Effects of Parental Viewing of Children’s Risk Behavior
on Home Safety Practices
Keri J. Brown,1 PHD, Michael C. Roberts,2 PHD, Sunnye Mayes,2 MA, and
Richard E. Boles,2 MS
1
Children’ s Research Institute, Columbus Children’ s Hospital and 2Clinical Child Psychology
Program, The University of Kansas
Objective To examine the effects of parent viewing of their child’s actual risk behavior on
home safety practices. Methods Sixty-one 4-to 7-year-old children and their caregivers
participated in a three session project. Parents were exposed to one of three videos: (a) their
own child with simulated home hazards, (b) a pilot child with hazards, or (c) a control child
development video. Observations of home hazards as well as parent measures of supervision
and vulnerability were completed pre and postintervention. Results Exposure to a video of
a parent’s own child playing with simulated hazards resulted in improved home safety practices.
Exposure to a pilot child interacting with home hazards did not increase parent safety behaviors. No group differences in levels of vulnerability were found. Conclusions Parental
attitudes are an important consideration in designing successful injury interventions. Increasing
parental awareness of their child’s risk to injury may be a valuable tool to change safety
behaviors.
Key words
childhood injury; home safety; parental vulnerability; injury prevention.
[parents] need to understand that injuries are by far the leading
killer of children and that despite the relatively low base rate,
children are more vulnerable to injury than to any other single
threat in their current world.
Peterson & Brown, 1994, p. 309
Unintentional injuries have been identified as a major
threat to the health and well-being of children (Healthy
People 2010, 2000; Roberts & Brooks, 1987). An alarming 91% of unintentional injuries and nearly one-half of
injuries that eventuate in death occur in or around the
place that children should feel safest, their home (National
Safety Council, 1987; Rivara, Calonge, & Thompson,
1989). According to Accident Facts (National Safety
Council, 1999), one fatality occurs every 19 minutes in a
residential setting. Although passive intervention strategies (e.g., flame retardant sleepwear) have effectively
decreased some household hazards (Baker, 1981), some
injury agents do not lend themselves to passive designs.
Thus, active prevention strategies (i.e., those that
require varying levels of parental effort) have emerged
but generally have a much lower success rate than passive strategies (Peterson, Farmer, & Mori, 1987).
There have been a number of widespread campaigns
to encourage various home safety behaviors. For example, one of the most well-publicized prevention practices
is storing poisons where children cannot reach them.
However, one-quarter of families do not properly store
toxic substances (Wortel & de Geus, 1993). More recently,
there has been increased advocacy for proper gun
storage, however, parents have reported that guns are
not stored properly in 40% of gun-owning homes, and
12–13% of guns remain unlocked and loaded (Connor &
Wesolowski, 2003; Farah, Simon, & Kellermann, 1999;
Senturia, Christoffel, & Donovan, 1994).
Despite attempts to increase parental awareness, there
have been several reports of parents underestimating
All correspondence concerning this article should be addressed to Keri J. Brown, Children’s Research Institute,
Columbus Children’s Hospital, Center for Injury Research and Policy, 700 Children’s Way, Columbus, Ohio 43205.
E-mail: [email protected].
Journal of Pediatric Psychology () pp. –, 
doi:./jpepsy/jsi
Advance Access publication February , 
Journal of Pediatric Psychology vol.  no.  © Society of Pediatric Psychology ; all rights reserved.

Brown, Roberts, Mayes, and Boles
children’s risk to injury (Gärling & Gärling, 1993).
Morrongiello and Dayler (1996) found that Canadian
parents were consistently able to identify home hazards
and potential consequences in hypothetical situations,
however, they did not report regularly worrying about
injury. Similarly, Peterson, Farmer, and Kashani (1990)
found that parents report low feelings of vulnerability to
childhood injuries; reporting on a 10-point Likert scale
that they worry less than a “2” about most injury items
and less than a “4” on all items. In a national telephone
survey, parents reported worrying more about kidnapping and drug abuse than childhood injury, although
the latter is a far more likely occurrence (Eichelberger,
Gotschall, Feely, Harstad, & Bowman, 1990).
These inaccurate risk estimations may lead to inadequate safety behaviors, although this has not been
empirically evaluated in the childhood injury literature.
Glik, Kronenfeld, and Jackson and colleagues (1991)
measured the relationship between perceived risk of
childhood injuries and demographic, sociocultural, and
situational variables. They found parents reported
increased home safety-proofing behaviors if they believed
these hazards were a threat to their children.
The relationship between an individual’s feelings of
vulnerability and resulting health beliefs, intentions, and
behaviors has been documented in the social psychology
literature. Individuals have been found to exhibit a
self-serving bias and report that negative life events are
more likely to occur to other people. This “illusion of
unique invulnerability” (Perloff, 1983, p. 47) has been
found in individuals’ perceptions of their physical health
(Weinstein, 1982) and for one’s own risk of injury
(Dolinski, Gromski, & Zawisza, 1987). This bias can be
applied to childhood injury; by extension, a caregiver
may feel that injuries are more likely to happen to other
people’s children. The relatively low base rate of major
injuries may fuel these misperceptions of risk. When
major injuries do not occur, assumptions that injuries
“won’t happen to me or my child” are sustained. In contrast, direct experience with injury appears to increase
awareness in parents (Glik, Kronenfeld, & Jackson, 1991)
although this outcome has not been found consistently
(Peterson, Bartelstone, Kern, & Gillies, 1995).
The Current Study
Given the high rate of unintentional injuries in the
home and the apparent misperceptions that parents have
regarding such incidents, there is a need to increase
parental awareness and understanding of childhood
injury and safety (Eichelberger et al., 1990). It has been
found that when unsupervised, many children interact
with disengaged hazards (Cataldo et al., 1992) and that
parents are likely to underestimate the likelihood of a
hazard-child interaction (Farah, Simon, & Kellermann,
1999; Jackman, Farah, Kellermann, & Simon, 2001).
However, whether “eyewitness” accounts of their child’s
true risk behavior leads to increases in parental safety
behaviors has not been examined.
The current study did not assess the rates of injury
pre and postintervention, rather, used process variables
(e.g., home safety proofing rates) as the primary dependent variable. This methodology was chosen due to the
relatively low base rate of injuries coupled with a short
follow-up and sample size that would have been epidemiologically small to detect significant change in injuries
per se.
Study Overview
In this study, parents were assigned to one of three conditions: the simulated hazard (SHC), video (VC), and
control (CC). The SHC, and to a lesser degree the VC,
was designed to increase the parents’ awareness of their
child’s vulnerability to home injury. The SHC was
intended to provide the parent with a more realistic view
of his or her child’s behavior and increase awareness of
the possibility that the child may come into contact with
a hazard when unsupervised. It was hypothesized that
the parents in the SHC would show increases in safety
behaviors based on this experience. The VC was anticipated to be less salient and therefore would increase
vulnerability to a lesser extent because parents are watching “others” and their experience with hazards. It may
have been more difficult for the parents to believe their
children would act in the same way as the children in
the video (e.g., “but I’ve taught my child better”). Nonetheless, the video condition was hypothesized to affect
vulnerability and, in turn, safety behaviors to a greater
extent than the control condition.
Method
Participants
Sixty-one children (31 males and 30 females) between
four and seven years of age (M = 6.02, SD = 1.13) were
recruited through afterschool programs, daycares, and
preschools. Of the 112 parents who provided contact
information, 72 (64%) enrolled, and 61 (55%) families
completed all three study visits. The most common reason
for termination was lack of time. Exclusion criteria
included child disability or a family injury requiring
Effects of Parental Viewing
hospitalization in the three months prior to study enrollment. Other study measures were collected and were not
utilized for the present study. These measures were of
child temperament, child vulnerability, family environment, parenting stress, and safety rules. This project was
approved by the institutional review board.
1998). Hazards were summed for each room and by hazard type; higher frequency indicated more hazard risk in
the child’s home. Four rooms were used in the analyses:
child bedroom, living room, kitchen and dining area,
and bathroom. An average of hazards by type and room
for each study group is presented in Table I.
Measures
Vulnerability Measure
Caregivers completed 12 items to assess perceptions of
their child’s vulnerability to injury in general (e.g., “The
chances of my child being injured are low” or “The possibility is high that my child will go to the doctor or
emergency room for an injury in the next two years”) as
well as vulnerability to specific household injuries (e.g.,
“My child is likely to touch a hot object” or “It is possible that my child will be cut by something in my home
and require stitches”). Higher scores on this measure
indicated higher feelings of their child’s vulnerability to
injury.
Family Information Form
Child’s age, ethnicity, and gender, as well as family income,
parental education and occupations were acquired with
a demographic form. Socioeconomic status was computed by calculating an income-to-needs ratio (total
family earnings divided by the current poverty rate for
the family size). In addition, parents were cued in a semistructured interview to provide child injury data for the
following injury categories: burn, head injury, poisoning, lacerations, choking, and water-related injuries. Injury
information included the injury type and severity (e.g.,
hospitalization). The frequency of reported injuries were
summed for a total child injury score.
The Home Assessment Prevention Inventory – Revised
(HAPI-R)
The HAPI-R is a home hazards checklist for items in six
categories: poisons, fire and electrical, firearms, drowning
hazards, and suffocation (Mandel, Bigelow, & Lutzker,
Supervision Vignette Measure (SVM)
This measure was designed for this study to assess
parental perceptions of common supervision practices
for 4-to 7-year-old children and was inspired by The
Supervision Rules Questionnaire (Peterson, Ewigman,
& Kivlahan, 1993). In the original measure, parents
reported their opinions regarding the acceptable amount
Table I. Mean Number of Home Hazards at Baseline and at Study Conclusion for Study Groups
Hazard Type
Simulated Hazard
Video Condition
Control Condition
4.00 (3.33)a
2.90 (2.55)
3.25 (3.45)
2.71 (3.33)b
2.50 (1.96)
3.50 (3.59)
1.67 (1.24)
1.20 (0.95)
1.75 (1.52)
1.12 (1.27)
0.80 (0.83)
1.65 (1.63)
0.78 (1.11)
0.45 (0.69)
0.95 (0.95)
0.24 (0.56)
0.50 (0.89)
0.80 (1.10)
0.39 (0.78)
0.20 (0.41)
0.10 (0.31)
0.24 (0.56)
0.15 (0.37)
0.05 (0.22)
2.94 (2.36)
2.35 (2.44)
2.10 (2.73)
1.71 (2.02)
1.65 (1.53)
2.73 (2.58)
Bathroom
2.65 (2.18)
1.85 (1.93)
2.44 (2.57)
1.76 (1.64)
1.90 (1.99)
2.56 (2.53)
Parent Bedroomc
0.86 (0.69)
0.73 (0.80)
1.00 (1.10)
0.63 (0.74)
0.57 (0.76)
0.86 (1.10)
Living Room
1.00 (0.75)
0.50 (0.95)
1.00 (1.59)
0.59 (1.50)
0.59 (1.50)
0.95 (0.95)
0.06 (0.25)
0.05 (0.23)
0.65 (1.50)
None found
0.05 (0.23)
0.77 (1.82)
Poisons
Sharp Objects
Fire Hazards
Firearms
Hazards by Room
Kitchen/Dining
Child’s Bedroom
The values listed in parentheses represent standard deviations.
a
The first value given in each item represent baseline hazard scores on 58 consenting homes.
b
The values on the second row represent postintervention hazard scores on the 57 consenting homes.
c
35 parent bedrooms were observed. These data were not used in total hazard analysis.


Brown, Roberts, Mayes, and Boles
of time that children of different ages could remain
unsupervised in various risk settings. A subset of these
items was expanded into vignettes. Four vignettes described home hazard scenarios in which medication, a
gun, a lit candle, or cleaning agents were present and the
hypothetical parent in the vignette is distracted or called
away from his or her child. Following each vignette, a
list of six supervision choices were read aloud and parents rated the acceptability of various parental behaviors
from “1” (parent behavior is completely unacceptable) to
“5” (completely acceptable parental supervision behavior). Vignettes were gender-matched to the study child.
Parental Affect Scale
Parents rated the extent to which they felt eight emotions (e.g., surprise, anger, pleasure) on a scale of 0 to 5,
with 0 being no emotion at all. The item responses for
“pleasure” and “pleased” were aggregated and averaged
to create a positive affect score, while “frightened” and
“concerned” were averaged to create a negative affect
score for group comparisons. Affect was reported immediately following the exposure to the respective videos at
Time 2.
Procedure
Parents were asked to complete two in-home and one
clinic-based study visit; each visit was 45–60 minutes in
length. All parents completed an informed consent form;
children gave assent. Each family received a food
voucher of $5 value for in-home sessions, and $10 value
for the clinic visit.
The following measures were counterbalanced and
presented at the initial home visit: Family Information
Form, Vulnerability Measure, and the SVM. Next, parents were asked to sign a separate consent form for a
home check to “take a general look at things that may
promote or impair child health” and a HAPI-R was completed for all consenting families.
Following the first visit, families were randomly
placed into one of three conditions: (a) a simulated hazard
condition, (b) a video condition (VC), or (c) a control
group (CC). Two researchers were present at each condition so that parents and children could complete their
separate study roles.
Simulated Hazard Condition
Upon arrival to the university clinic, the caregiver examined the simulated hazard room and rated his or her
child’s risk using the SHS. The hazard room housed the
following disengaged hazards: a cigarette lighter (plugged
with glue), a pill dispenser (with candy inside), a slightly
open buck-knife (grinded to sharpness of a butter knife),
a soldering iron (with electrical wires cut), a CO2
powered BB-gun (with no BBs and trigger lock on), and
spray cleaner (bottle containing nontoxic colored water).
In addition to the simulated hazards, the room contained seven common toys, including a puzzle, Etch-aSketch™, and Play-doh™. While parents completed
study questionnaires, the child was introduced to the
hazard playroom by a researcher. Soon after entering the
playroom, the child was asked to wait in the room while
the researcher retrieved materials. The child’s play was
observed through a one-way mirror and videotaped for
15 minutes. The investigator recorded the frequency and
duration of each hazard contact and noted the time
marker for the contact(s). Immediately following the
taping, this information was given to a second research
assistant who cued the tapes and subsequently showed
these interactions to the child’s parent(s). Parents viewed
the child’s hazard contacts (but no other room activity)
in order not to reinforce that their children did not get
hurt when playing with the objects. Children in the SHC
demonstrated an average of 2.26 hazard interactions,
with 68.4% of children touching at least one hazard
item. In order of frequency (with percentages in parentheses), children in the simulated hazard condition
touched the gun (47.4%), knife (42.1%), soldering iron
(26.3%), cigarette lighter (21.1%), medication (21.1%),
and cleaner (10.5%).
Video Condition
Parents were shown the simulated hazard room and
asked to complete the SHS. In the same fashion as the
SHC, children were introduced to the room and child
behavior was recorded as part of another study. Parents
were not told of their child’s behavior and instead
watched a video of a same-sex pilot child in the hazard
room. The 2-minute video depicted a 5-year-old of the
same gender as the child participant interacting with
three hazard room items: the pills, the lighter, and the
gun.
Control Condition
Parents and children were separated upon arrival to the
clinic. Parents completed the SHS (without seeing the
hazard room or contents) and then viewed a 2-minute
clip of a child development video. In this condition,
children were not exposed to the simulated hazard environment, but questions were asked about household
rules for another study.
After viewing each respective video presentation, all
parents completed the Affective Rating Scale and the
Vulnerability Measure. Parents were given the opportunity to choose from donated household items, including
Effects of Parental Viewing
safety-related (e.g., cabinet locks) and non-safety related
products (e.g., pens) of approximate equal monetary
values. This procedure removed financial or transportation barriers to obtaining safety products and ensured
that all parents had equal access to these items should
they wish to install the items in their home.
Two weeks following the clinic visit, the investigator examined the home environment with the HAPI-R
for a second time and administered the SVM. At conclusion of the visit, parents were told of the home safety
nature of the study, given specific information about
hazards in their home, and given the opportunity to ask
safety questions. Parents were told that one of the most
effective means of keeping their children safe at home is
keeping hazards inaccessible.
Results
Sample Characteristics
Mothers most often completed the study measures
(82%). The background of the sample was representative of the Midwestern target area: 77% Caucasian, 8.5%
African-American, 6% Bi-racial, 5% Native American,
1.5% Hispanic, and 1.5% Asian. As seen in Table II,
there were no significant group differences on major
demographic variables.
Multivariate Analyses of Parental Safety Measures
To examine the effect of group membership on resulting parental home safety behavior, a one-way multivariate analysis of covariance (MANCOVA) was conducted.
The two dependent variables were supervision scores
and the frequency of safety hazards at postintervention.
Pretest scores on supervision vignettes and home hazard assessments were entered as covariates to control
for parental practices prior to the intervention. The
multivariate design was employed to assess overall effects
and to control for inflations in alpha level that can
result from multiple tests (Bock & Haggard, 1968). The
MANCOVA was significant, Wilks’ Λ = .761, F (4, 100) =
3.649, p = .008. As indicated by partial eta squared,
group accounted for 13% of the variance of the dependent variables after pretest scores were taken into
account. Pretest scores for supervision and baseline
HAPI-R scores accounted for a large proportion of the
variance, both Fs (2, 50) > 45.73, both ps < .001,
accounting for 85.3% and 64.9% of the variance,
respectively.
Table II. Demographic Statistics for Study Groups
Variable
Age
SHC (n = 19)
6.2 (1.2)
VC (n = 20)
5.9 (1.0)
CC (n = 22)
6.0 (1.2)
Test Statistic
p-value
a
.738
.710b
.701
.315
Gender
% male
58
55
46
% 4–5
42
60
50
% 6–7
58
40
50
Age of child
1.26b
.533
Gross family income (%)
Under 10,000
11
10
9
10,001–20,000
11
15
32
20,001–40,000
16
20
9
40,001–50,000
0
15
14
50,001–75,000
32
15
5
Over 75,000
32
35
23
Income to Needs Ratio
3.2 (1.8)
2.6 (1.7)
2.7 (1.9)
.643a
.530
Children in Household
2.2 (1.1)
2.3 (.55)
2.0 (.84)
.648a
.527
Maternal educationc
15.8 (2.0)
15.5 (2.3)
15.2 (2.1)
.493a
.613
Paternal educationc
15.1 (2.4)
14.9 (2.2)
15.1 (2.8)
.028a
.972
Maternal age
35.6 (7.7)
34.0 (3.9)
31.5 (6.5)
2.20a
.120
Paternal age
37.9 (9.6)
36.3 (5.4)
33.9 (8.0)
1.05a
.358
2.51a
.090
Pretest Supervision Score
3.38 (0.51)
3.04 (0.48)
2.94 (0.46)
Pretest Home Hazards
6.01 (4.80)
4.70 (3.90)
5.25 (3.90)
Mean (SD) and percentages of sample given.
a
Differences in group values assessed using a One-Way ANOVA.
b
Differences in frequencies assessed using Chi-Square.
c
Parental education measured in years of school completed.
.441a
.646


Brown, Roberts, Mayes, and Boles
Table III. Means, Standard Deviations, and Follow-up Analysis of Covariance (ANCOVA) Results for Parent Safety Behaviors
Group
Dependent Variable
Simulated Hazard
ANCOVA
Video Condition
Control Condition
F(2, 56)
p
3.61
.034
4.91
.011
Post-Supervision Score
Means
3.04 (0.61)
2.92 (0.48)
2.98 (0.68)
Adjusted Means†
2.83a (0.09)
2.94 (0.08)
3.14a (0.08)
Post Home Hazards Score
Means
4.00 (4.61)
4.02 (3.50)
5.05 (5.01)
Adjusted Means†
3.38a (0.43)
4.63 (0.39)
5.00a (0.40)
a
Pairwise comparisons with like superscripts were statistically significant using Holm’s sequential Bonferroni procedure.
†Standard Errors are reported for adjusted means.
Subsequent univariate one-way analyses of covariance (ANCOVA) were significant for group and were
followed by post hoc t tests. Both the supervision and
home hazard ANCOVAs were significant for group,
respectively. Parents in the simulated hazard group had
significantly fewer hazards on the HAPI-R and had more
conservative supervision scores than the control group.
The simulated hazard condition did not have scores that
were significantly different than the video condition.
Likewise, the video condition and the control condition
were not statistically different on the outcome variables,
as indicated in Table III.
Perceptions of vulnerability
Pearson correlation coefficients indicated that parental
report of vulnerability following the intervention was
not significantly related to the frequency of home hazards at follow-up, r = .246, p = .076. Additionally, parental reports of vulnerability following the intervention
was not found to be related to parental reports on the
supervision measure, r = −.004, p = .977. Given these
unexpected results, further testing of the relationship
between vulnerability and outcome via planned regression analyses was not warranted.
An ANCOVA was conducted to examine the group
effect on reported vulnerability. The independent variable, group, included three levels: simulated hazard,
video, and control condition. The dependent variable
was the reported vulnerability following intervention
and the covariate was pretest vulnerability scores. The
ANCOVA was marginally significant, as indicated in
Table IV. Planned pairwise comparisons were conducted
to evaluate the differences among the group means as
guided by the research question. In order to control for
Type 1 error across all comparisons, the Holm’s sequential Bonferroni method was utilized. There were no significant differences in vulnerability between any two
groups.
Affective rating
A multivariate analysis of variance was conducted to
account for multiple affective comparisons. The MANOVA
yielded significant group effect, Wilks’ Λ = .490, F (4,
100) = 10.72, p < .001. The multivariate η2 was quite
strong, .30. Table V contains the means and the standard
deviations for the dependent variables for the three groups.
Analysis of variance (ANOVA) for each independent variable was conducted as follow-up tests to the
MANOVA. Using the Bonferroni method, each ANOVA
was tested at the .025 level. The ANOVAs for the positive affective and the ANOVA for the negative affective
scores were significant, F (2, 51) = 12.88, p < .0001,
partial η2 = .34 and F (2, 51) = 10.82, p < .0001, partial
η2 = .30, respectively.
Post hoc analyses to the univariate ANOVA for the
affective scores were conducted. Because the variances
among the three groups ranged from 1.2 to 3.5 for the
positive affective scores, we chose not to assume that the
variances were homogenous and conducted post hoc
comparisons using the Dunnett’s C test, a test that does
Table IV. Means, Standard Deviations, and Analysis of Covariance (ANCOVA) Results for Vulnerability
Group
Variable
ANCOVA
Simulated Hazard
Video Condition
Control Condition
2.47 (.78)
2.50 (.77)
2.00 (.70)
Means
2.43 (.80)
2.73 (.76)
1.99 (.50)
Adjusted Means
2.36 (.13)
2.63 (.14)
2.18 (.14)
Pretest Vulnerability
F(2, 50)
p
2.61
.083
Post Vulnerability Score
Standard Errors are reported for adjusted means.
Effects of Parental Viewing
Table V. Group Differences in Reported Affect Following Intervention
Positive Affect
Group
Negative Affect
M
SD
Simulated Hazard
5.58*
3.5
3.42
3.5
Video Condition
.78
1.2
6.33*
3.1
2.59
3.4
1.71
2.6
Control Condition
M
SD
*Means are significantly different at p < .05 than groups without asterisks as
determined by Dunnett’s C.
not assume equal variances among the three groups.
Unexpectedly, the VC reported higher negative emotions
(e.g., fear) while the SHC reported significantly higher
positive emotions (e.g., pleasure) at the conclusion of the
intervention, as seen in Table V.
Discussion
This study sought to increase home safety behaviors by
providing parents with insurmountable “eyewitness”
evidence that their children are likely to handle home
hazards when left accessible. As has been reported in
previous studies (Connor & Wesolowski, 2003), many
children will touch accessible simulated hazards. Unlike
previous studies, this project attempted to capture quantifiable change in parent behaviors based on this hazard
room experience. Home observations pre and postintervention indicated that parents who viewed their own
child’s risk behavior decreased their home hazards. Further, parents in the simulated hazard condition also
reported more conservative responses to the acceptability of various common supervision practices. Thus, the
intervention was successful in increasing the home
safety behaviors of parents and lends support to the
work of Glik et al. (1991) suggesting parental home
safety vigilance increases when parents believe injuries
are a real threat. As proxy variables for home injury, it is
hoped that the reported increase in awareness of supervision and decrease in home hazards for the SHC will
lead to fewer child injuries in the home.
A cautionary note is necessary, however, although it
was assumed that increased parental vulnerability would
be the mechanism by which this change occurred,
vulnerability was not significantly elevated immediately
following the intervention. It is theoretically possible
that a “sleeper” effect for vulnerability was present, with
increased feelings of vulnerability emerging after parents
returned to their home environments. However, the
validity of this argument cannot be substantiated by this
study. Alternatively, parents in the simulated hazard
group may have been more vigilant about safety because
of increased attention at the clinic visit. Anecdotally, it
was noted that simulated hazard parents were extremely
interested in their child’s behavior in the room; thus, the
novelty of the experience may have made safety messages more meaningful to parents.
The hypothesis that parents in the video condition
would also report significant increases in safety behaviors was not supported, although safety behavior scores
were elevated and not statistically different than the
simulated hazard condition. Without the undeniable
evidence of seeing their own child interact with hazards,
these parents may have succumbed to the “illusion of
unique invulnerability” (Perloff, 1983, p. 47). In addition, although the child participant and the model child
in the video were matched for gender, the videos were
not specific to child ethnicity or age. A video of a child
that was matched to all demographic variables of the
study child may have resulted in more pronounced
change in parental safety behavior for the video condition. Given that the safety behaviors of families in the
video condition were elevated and not statistically different than the condition whereby parents saw their own
child interact with hazard props, a video intervention to
reduce childhood home injuries should continue to be
researched. A video that is tailored to the specific environmental risks present for a particular child (e.g., guns
in the home) and matched on child and family demographics may be a worthwhile and cost-effective method
of increasing parent awareness of risk.
As a manipulation check of the intervention, parents were asked to rate their affect immediately following the intervention experience. Unexpectedly, parents
in the simulated hazard group reported being “pleased”
and “proud,” whereas parents in the video condition
reported feelings of “fear” and “concern.” Although a
large percentage of children in this condition did interact with a potentially hazardous item, parents in general
were not significantly alarmed. This finding demonstrates
the importance of testing researcher assumptions.
There are several possible explanations for the lack
of predicted increase in vulnerability and reportedly
positive affect in the SHC. First, children did not touch
as many hazardous items on average as the pilot children in the video. Thus, the effect of viewing a child
interacting with hazards may not have been as powerful
of a manipulation in the simulated hazards condition as
in the video condition. This difference may have led to
the variations in reported affect for the parents in the
SHC and the VC. Second, parents may have attempted to
rationalize their children’s hazardous behaviors. For
example, the parents may have been relieved that the


Brown, Roberts, Mayes, and Boles
child only touched a few hazards or did not touch the
hazard that they perceived to be the most dangerous
(e.g., the gun). In retrospect, it may have been more
salient (and less “anxiety relieving”) to have shown the
parent only the hazard interactions without parental
knowledge of how many potential hazards were made
available to the child. Third, hazard “interaction” was
operationally defined as any contact with a hazard item,
regardless of the duration of contact. Parents, however,
may define “risky behavior” as more prolonged contact.
Additionally, these parents may have believed that their
presence could have intercepted the child-hazard interaction to prevent injury. The impact of viewing their
child interacting with hazards, therefore, would have
been mitigated. Although positive feelings were recorded
for the SHC parents, these feelings did not impede their
ability to take protective safety actions, as seen by the
differences in group safety practices at the follow-up
home visit.
Theoretical Considerations
The hazard room experience resulted in quantifiable
changes in parent behaviors. Although improvements in
home safety were found, hazards were still present in
every home at follow-up, regardless of group membership. The level of hazard reduction in the present study
was less notable than the percentage of hazard reduction
reported for participants in an in-home, multicomponent, four session safety training program designed for
families who were either at risk for maltreatment or families of children who were victims of maltreatment
(Gershater-Molko, Lutzker, & Wesch, 2003). Similarly,
King et al. (2001) reported that a single home-visit with
provision of educational material and intervention resulted
in significant home safety modifications at 4-months
postintervention. However, that study used parent reports
via telephone interviews rather than follow-up home
visits. Nonetheless, the observed lack of universal adoption of safety devices in the present study, even when
safety information and devices were free and accessible,
may have been related to the amount of effort needed on
behalf of the parent. The parental vigilance and continuous demands of the more active measures (e.g., keeping
frequently used medicines properly stored) may have
been more resistant to change, whereas other safety
strategies (e.g., gun locks) may demand less daily effort.
Studies have identified higher rates of compliance for
injury prevention strategies that demand fewer repeated
actions (Dershewitz, 1979).
Wilson and Baker (1987) proposed a continuum of
passive to active safety behaviors. It is conceivable that
increasing parental understanding of real probability of
hazard interactions might increase parental willingness
to engage in safety behaviors that are higher on the continuum. However, without accurate parental awareness
of their child’s vulnerability to injury, parents might be
unmotivated to perform safety behaviors beyond passive
intervention. One exception might be those interventions
that have been found effective by supplying secondary
reinforcers, thereby eliminating the need to attend to
hypothesized intermediate variables (Roberts & Turner,
1986).
Study Limitations
Although the results of the present study provide some
promising contributions to the injury literature, a number of limitations require that the results be interpreted
with some caution.
Even though the study was presented in terms of
“health promotion and safety,” parents were likely to
deduce that it was primarily targeted to home safety. A
self-selection bias may have been present in those parents who elected to participate. The self-selection bias
may have also extended to the families who chose to
participate in all 3 visits. Nonetheless, this understanding of the study was present for each condition and this
potential confound should not have affected the significant group effects. Another potential source of bias was
the unknown parent–child communications regarding
this project. Although parents were asked to refrain
from discussing aspects of the study with their children
until the final study session, parents have been found to
rely heavily on talking to their children as an effort to
keep their children safe (Gärling, 1989).
Other limitations are due to measurement and statistical power issues. There is a lack of well-validated
injury vulnerability instruments in the literature, thus
the measure of vulnerability was constructed for the
present study and has limited psychometric strength. In
addition, our modest sample size may not have had
sufficient power to detect significant differences in vulnerability. A retrospective power analysis with the present
sample size and alpha set at .05 indicated that power was
.59 for the univariate ANOVA on vulnerability, below
the conventional .80 mark.
Future Directions
The results of this study suggest a number of future
directions for research. First, additional research on the
role of parental perceptions of risk and vulnerability to
childhood injury prevention is warranted. Although the
malleability of parent risk awareness was established in
Effects of Parental Viewing
the current study, questions as to its overall utility in
prevention interventions remain. Future research should
attempt to gain a better understanding of the parental
characteristics that might motivate parents to preventive
action by increasing injury risk awareness. A series of
related questions would include the following: Are there
certain periods in the child’s development that parents
feel more or less vulnerable to injury? Are there particular individual and/or cultural beliefs that make parents feel less vulnerable? Finally, does adding a risk
awareness component to more traditional interventions
that have had low success rates increase the parental
adoption of the suggested safety practices?
Because the study intended to examine vulnerability changes as a result of treatment condition, any gender differences were less important to the levels of
vulnerability affected. Future studies may need to consider the meaning of the relationship between gender,
vulnerability, and risk awareness in more depth. That
is, rather than just finding a relationship of gender to
vulnerability and behavior, the theoretical interpretations might be important. For example, what information are parents using to make supervisory decisions,
and what are parents telling themselves or reassuring
themselves about their children that influences how
they structure their children’s environments or control
their behavior.
Providing parents with realistic views of their children’s behavior may be a credible motivational strategy,
however, a more cost and time efficient method of delivery should be investigated. Further, longitudinal research
is needed to determine the stability of household changes
postintervention and whether the increased vigilance in
parents translates into reduction of childhood injuries.
Observation of supervision behaviors would improve
upon this investigation’s use of self-reported vignettes.
This study leads credence to the complex nature of
unintentional injury in childhood. The findings confirm
that there are parental attitudes about safety that preclude their ability to make changes to keep their children safe. Future studies should continue to explore
behavioral and attitudinal variables that may impact the
successful implementation of safety programs.
Acknowledgments
This study was funded by the Rebecca Routh Coon Injury
Prevention Grant (now entitled the Lizette PetersonHomer Memorial Injury Grant) awarded by the Society
of Pediatric Psychology. The authors wish to acknowledge Barbara Morrongiello, PHD, of the University of
Guelph for her thoughtful comments in the early stages
of this research.
Received December 15, 2003; revisions received June 7,
2004, and August 25, 2004; accepted August 26, 2004
References
Baker, S. P. (1981). Childhood injuries: The community
approach to prevention. Journal of Public Health
Policy, 3, 235–246.
Bock, R. D., & Haggard, E. A. (1968). The use of
multivariate analysis of variance in behavioral
research. In D. Whitla (Ed.), Handbook of
measurement and assessment in behavioral sciences
(pp. 100–142). Boston: Addison-Wesley.
Cataldo, M. F., Finney, J. W., Richman, G. S., Riley, A.
W., Hook, R. J., Brophy, C. J., & Nau, P. A. (1992).
Behavior of injured and uninjured children and
their parents in a simulated hazardous setting.
Journal of Pediatric Psychology, 17, 73–80.
Connor, S. M., & Wesolowski, K. L. (2003). They’re
too smart for that: Predicting what children
would do in the presence of guns. Pediatrics, 111,
e109–e114.
Dershewitz, R. A. (1979). Will mothers use free
household safety devices? American Journal of
Diseases of Children, 133, 61–64.
Dolinski, D., Gromski, W., & Zawisza, E. (1987).
Unrealistic pessimism. Journal of Social Psychology,
127, 511–516.
Eichelberger, M. R., Gotschall, C. S., Feely, H. B.,
Harstad, P., & Bowman, L. M. (1990). Parental
attitudes and knowledge of child safety: A National
Survey. American Journal of Disease of Children, 144,
714–720.
Farah, M. M., Simon, H. K., & Kellermann, A. (1999).
Firearms in the home: Parental perceptions.
Pediatrics, 104, 1059–1063.
Gärling, A. (1989). Parents’ heuristics for judging
children’s accident risk. Scandinavian Journal of
Psychology, 30, 134–145.
Gärling, A., & Gärling, T. (1993). Mothers’ supervision
and perception of young children’s risk of unintentional injury in the home. Journal of Pediatric
Psychology, 18, 105–114.
Gershater-Molko, R. M., Lutzker, J. R., & Wesch, D.
(2003). Project SafeCare: Improving health, safety,
and parenting skills in families reported for, and
at-risk for child maltreatment. Unpublished
manuscript.


Brown, Roberts, Mayes, and Boles
Glik, D. C., Kronenfeld, J., & Jackson, K. (1991).
Predictors of risk perceptions of childhood injury
among parents of preschoolers. Health Education
Quarterly, 18, 285–301.
Healthy People 2010. (2000). Centers for Disease Control
and Prevention. Retrieved on October 1, 2003 from
http://www.health.gov/healthypeople/document.
Jackman, G. A., Farah, M. M., Kellermann, A. L., &
Simon, H. K. (2001). Seeing is believing: What do
boys do when they find a real gun? Pediatrics, 107,
1247–1250.
King, W. J., Klassen, T. P., LeBlanc, J., Bernard-Bonnin,
A. C., Robitaille, Y., Pham, B., et al. (2001). The
effectiveness of a home visit to prevent childhood
injury. Pediatrics, 108, 382–388.
Mandel, U., Bigelow, K. M., & Lutzker, J. R. (1998).
Using video to reduce home safety hazards with
parents reported for child abuse and neglect. Journal
of Family Violence, 13, 147–162.
Morrongiello, B. A., & Dayler, L. (1996). A communitybased study of parents’ knowledge, attitudes and
beliefs related to childhood injuries. Canadian
Journal of Public Health, 87, 383–388.
National Safety Council. (1987). Accident facts. Chicago,
IL: National Safety Council.
National Safety Council. (1999). Accident facts. Chicago,
IL: National Safety Council.
Perloff, L. S. (1983). Perception of vulnerability to
victimization. Journal of Social Issues, 39, 41–61.
Peterson, L., Bartelstone, J., Kern, T., & Gillies, R.
(1995). Parents’ socialization of children’s injury
prevention: Description and some initial parameters. Child Development, 66, 224–235.
Peterson, L., & Brown, D. (1994). Integrating child
injury and abuse-neglect research: Common
histories, etiologies, and solutions. Psychological
Bulletin, 116, 293–315.
Peterson, L., Ewigman, B., & Kivlahan, C. (1993).
Judgments regarding appropriate child supervision
to prevent injury: The role of environment risk
and child age. Child Development, 64, 934–950.
Peterson, L., Farmer, J., & Kashani, J. H. (1990).
Parental injury prevention endeavors: A function
of health beliefs? Health Psychology, 9, 177–191.
Peterson, L., Farmer, J., & Mori, L. (1987). Process analysis of injury situations: A complement to epidemiological methods. Journal of Social Issues, 43, 33–44.
Rivara, F. P., Calonge, N., & Thompson, R. S. (1989).
Population-based study of unintentional injury
incidence and impact during childhood. American
Journal of Public Health, 79, 990–994.
Roberts, M. C., & Brooks, P. H. (1987). Children’s
injuries: Issues in prevention and public policy.
Journal of Social Issues, 43, 1–12.
Roberts, M. C., & Turner, D. S. (1986). Preventing
death and injury in children: A synthesis of child
safety seat efforts. Health Education Quarterly, 11,
181–193.
Senturia, Y. D., Christoffel, K. K., & Donovan, M. (1994).
Children’s household exposure to guns: A pediatric
practice-based survey. Pediatrics, 93, 469–475.
Weinstein, N. D. (1982). Unrealistic optimism about
susceptibility to health problems. Journal of
Behavioral Medicine, 5, 441–460.
Wilson, M., & Baker, S. (1987). Structural approach to
injury control. Journal of Social Issues, 43, 73–86.
Wortel, E., & de Geus, G. H. (1993). Prevention of
home related injuries of pre-school childrens’ safety
measures taken by mothers. Health Education
Research: Theory and Practice, 8, 217–231.