Baier, Emmanuel Muller, Jean-Alain Héraud,
Volker Tippmann (Eds.): Strategies for bilateral research cooperation: French-German
experience, Fraunhofer Verlag, pp. 31-40
Bateson, M., Nettle, D., & Roberts, G. (2006). Cues of being watched enhance cooperation in a
real-world setting. Biology Letters, 2(3), 412-414.
Bergeron, J., De Lavalette, B. C., Tijus, C., Poitrenaud, S., Leproux, C., Thouez, J. P., & Rannou,
A. (2008). Effets des caractéristiques de l’environnement sur le comportement des piétons à des
intersections urbaines. Le piéton et son environnement. Quelles interactions [The pedestrian and its
environement, what interactions ], 163-174.
Bungum, T. J., Day, C., & Henry, L. J. (2005). The association of distraction and caution displayed
by pedestrians at a lighted crosswalk. Journal of Community Health, 30(4), 269-279.
Conseil National de sécurité routière (2019). Enjeux d’accidentologie des piétons. [French national
center for road security (2019)Pedestrian accident issues]. https://www.onisr.securite-
MM
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
33
routiere.gouv.fr/sites/default/files/2019-07/CNSR%20-%20Comité%20des%20experts%20-
%20note%20piétons%20-%20VF.pdf. 2021/05/16
Daamen, W., & Hoogendoorn, S. P. (2003). Experimental research of pedestrian walking
behavior. Transportation Research Record, 1828(1), 20-30.
Ding, T. Q., Li, W., Xi, J. F., Zheng, L. L., & Wang, S. L. (2014). Analysis of Pedestrian Safety at
Signalized Intersections Based on Pedestrian Crossing Behavior. In CICTP 2014: Safe, Smart, and
Sustainable Multimodal Transportation Systems (pp. 2291-2303).
Di Stasi, L. L., Megías, A., Cándido, A., Maldonado, A., & Catena, A. (2014). The influence of
traffic signal solutions on self-reported road-crossing behavior. The Spanish Journal of
Psychology, 17, E103.
Dommes, A., Granié, M. A., Cloutier, M. S., Coquelet, C., & Huguenin-Richard, F. (2015). Red
light violations by adult pedestrians and other safety-related behaviors at signalized
crosswalks. Accident Analysis and Prevention, 80, 67-75.
Elie, C., & Colombet, I. (2011). Méthodes d’estimation de la reproductibilité [Methods to estimate
replicability]. Sang Thrombose Vaisseaux, 23(3), 138-145.
Elliot, A. J., & Maier, M. A. (2007). Color and psychological functioning. Current Directions in
Psychological Science, 16(5), 250-254.
Elliott, B. (2003). The psychology of fear appeals revisited. In Proceedings of the Australasian
Road Safety Research, Policing and Education Conference (Vol. 7, No. 2, pp. 266-273). Monash
University.
Ernest-Jones, M., Nettle, D., & Bateson, M. (2011). Effects of eye images on everyday cooperative
behavior: A field experiment. Evolution and Human Behavior, 32(3), 172-178.
MF
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
34
Faria, J. J., Krause, S., & Krause, J. (2010). Collective behavior in road crossing pedestrians: The
role of social information. Behavioral Ecology, 21(6), 1236-1242.
Francey, D., & Bergmüller, R. (2012). Images of eyes enhance investments in a real-life public
good. PLoS One, 7(5), e37397.
Giuffrè, T., Campisi, T., & Tesoriere, G. (2017). Implications of adaptive traffic light operations on
pedestrian safety. IOSR J. Mech. Civ. Eng, 13, 58-63.
Granié, M. A., & Espiau-Nordin, G. (2008). Etude qualitative du comportement piéton de
collégiens par la méthode de l’autoconfrontation. Territoire en mouvement Revue de géographie et
aménagement [Territory in Movement Journal of Geography and Planning], 1, 39-57.
Granié, M. A. (2010). Socialisation au risque et construction sociale des comportements de l’enfant
piéton: éléments de réflexion pour l’éducation routière. Enfances, familles, générations [Childhood,
Famillies, generations], 12, 88-110.
Greenwald, M. J., & Boarnet, M. G. (2001). Built environment as determinant of walking behavior:
Analyzing nonwork pedestrian travel in Portland, Oregon. Transportation Research
Record, 1780(1), 33-41.
Guéguen, N., Eyssartier, C., & Meineri, S. (2016). A pedestrian's smile and drivers' behavior: When
a smile increases careful driving. Journal of Safety Research, 56, 83-88.
Guéguen, N., Meineri, S., & Eyssartier, C. (2015). A pedestrian’s stare and drivers’ stopping
behavior: A field experiment at the pedestrian crossing. Safety Science, 75, 87-89.
Holland, C., & Hill, R. (2007). The effect of age, gender and driver status on pedestrians’ intentions
to cross the road in risky situations. Accident Analysis and Prevention, 39(2), 224-237.
M?
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
35
Hoekstra, T., & Wegman, F. (2011). Improving the effectiveness of road safety campaigns: Current
and new practices. IATSS research, 34(2), 80-86.
Jay, M., Régnier, A., Dasnon, A., Brunet, K., & Pelé, M. (2020). The light is red: Uncertainty
behaviours displayed by pedestrians during illegal road crossing. Accident Analysis and
Prevention, 135, 105369.
Keegan, O., O’Mahony, M. (2003). Modifying pedestrian behaviour. Transportation Research Part
A: Policy and Practice, 37, 889–901. https://doi.org/10.1016/S0965-8564(03)00061-2
Leden, L. (2002). Pedestrian risk decrease with pedestrian flow. A case study based on data from
signalised intersections in Hamilton, Ontario. Accident Analysis and Prevention, 34, 457–464.
Lipovac, K., Vujanic, M., Maric, B., Nesic, M. (2013). The influence of a pedestrian countdown
display on pedestrian behaviour at signalised pedestrian crossings. Transportation Research Part F:
Traffic Psychology and Behaviour, 20, 121–134.
McIlroy, R. C., Nam, V. H., Bunyasi, B. W., Jikyong, U., Kokwaro, G. O., Wu, J., ... & Stanton, N.
A. (2020). Exploring the relationships between pedestrian behaviours and traffic safety attitudes in
six countries. Transportation Research Part F: Traffic Psychology and Behaviour, 68, 257-271.
Mueller, B. A., Rivara, F. P., Lii, S. M., & Weiss, N. S. (1990). Environmental factors and the risk
for childhood pedestrian-motor vehicle collision occurrence. American Journal of
Epidemiology, 132(3), 550-560.
Nasar, J., Hecht, P., & Wener, R. (2008). Mobile telephones, distracted attention, and pedestrian
safety. Accident Analysis and Prevention, 40(1), 69-75.
Northover, S. B., Pedersen, W. C., Cohen, A. B., & Andrews, P. W. (2017). Artificial surveillance
cues do not increase generosity: Two meta-analyses. Evolution and Human Behavior, 38(1), 144-
153.
M3
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
36
Observatoire national interministériel de la sécurité routière [ONISR] (2019).
https://www.onisr.securite-routiere.gouv.fr/sites/default/files/2020-09/
Bilan_2019_version_site_internet_24_sept.pdf . 2021/05/16
Osman, L. M. (1982). Conformity or compliance? A study of sex differences in pedestrian
behaviour. British Journal of Social Psychology, 21(1), 19-21.
Panagopoulos, C. (2014). Watchful eyes: Implicit observability cues and voting. Evolution and
Human Behavior, 35(4), 279-284.
Papadimitriou, E., Theofilatos, A., & Yannis, G. (2013). Patterns of pedestrian attitudes,
perceptions and behaviour in Europe. Safety Science, 53, 114-122.
Pelé, M., Bellut, C., Debergue, E., Gauvin, C., Jeanneret, A., Leclere, T., ... & Sueur, C. (2017).
Cultural influence of social information use in pedestrian road-crossing behaviours. Royal Society
Open Science, 4(2), 160739.
Pelé, M., Deneubourg, J. L., & Sueur, C. (2019a). Decision-making processes underlying pedestrian
behaviors at signalized crossing: Part 1. The first to step off the kerb. Safety, 5(4), 79.
Pelé, M., Deneubourg, J. L., & Sueur, C. (2019b). Decision-making processes underlying
pedestrian behaviors at signalized crossings: Part 2. Do pedestrians show cultural herding
behavior?. Safety, 5(4), 82.
Powell, K. L., Roberts, G., & Nettle, D. (2012). Eye images increase charitable donations: Evidence
from an opportunistic field experiment in a supermarket. Ethology, 118(11), 1096-1101.
Rosenbloom, T. (2009). Crossing at a red light: Behaviour of individuals and
groups. Transportation Research Part F: Traffic Psychology and Behaviour, 12(5), 389-394.
M4
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
37
Sénémeaud, C., Sanrey, C., Callé, N., Plainfossé, C., Belhaire, A., & Georget, P. (2017). The
watching-eyes phenomenon and blood donation: Does exposure to pictures of eyes increase blood
donation by young adults?. Transfusion and Apheresis Science, 56(2), 168-170.
Shiwakoti, N., Tay, R., & Stasinopoulos, P. (2019). Development, testing, and evaluation of road
safety poster to reduce jaywalking behavior at intersections. Cognition, Technology and Work,
22(2), 1-9.
Smetana, J. G. (1997). Parenting and the development of social knowledge reconceptualized: A
social domain analysis. Parenting and Children's Internalization of Values: A Handbook of
Contemporary Theory, 162-192.
Sueur, C., Class, B., Hamm, C., Meyer, X., & Pelé, M. (2013). Different risk thresholds in
pedestrian road crossing behaviour: A comparison of French and Japanese approaches. Accident
Analysis and Prevention, 58, 59-63.
Sueur C., Piermattéo A., & Pelé M. (2022). Eyes image effect in the context of pedestrian safety: A
French questionnaire study. F1000Research 11, 218
Tezcan, H.O., Elmorssy, M., Aksoy, G. (2019). Pedestrian crossing behaviour at midblock
crosswalks. Journal of Safety Research, 71, 49–57.
Thaler, R. H., & Sunstein, C. R. (2009). Nudge: Improving decisions about health, wealth, and
happiness. Penguin (Introduction).
Thompson, S. J., Fraser, E. J., & Howarth, C. I. (1985). Driver behaviour in the presence of child
and adult pedestrians. Ergonomics, 28(10), 1469-1474.
Tom, A., Granié, M.-A. (2011). Gender differences in pedestrian rule compliance and visual search
at signalised and unsignalized crossroads. Accident Analysis and Prevention, 43, 1794–1801.
M@
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
38
United Nations (2019). World Urbanization Prospects.
https://population.un.org/wup/Publications/Files/WUP2018-Report.pdf. 2022/05/16
& O$ 5 $ P L 5 $ )1>11*$  !#! "#
-! & !! "  !"   !$
Transportation Research Part F: Tra$c Psychology and Behaviour, 871M391F@$
Vujanić, M., Pešić, D., Antić, B., Smailović, E. (2014). Pedestrian risk at the signalised pedestrian
crossing equipped with countdown display. International Journal for Traffic and Transport
Engineering, 4, 52–61.
Wang, W., Guo, H., Gao, Z., & Bubb, H. (2011). Individual differences of pedestrian behaviour in
midblock crosswalk and intersection. International Journal of Crashworthiness, 16(1), 1-9.
Wundersitz, L., & Hutchinson, T. (2011). What can we learn from recent evaluations of road safety
mass media campaigns?. Journal of the Australasian College of Road Safety, 22(4), 40-50.
Yüksel, A. (2009). Exterior color and perceived retail crowding: Effects on tourists’ shopping
quality inferences and approach behaviors. Journal of Quality Assurance in Hospitality and
Tourism, 10, 233–254.
Zhang, C., Chen, F., Wei, Y. (2019). Evaluation of pedestrian crossing behaviour and safety at
uncontrolled mid-block crosswalks with different numbers of lanes in China. Accident Analysis and
Prevention, 123, 263–273.
Zhang, C., Zhou, B., Chen, G., & Chen, F. (2017). Quantitative analysis of pedestrian safety at
uncontrolled multi-lane mid-block crosswalks in China. Accident Analysis and Prevention, 108, 19-
26.
M2
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
39
Appendix
Tables:
Table A.1. Number of pedestrians observed per site and condition to number of hesitations (H) and crossings with at
least one child (C).
Site Nuée Bleue Quai de Paris Université
Pedestrians H C Pedestrians H C Pedestrians H C
Control
condition
(no sign)
257 14 2 129 1 5 437 19 4
Control
condition
(flowers)
223 15 8 105 0 9 324 13 2
Nudge
(child's
eyes)
369 18 11 98 3 5 322 6 3
Nudge
(woman’s
eyes)
329 7 13 108 1 2 266 5 3
Total: 1178 54 34 440 5 21 1349 43 12
Table A.2: Results of the Generalized Linear Model 1 (quasi-Poisson law) for testing the red-light crossing rate (Ratio
PR).
Estimate Std. Error t-value Pr(>|t|)
(Intercept) -0.174 0.061 -2.866 0.008 **
Sign: Flowers 0.001 0.034 0.017 0.986
Sign: Child’s eyes 0.014 0.034 0.420 0.678
Sign: Woman’s eyes 0.007 0.034 0.201 0.842
Tram: Yes 0.194 0.034 5.642 < 0.001 ***
Number of cars -0.002 < 0.001 -4.297 < 0.001 ***
Signif. codes: ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1
Table A.3: Results of the Generalized Linear Model 2 (binomial law) for testing the light’s color when pedestrians are
crossing (green or red).
Estimate Std. Error z-value Pr(>|z|)
(Intercept) -0.534 0.147 -3.619 < 0.001 ***
Sign: Flowers -0.108 0.131 -0.822 0.411
Sign: Child’s eyes -0.312 0.130 -2.396 0.017 *
F>
880
881
882
883
884
885
886
887
888
889
890
40
Sign: Woman’s eyes -0.140 0.130 -1.081 0.280
Age -0.009 0.003 -2.626 0.009 **
Sex (man) -0.474 0.098 -4.833 < 0.001 ***
Accompanied: Yes -0.452 0.131 -3.444 0.001 ***
Distractors 0.164 0.078 2.111 0.035 *
Group 0.200 0.019 10.280 < 0.001 ***
Order of departure -0.313 0.036 -8.752 < 0.001 ***
Children 1.066 0.206 5.168 < 0.001 ***
Signif. codes: ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1
Estimate Std. Error z-value Pr(>|z|) 
(Intercept) 0.428 0.119 3.596 0.000 ***
Sign: Flowers -0.029 0.109 -0.263 0.792
Sign: Child’s eyes -0.093 0.105 -0.890 0.373
Sign: Woman’s eyes -0.132 0.107 -1.235 0.217
Age -0.007 0.003 -2.642 0.008 **
Sex (man) -0.475 0.077 -6.145 < 0.001 ***
Accompanied: Yes -0.280 0.099 -2.818 0.005 **
Distractors 0.305 0.068 4.469 < 0.001 ***
Group 0.199 0.021 9.679 < 0.001 ***
Order of departure -0.145 0.021 -6.761 < 0.001 ***
Children 0.470 0.199 2.360 0.018 *
Signif. codes: ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1
Table A.4: Results of the Generalized Linear Model 3 (binomial law) for testing the presence or absence of waiting time
for all pedestrians.
Estimate Std. Error t-value Pr(>|t|)
(Intercept) 0.064 0.004 14.867 < 0.001 ***
Sign: Flowers 0.005 0.004 1.297 0.195
Sign: Child’s eyes 0.006 0.004 1.683 0.093 .
Sign: Woman’s eyes -0.001 0.004 -0.142 0.887
Age < 0.001 < 0.001 2.168 0.030 *
Sex (man) 0.001 0.003 0.466 0.641
Accompanied: Yes -0.004 0.004 -1.136 0.256
Distractors -0.011 0.002 -5.377 < 0.001 ***
Group -0.002 < 0.001 -3.404 0.001 ***
Order of departure 0.001 0.001 1.233 0.218
Children -0.006 0.004 -1.662 0.097 .
F;
891
892
893
894
895
41
Signif. codes: ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1
Table A.5: Results of Generalized Linear Model 4 (gamma law) for testing the waiting time of pedestrians.
Table A.6: Results of the Generalized Linear Model 5 (quasi-Poisson law) for testing the time between the first car and
pedestrian crossing for pedestrians who have crossed at a red light (risk taking).
Estimate Std. Error t-value Pr(>|t|)
(Intercept) 1.521 0.029 52.745 < 0.001 ***
Sign: Flowers -0.056 0.025 -2.201 0.028 *
Sign: Child’s eyes -0.019 0.024 -0.796 0.426
Sign: Woman’s eyes 0.020 0.024 0.812 0.417
Age 0.002 0.001 3.655 < 0.001 ***
Sex (man) -0.052 0.017 -2.989 0.003 **
Accompanied: Yes 0.011 0.022 0.499 0.618
Distractors 0.014 0.016 0.906 0.365
Group 0.009 0.005 1.896 0.058 .
Order of departure 0.014 0.006 2.539 0.011 *
Children -0.076 0.069 -1.100 0.271
Interaction group: Order -0.003 0.001 -3.542 < 0.001 ***
Signif. codes: ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1
Table A.7: Results of the Generalized Linear Model 6 (binomial law) for testing the presence or absence of pedestrian
hesitation for those who have crossed at a red light.
Estimate Std. Error z-value Pr(>|z|)
(Intercept) -3.179 0.346 -9.183 < 0.001 ***
Sign: Flowers 0.196 0.283 0.693 0.488
Sign: Child’s eyes -0.181 0.291 -0.622 0.534
Sign: Woman’s eyes -0.930 0.376 -2.470 0.013 *
Sex (man) 0.051 0.221 0.230 0.818
Age 0.002 0.007 0.286 0.775
Accompanied: Yes 0.732 0.250 2.928 0.003 **
Distractors -0.005 0.212 -0.026 0.979
Group 0.052 0.037 1.429 0.153
Order of departure -0.121 0.067 -1.819 0.069 .
Children -0.688 1.005 -0.685 0.493
Signif. codes: ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1
F1
896
897
898
899
900
901
902
903
904
42
Figure:
Figure A.1: Risk time according to the experimental condition (GLM 5, quasi-Poisson law); pairwise t-test, p-value <
0.05.
FM
905
906
907
908
909
910
911
43
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.