Behavior
and
Parental
Expectations
of Child
Pedestrians
Rosemary C. Dunne, MS*; Kenneth N. Asher, PhD*; and Frederick P.
Rivara, MD, MPH*
ABSTRACT. Pedestrian injuries remain the most
corn-mon cause of death from trauma for young school-age
children. This study was based on the hypothesis that
parents’ abilities to accurately assess their children’s
street-crossing skills vary with the crossing test and age
of the children, being less accurate for younger children. Children at three developmental levels (aged 5 through
6,
7 through 8, and 9 through 10 years) and their parents were evaluated on four street-crossing tests and a control vocabulary test. For each test, children’s answers were compared to parents’ estimates of their children’s per-formance. Parents overestimated the abilities of their 5-through 6-year-olds on all four tests (P < .01). Parents overestimated the abilities of 7- through 8-year-olds on two of the tests (P < .05) and parents accurately assessed the abilities of the 9- through 1O-year-olds. On the vo-cabulary test, parents overestimated their children’s per-formance at all age levels (P < .01). The results support the hypothesis and indicate that parents’ expectations for their children’s pedestrian skills are least accurate for5- and 6-year-olds, with the mismatch decreasing as
children get older. Inaccurate expectations of children’s pedestrian skills may be a fruitful target for injury pre-vention programs. Pediatrics 1992;89:486-490; injury, pe-destrian, children behavior, parent supervision
Each year in the United States, more than 50 000
children are injured as pedestrians, of whom
approx-imately 1800 die, 1 8 000 are admitted to the hospital,
and 5000 have significant long-term sequelae.’
Al-though there has been a decrease in pedestrian fatality
rates, pedestrian injuries remain the most common
cause of death from trauma for the 5- to 9-year-old
age group and are second only to cancer in accounting
for mortality of young school-age children.2 The
de-velopinent and implementation of effective
preven-tion progranis are necessary if further significant
de-creases in morbidity and mortality are to occur;
im-proveinents in trauma care alone will not make the
difference.
Parents must be a key element of any program
designed to reduce child pedestrian injuries, as
par-ents set the expectations for children’s performance
in traffic.’ Surveys suggest that parents have
unreal-istic expectations of their children’s pedestrian
skills.35 One of the causes of child pedestrian injuries
From tht, ‘Harborview Injury Prevention and Research Center and the
L)epartnicnts of Pediatrics and Epidemiology. University of Washington,
Seattle.
Received for publication May 28, 1991; accepted Aug 21, 1991. Reprint rcqut’sts to (F P.R.) Harborview Injury Prevention and Research Center, Matistop ZX-1tJ. 325 Ninth Ave. Seattle, WA 98104.
l’EL)lAlRICS (ISbN 0031 4005). Copyright .c 1992 by the American
ALad-cloy of Pediatrics
may be a relative mismatch between children’s skills
and parents’ and society’s expectations.6
This study was undertaken to test the hypothesis
that such a mismatch exists and that parents’ abilities
to assess their children’s pedestrian skills vary with
children’s age and the difficulty of the crossing task.
With these results, it would be possible to design
injury prevention programs that take into account
both child development and parent perception
fac-tors.
METHODS
The intervention consisted of parents and children
independ-ently completing a set of street-crossing tests. The parents were
required to complete the same street-crossing tests as their children
but to respond in the way they thought their children would.
Differences between children’s performance and parents’
percep-tions of how their children would perform (parent-child
differ-ences) were calculated to identify the mismatch between parents’
knowledge of their children’s street-crossing skills and the actual
skills of their children.
Study Population and Recruitment Procedures
Two hundred forty parent-child dyads participated in the study.
The children were grouped into three age categories (5 through 6,
7 through 8, and 9 through 10 years). Families were recruited from two public schools, chosen because of proximity to the testing site. Parents were contacted initially by letter stating the purpose of the
study and explaining their involvement. The letter was followed
by a telephone call asking parents to participate and, if they
consented, scheduling the study procedures. These included written
materials, observation, and interview. Families received $50 at
completion of the testing procedures. The study was approved by
the human subjects committees of the University of Washington
and of the Seattle Public Schools. Written consent was obtained from both parents and children.
Assessment Package
The assessment package consisted of a demographic question-naire, standardized psychological tests, and four street-crossing
assessments (Table 1).
Demographic information, obtained from the parents by
self-administered questionnaire, included birth order of the child,
mar-ital status of the parent, relationship of the adult respondent to the
child, parent age, and socioeconomic status of the family. Standardized Tests A vocabulary test, a subtest of the
Stanford-Binet Intelligence Scale (Form L-M),7 was used to assess parents’
ability to estimate performance of their children on a skill unrelated
to street crossing, ie, as a control test. The vocabulary test was
administered in the home and consisted of a list of words read
aloud to children for which they were required to give definitions.
The parents were read the words, but were asked to estimate which
ones they thought their children would be able to define correctly. The Parent Accident Locus of Control Scale,’ the Children’s
Health Locus of Control Scale,9 and the Connors Behavior
Questionnaire’#{176} were administered to correlate locus of control and
behavior problems with crossing performance. However, these tests
proved to have poor correlation with crossing skills and will not be
TABLE 1. Tests Used in Assessing Parent-Child Differences
Test Purpose Location Parents’
Estimate
of Child’s
Child
Response
Demographic questionnaire Sample description Home X
Vocabulary test of Stanford-Binet Skill unrelated to street crossing Home X X
Intelligence Scale Street-crossing assessments
General traffic test Knowledge of traffic safety Home X X
Map test Abstract test of street-crossing skill Home X X
Simulated street-crossing test Street-crossing skills on simulated
street
Street site X X
Realistic street-crossing test Street-crossing skills on actual street
under controlled conditions
Street site X X
Street-Crossing Assessments The street-crossing tests followed the
same procedure as the vocabulary test in that the parents and
children conducted them independently in the absence of the other
subject, with the parents always responding to the test the way
they thought their children would. Discussion of the tests among
parents and children was prohibited until both had completed the
study. The street-crossing assessments consisted of the following: 1. General Traffic Test. The children and parents were required
to answer basic general traffic safety questions with the aid of
pictures representing different situations, eg, crossing between cars.
2. Map Test. This was an aerial representation of two adjacent
blocks-one with a stop sign and one with a pedestrian
signal-and four cars. The test was administered twice per block, with
different starting points each time. The subjects were requested to
draw a line indicating the path they would take to reach a
desti-nation at the other end of the block and on the opposite side of
the street as if it were a real street-crossing situation. The subjects were also asked questions about street-crossing behaviors in
rela-tion to the chosen route of crossing.
Both of the above traffic tests were conducted in the home. The
remaining two traffic tests were conducted outside on three
adja-cent quiet city blocks, which were the same for all subject dyads.
3, Simulated Street-Crossing Test. This method of street
simu-lation was suggested by the method used by Young and Le’ in
their study on child pedestrian behavior. The children and parents
were asked to cross a simulated street as if they were crossing the
real street. A plastic street with curb markings was laid widthwise
on the sidewalk at the corner adjacent to the real street. The subjects were required to cross the plastic street as if they were crossing at
that point on the parallel street, that is, taking into account the
traffic on the adjacent street when crossing the plastic street. The
simulated street crossing was conducted at both ends of a block;
one end had a stop sign while the other end was unregulated. The
observers waited until there was a moving vehicle within one block
before they asked the subjects to start crossing the simulated street.
The parents completed the same test as they thought the children would.
4. Realistic Street-Crossing Test. The children and parents were
asked to cross a real street. The configuration of the realistic street
crossing was similar to the map-drawing test. It was conducted on
two adjacent blocks, with a stopsign at the end of one block and a pedestrian signal at the end of the other block. The subjects were required to make two crossings per block. The goal of each crossing
was to reach a flag placed at the other intersection diagonally opposite to the children (ie, across the street and down the block). The children were instructed to cross the way they would if they were alone. To avoid danger to subjects during this test the blocks were cordoned off from through-traffic. To simulate street traffic a
car was driven up and down the street by a research assistant. The
car was positioned so as to approach the location where children
were preparing to cross. The parents followed the same procedure,
responding in the way they thought the children would.
Randomization
The order of presentation of the street-crossing procedures was randomized for all families. This consisted of three different
ran-dom assignments: (1) test sequence-home-administered
street-crossing tests vs simulated street-crossing tests vs realistic
street-crossing tests; (2) starting block-stop sign block vs pedestrian
signal block for the map test and realistic street test; and (3) starting
point on block-regulated vs unregulated end.
Scoring Procedures
The vocabulary test consisted of 25 words. The general
street-crossing test consisted of I 0 questions. The score was the total
number correct for each of these two assessments.
The remaining tests’ variables were coded separately. Each
variable had a possible score ranging from 0 to 4. In addition, total scores were calculated by adding the scores of each of the variables. The map test consisted of I 2 questions. The children were scored on the answers to six distinct street-crossing behaviors: walking
location, crossing location, stopping before crossing, search
behav-ior before crossing, traffic location while crossing, and search
behavior while crossing. Possible scores ranged from 0 to 48.
The simulated crossing test consisted of two crossing tests, scored
on three of the measurement variables listed above: search behavior
before crossing traffic location while crossing and search behavior while crossing. Possible scores ranged from 0 to 24.
The realistic crossing test consisted of four crossing tests, with the same six measurement variables as in the map test. Possible
scores ranged from 0 to 96.
Data Collection and Reliability
The observers used clipboards with behavioral checklists and
walked behind the subjects as they carried out the crossing tests.
To determine reliability two observers were present on 28% of the
observations on the street; one observer was used the remainder of the time.
Reliability was calculated by examining the number of times there was perfect agreement and the number of times the observers
disagreed by 1, 2, 3, or 4 points. Of the 28% of observations with
two observers, 85% of the time there was perfect agreement, 7%
of the time observers disagreed by 1 point, 6% by 2 points, and
2% by 4 points. The average difference in ratings was 0.27 points.
Analysis
Total scores for the street-crossing tests and vocabulary test were
converted into percent correct ([score obtained divided by maxi-mum score possible] x 100%) before the analyses were performed. In addition, individual scores for each variable of the map drawing, the simulated, and the realistic tests were analyzed. Mean scores were calculated for parents and children in each of the three age
groups (5 through 6, 7 through 8, and 9 through 10 years). Mean
differences (parent-child) were also calculated. t Tests and analyses of variance were conducted to identify significant differences.
The demographic data were analyzed using x2and analyses of variance.
RESULTS
Demographic Characteristics
Two hundred forty dyads completed the study, 79
in the 5- through 6-year age category, 80 in the
7-through 8-year age group, and 81 in the 9- through
demo-S correct
VOCAB-T GEN-T MAP-T SIMULAT-T REALISTIC-I
TEST
- 5-6 YRS 7-8 YRS 9-10 YRS
MAP-I SIMULAT-T REALISTIC-I
TEST
- 5-6 YRS 7-8 YRS
E1
9-10 YRSgraphically comparable (Table 2). The most frequent
adult respondents were mothers. Study children were
primarily first or second children, with oldest children
comprising 52% to 60% of the groups. Families who
participated were most commonly of high
socioeco-nomic status (SES), ie, at least one parent holding a
four-year college degree and/or pursuing a
semi-professional career. Middle SES families, ie, parents
who had at least 12 years of schooling and/or a
semiprofessional job, were the second most common
participants, followed by low SES families, which
included parents who had not finished high school
and/or worked as skilled or unskilled manual
em-ployees.
Children’s Performance
Children’s scores on all tests (vocabulary test and
the four street-crossing tests) improved significantly
with age (Fig 1). The performance of the 5- through
6-year-olds was significantly different on the general
test (P < .01), the map test (P < .01), and the realistic
street-crossing test (P < .05) than that of the other
two age groups. On these three tests 7- through
8-year-olds did not perform significantly better than
9-through 1 0-year-olds, while on the simulated
street-TABLE 2. Demographic Cha racteris tics of Families*
Characteristic Age Groups (y)
5-6 7-8 9-10
Pairs per group, n 79 80 81
% Children female 54 51 49
Birth order, %
1 52 60 53
2 38 30 25
3+ 10 10 22
Adult respondent, %
Mother 73 82 74
Father 19 14 20
Other 8 4 6
Married parents, % 72 68 62
Mean parent age, y 37.6 37.6 39.2
Socioeconomic status, %
High 56 44 41
Middle 26 30 33
Low 18 25 26
* No significant differences in variables by age.
crossing test the 9- through 10-year-olds performed
significantly better than the 5- through 6-year-olds
(P < .01). Across all ages children performed best on
the general test and worst on the simulated
street-crossing test.
Parents’ Perceptions of Children’s Performance
Parents’ scores differed significantly by children’s
age only on the vocabulary test (P < .01), in which
parents expected older children to have better
vocab-ulary knowledge than did parents of younger children
(Fig 2). Parents’ perceptions of children’s performance
on the street-crossing tests were not significantly
dif-ferent across the three age groups. Parents of younger
children tended to expect as much of them as parents
of older children. Similar to the children’s
street-crossing tests scores, parents expected their children
to score highest on the general test and lowest on the
simulated test.
Parent-Child Differences
Parents significantly overestimated their children’s
performance on the vocabulary test in all three age
groups (P < .01); there were no significant variations
by age. In contrast to the vocabulary test,
parent-child difference scores within each of the four
street-crossing tests varied significantly by children’s age (P
< .01) (Fig 3). Significant differences existed between
the 5- through 6-year-olds and the other
two
agegroups on the general test, the map test, and the
realistic street-crossing test. Analyses within age and
street-crossing test indicated that parents significantly
overestimated the abilities of 5- through 6-year-olds
on the general test and the realistic street-crossing test
(P < .01); parent-child differences on these tests for
the 7- through 8- and 9- through 10-year-olds were
nonsignificant. On the map test and the simulated
test, parents of 5- through 6-year-olds overestimated
their children’s skills (P < .01) as did parents of
7-through 8-year-olds (P < .05). There were no
signif-icant differences between parents’ and children’s
scores for the 9- through 10-year-olds.
Variations in scores for the street-crossing tests
differed by age (P < .01). In the youngest age group,
differences between parents’ and children’s scores for
Fig 1. Children’s scores by age and test. Analysis of variance was
used to test for a significant relationship of age to scores within each test (*P < .01; ** < .05).
80
60
40
20
0
Fig 2. Parents’ predictions of children’s scores by age and test. Analysis of variance was used to test for a significant relationship
- Y#ILK LOCATION STOPPING
- SEARCH ENVIRONMENT 1 SEARCH WHILE CR S correct
VOCAB-T GEN-I MAP-I SIMULAI-I REALISTIC-I
CONDITIONS
5-6 YRS 7-b YRS L a-io YRS
Fig 3. Differences between parents’ predictions and children’s
actual scores by age and test. Analysis of variance was used to test for significant relationship of age to scores within test (*P < .01).
Tests were used to test for significance of mean difference within
each age and test (P < .01; < .05).
the street-crossing tests were highest for the map test
(23%) and lowest for the general test (7%). Difference
scores within age and test were less than 5% for the
two older age groups.
Variation of Parent-Child Differences With Gender
Parent-child differences were examined for the
ef-fects of gender within the three age groups. In the
youngest age group, parents overestimated the skills
of girls more than those of boys, although the only
significant male-female effect was on the map test (P
< .05). This variation between boys and girls
de-creased with age and was not significant in older
children.
Specific Street-Crossing Tasks
We examined specific crossing tasks within the
realistic test and the map test to determine whether
the parent-child differences were confined to specific
components or occurred for all tasks involved with
crossing. The former was chosen for analysis as it was
the test most comparable with a real crossing
situa-tion. The latter was also analyzed because it was a
written representation of the realistic street-crossing
test. Six crossing tasks were measured: walking
be-havior (sidewalk/street), crossing location on street,
stopping behavior prior to crossing, search behavior
before crossing, crossing environment, and search
while crossing.
On the realistic street-crossing test, parents of
5-through 6-year-olds overestimated the skills of
cross-ing location, stopping before crossing, crossing
envi-ronment (P < .01), walking behavior, and search
before crossing (P < .05) (Fig 4). Similar parent
over-estimation of children in this age group on the map
test (data not shown) existed in all variables with the
exception of search before crossing; in addition,
search while crossing was significant (P < .01).
For the 7- through 8-year-old age group, crossing
location was the only task of the realistic street test
in which there was a significant parent-child
differ-ence. For the map test, differences were significant
for crossing location, search before crossing, and
search while crossing (P < .01).
Fig 4. Differences between parents’ predictions and children’s subscores on realistic street-crossing test by age and task. t Tests
were used to test for significance of mean difference within each
age and task (*J .01; P < .05).
This trend of significant differences being more
task specific in the 7- through 8-year-old group was
also seen in the 9- through 10-year-old group.
Cross-ing location (P < .01) and search before crossing (P
< .05) were significant for the realistic crossing task.
On the
map test thetwo
search subtasks weresignif-icant (P < .0 1).
DISCUSSION
The study was designed to determine the
cone-spondence between children’s pedestrian skills and
parents’ expectations of those skills. More specifically,
our aim was to test the hypothesis that parents’
abilities to assess their children’s pedestrian skills vary
with age of their children and the difficulty of the
street-crossing test.
As expected, children’s street-crossing behavior
im-proved with age, with differences in skill levels for
the street-crossing tests greatest between the youngest
and the two older age groups. However, parents’
perceptions of their children’s skill level on the
street-crossing tests did not vary by age. Parents of
5-through 6-years-olds did not differ from parents of
9- through 10-year-olds in rating skill level for
cross-ing streets. This same discrepancy was not found for
a more general measure of development, vocabulary
skills, in which parents of younger children did not
differ from parents of older children in accurately
estimating vocabulary skill. This overestimation of
children’s street-crossing skills decreased with age,
due to children’s skills acatching up” to parents’
esti-mates. Parents overestimated all crossing behaviors
in the youngest age group while effects were more
task specific in the two older age groups. However,
in all three age groups parents significantly
overesti-mated the task of crossing location, ie, parents
thought their children would cross at a safe, regulated
intersection rather than an unregulated intersection
much more often than they actually did. This lack of
knowledge has important implications for parent
ed-ucation and pedestrian skills training since an
in-creased risk of injury for children at unregulated
crosswalks has been noted.’
Parents’ estimates were most accurate in all ages
for the most “lifelike” street-crossing situation, ie, the
at Viet Nam:AAP Sponsored on September 1, 2020
www.aappublications.org/news
realistic street-crossing test. Children also scored
higher on this test than on the simulated test.
Inability of parents to estimate accurately their
children’s knowledge and behavior is not limited to
street-crossing skills. Johnson et al,’2 in their study
assessing daily management of diabetes in children
aged 6-19 years, found that agreement between
chil-dren’s and parents’ reports was highest for the
10-15-year-olds and lowest for the 6-9-year-olds and
1 6-19-year-olds. Yarmey and Rosenstein’3 explored
the extent to which parents can predict children’s
responses to questions about safe and dangerous
sit-uations. Children who participated were 5, 8, and 12
years of age. Similar to the present findings, parents
in general tended to overestimate safety knowledge
of all the children, but this was most evident in the
youngest age group (5-year-olds).
There are several limitations to the study. The small
sample size and relative homogeneity of the sample
precluded our analyzing the data for SES or racial
differences. Likewise, the relatively small number of
fathers participating prevented us from studying
dif-ferences in perceptions between mothers and fathers.
Because of safety concerns subjects could not be
tested under absolutely realistic conditions. The
pres-ence of observers during testing may have had a
positive influence on the children’s behavior, ie, it is
likely that children performed better on the
street-crossing test than they normally would. This effect
was probably more pronounced for the 7- through
8-and 9- through 10-year-olds than the youngest
chil-dren because of increased experience with testing
situations and increased knowledge and practice in
crossing streets. Observations of children conducted
in natural settings by nonobtrusive observers indicate
only 20% to 40% of children correctly perform any
of the tasks during street crossing.’4 This is in contrast
to 70% to 80% correct behavior in the realistic test of
the present study.
Simulated environments might not be the best
method of testing children’s pedestrian skills or
par-ents’ perceptions of their children’s skills. Indeed, if
testing had been carried out in completely natural
settings, parent-child differences might have been
higher. Thus the findings in the present study
prob-ably underestimate the true difference between actual
street-crossing skills and parents’ expectations.
Prevention of childhood pedestrian injuries rests
on a multifaceted approach of which parent education
and pedestrian skills training programs are important
components. Study findings showed that parents of
5- through 6-year-olds had unrealistic expectations
of their street-crossing skills. Overestimations of
chil-dren’s street-crossing skills for 7- through 8- and
9-through 10-year-olds were less evident. This is of
grave concern because children 5-9 years of age are
the group who are at greatest risk for pedestrian
injuries.’ Within this high risk group fatal and
non-fatal pedestrian injuries occur most frequently for the
5- and 6-year-olds compared with 7- and 8- or 9- and
10-year-olds’5 (personal communication in writing
with National Center for Health Statistics, 1991). This
is not a surprising finding when one considers that
the greatest mismatch between parents’ expectations
of their children’s street-crossing skills and their
ac-tual skills occurs at this age. Parents of children in
the youngest age group should be made more aware
of their children’s street-crossing skills. In addition,
they need to be educated about children’s
develop-mental abilities and capabilities as pedestrians. Public
health professionals should develop pedestrian skills
training programs tailored to specific ages with an
emphasis on increased responsibility being given to
children as they get older.
This model of parents’ perceptions of children’s
skills may also be applicable to other childhood injury
problems, as it may identify other age-related
unreal-istic expectations of children’s skill levels and
devel-opmental capabilities that ultimately contribute to
increased risk of childhood injury.
ACKNOWLEDGMENTS
This project was supported by grant I RO1 HD25404 from the
National Institute of Child Health and Human Development and
grant R49/CCR002570 from the Centers for Disease Control.
We thank Valerie Pollet and Valerie Turnbull for assistance in
data collection and Robert Soderberg for computer assistance.
REFERENCES
1. Rivara FP. Child pedestrian injuries in the United States: current status
of the problem, potential interventions, and future research needs. AJDC. 1990;144:692-696
2. Accident Facts. Chicago. IL: National Safety Council; 1988
3. Foot HC, Chapman AJ, Wade FM. Pedestrian accidents: general issues
and approaches. In: Chapman AJ, Foot HC, Wade FM, eds. Pedestrian Accidents. Chichester, England: John Wiley & Sons Ltd; 1982:1-37
4. Thackerey RM, Dueker RL. Child Pedestrian Supervision/Guidance. Wash-ington, DC: US Dept of Transportation. National Highway Traffic Safety Administration; January 1983. DOT HS-806-519
5. Rivara FP, Bergman AB, Drake C. Parental attitudes and practices toward children as pedestrians. Pediatrics. 1989;84: I 017-1021
6. Rivara FP. Epidemiology of childhood injuries. In: Matarazzo JD, Weiss SM, Herd JA, Miller NE, Weiss SM, eds. Behavioral Health: A Handbook of Health Enhancement and Disease Prevention. Chichester, England: John
Wiley & Sons Ltd; 1984:1003-1020
7. Terman LM, Merrill MA. Stanford-Binet Intelligence Scale. Boston, MA:
Houghton-Mifflin; 1970
8. Coppens NM. Cognitive development and locus of control as predictors
of preschoolers’ understanding of safety and prevention classes. I Appl Dev Psychol. 1985;6:43-55
9. Parcel GS, Meyer MP. Development of an instrument to measure chil-dren’s health locus of control. Health Educ Monogr. 1978;6:149-159
I 0. Gayette CH, Conners CK, Ulrich RF. Normative data for revised Conners
Parent and Teacher Rating Scales. /Abnorm Child Psycho!. 1978;6:221-236
I 1. Young DS, Lee DN. Training children in road crossing skills using a roadside simulation. Accid Anal Prey. 1987;19:327-341
12. Johnson SB, Silverstein J, Rosenbloom A, et al. Assessing daily
manage-ment in childhood diabetes. Health Psychol. 1986;5:545-564
13. Yarmey AD, Rosenstein SR. Parental predictions of their children’s
knowledge about dangerous situations. Child Abuse Negl.
1988;12:355-361
14. Rivara FP, Booth CL, Bergman AB, et al. Prevention of injuries to children: effectiveness of a school training program. Pediatrics. 1991;88:770-775
1 5. American Automobile Association. AAA Pedestrian Safety Report. Special
