The Influence of Interactions with Dogs on Affect, Anxiety, and

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=hcap20

Journal of Clinical Child & Adolescent Psychology

ISSN: 1537-4416 (Print) 1537-4424 (Online) Journal homepage: http://www.tandfonline.com/loi/hcap20

The Influence of Interactions with Dogs on Affect,

Anxiety, and Arousal in Children

Molly K. Crossman, Alan E. Kazdin, Angela Matijczak, Elizabeth R. Kitt &

Laurie R. Santos

To cite this article: Molly K. Crossman, Alan E. Kazdin, Angela Matijczak, Elizabeth

R. Kitt & Laurie R. Santos (2018): The Influence of Interactions with Dogs on Affect,

Anxiety, and Arousal in Children, Journal of Clinical Child & Adolescent Psychology, DOI:

10.1080/15374416.2018.1520119

To link to this article: https://doi.org/10.1080/15374416.2018.1520119

Published online: 30 Oct 2018.

Submit your article to this journal

View Crossmark data

The Inﬂuence of Interactions with Dogs on Affect,

Anxiety, and Arousal in Children

Molly K. Crossman, Alan E. Kazdin, Angela Matijczak, Elizabeth R. Kitt, and

Laurie R. Santos

Department of Psychology, Yale University

Interactions with animals represent a promising way to reduce the burden of childhood

mental illness on a large scale. However, the speciﬁc effects of child–animal interactions are

not yet well-established. This study provides a carefully controlled demonstration that

unstructured interactions with dogs can improve clinically relevant symptoms in children.

Seventy-eight children (55.1% female, 44.9% male) ages 10 to 13 (M = 12.01, SD = 1.13)

completed the Trier Social Stress Test for Children, followed by (a) interaction with a dog,

(b) a tactile-stimulation control condition, or (c) a waiting control condition. The Positive

and Negative Affect Schedule for Children, Short Form and the State/Trait Anxiety

Inventory for Children were completed at baseline and posttest, and salivary cortisol was

assessed at 5 time points. Adjusting for baseline scores, participants in the experimental

condition showed higher scores on the Positive Affect scale than participants in both control

conditions and lower scores on the State/Trait Anxiety Inventory for Children than partici-

pants in the waiting control condition at posttest. Negative affect was not assessed reliably,

and we detected no effect of the interactions on salivary cortisol, as measured by area under

the curve with respect to ground. Brief, unstructured interactions with dogs boosted chil-

dren’s positive emotions and reduced anxiety. Additional research is needed to further clarify

which features of the interactions produce these beneﬁts and the extent to which interactions

with animals offer beneﬁts that exceed the effects of other common coping strategies,

activities, and interventions.

Mental illness is the leading source of childhood disability

and a top contributor to the overall burden of disease

(Erskine et al., 2015). In the United States, approximately

one in ﬁve children are affected by mental illness each year,

with an annual ﬁnancial cost of approximately $247 billion

(Centers for Disease Control, 2013; Kessler et al., 2005).

These estimates do not include consideration of subthres-

hold symptoms, which add considerably to the burden of

disease (Balzás et al., 2013). Despite the considerable pre-

valence and consequences of childhood mental illness and

subthreshold symptoms, roughly 80% of children in need

do not receive treatment (Kataoka, Zhang, & Wells, 2002).

Simple, scalable approaches to reducing the burden of

childhood mental illness and subthreshold symptoms are

urgently needed.

Animal-assisted activities (AAAs) represent a promising

type of intervention to bridge the gap between children in

need of mental health services and those receiving treat-

ment. AAAs are unstructured programs that provide parti-

cipants with opportunities to interact with animals (most

commonly domestic dogs; Canis familiaris), with the goal

of improving mental health and alleviating stress

(International Association of Human–Animal Interaction

Organizations, 2013). AAAs are promising as a method

for improving children’s mental health because they are

appealing, enjoyable, credible, low cost, and easy to imple-

ment (Crossman & Kazdin, 2015). Indeed, AAAs are

already prevalent as a strategy for improving children’s

mental health. Prominent examples include AAAs for

pediatric patients, child victim witnesses, children who

have experienced traumatic events, students learning to

read, and even those without any special status at all. The

Correspondence should be addressed to Molly K. Crossman,

Department of Psychology, Yale University, Box 208205, New Haven,

CT 06520-8205. E-mail: molly[email protected]

Color versions of one or more of the ﬁgures in the article can be found

online at www.tandfonline.com/hcap.

Journal of Clinical Child & Adolescent Psychology, 00(00), 1–14, 2018

ISSN: 1537-4416 print/1537-4424 online

DOI: https://doi.org/10.1080/15374416.2018.1520119

rationale is that contact with the animals may improve

mood and affect and alleviate anxiety and stress by promot-

ing emotion-focused coping, serving as sources of social

support, providing enjoyable experiences, and/or transmit-

ting positive emotions through emotional contagion

(Crossman, 2017; Kruger, Trachtenberg, & Serpell, 2004).

Preliminary evidence in support of the effects of AAAs

on children’s affect, anxiety, and stress-related symptoms

comes from a small number of rigorously designed studies

(e.g., Beetz, Julius, Turner, & Kotrschal, 2012; Kerns,

Stuart-Parrigon, Coifman, Van Dulmen, & Koehn, 2017;

Kertes et al., 2017). These studies have focused primarily

on interactions with dogs and have documented the effects

of these interactions on self-reported, behavioral, and phy-

siological outcomes. However, the majority of research in

this area has focused on adults (Nimer & Lundahl, 2007;

Souter & Miller, 2007). In addition, most studies have

focused on whether interactions with dogs buffer against

the negative effects of stress (i.e., prevent increases in

symptoms during stress exposure; Beetz et al., 2012;

Kerns et al., 2017; Kertes et al., 2017; Krause-Parello,

Thames, Ray, & Kolassa, 2018; Vagnoli et al., 2015).

Although some AAAs focus on supporting children during

stress exposure (e.g., while giving testimony in court),

many focus on alleviating symptoms after exposure to an

acute stressor (e.g., at the sites of tragedies/natural disas-

ters), or among children facing conditions of chronic stress

(e.g., children undergoing long-term hospitalizations), and

it is not yet clear whether ﬁndings on the stress-buffering

effects of AAAs will apply in situations where the goal is to

promote recovery following a stressful event. Most impor-

tant, the bulk of this research has been exploratory in nature

(e.g., Kaminski, Pellino, & Wish, 2002; Tsai, Friedmann, &

Thomas, 2010). As a result, the effects of AAAs on chil-

dren are not yet well-established.

The preliminary state of the literature has resulted in one

particularly important gap in our understanding of the

effects of AAAs for children. Namely, it is not yet clear

whether the interactions with the animals are really the

active ingredient (Crossman, 2017; Marino, 2012). In

other words, we cannot yet say whether the improvements

observed in studies of AAAs can really be attributed to the

interactions with the animals, rather than to other aspects of

the interventions or to more general factors, such as

engagement in any distracting or appealing activity

(Marino, 2012). The issue is that the majority of studies

have relied exclusively on no-treatment control conditions,

if they have included any control condition at all (e.g.,

Braun, Stangler, Narveson, & Pettingell, 2009; Kerns

et al., 2017; Vagnoli et al., 2015). Studies that use no-

treatment control conditions demonstrate that AAAs can

alleviate clinically relevant symptoms in children and sug-

gest that the observed improvements are not simply due to

extratherapeutic factors. However, these studies do not

clarify what speciﬁc component of the intervention pro-

duced the improvements.

A handful of studies have attempted to isolate the effects

of the interactions with the animals. The most common

approach has been to compare the effects of support from

an animal to support from a human (e.g., a parent or a

friendly confederate; Beetz et al., 2012; Beetz et al., 2011;

Kertes et al., 2017

). In some cases, these studies have also

ncluded support from toy dogs as an additional control

(Beetz et al., 2012; Beetz et al., 2011). Findings from these

studies are mixed; some show positive effects of support

from a dog on physiological arousal but not perceived stress,

and some show positive effects on perceived stress but not

physiological arousal (e.g., Beetz et al., 2012;Kertesetal.,

2017). These ﬁndings provide initial hints that animals may

be effective sources of social support for children. However,

key questions remain. For example, do the effects of AAAs

exceed the effects of other common approaches to improving

children’s affect and anxiety? One additional study

attempted to isolate the effects of AAAs by ruling out

participation in any distracting activity as an alternative

explanation for the beneﬁts (Barker, Knisely, Schubert,

Green, & Ameringer, 2015). However, this study showed

no differences between the effects of the two conditions,

highlighting the fact that it is not yet clear whether interac-

tions with animals convey speciﬁc beneﬁts for children.

Observational ﬁndings from prior studies suggest one

especially important plausible alternative explanation for

the beneﬁts of AAAs. These studies have shown that chil-

dren who engage in more physical contact with the dogs

over the course of their interactions show lower levels of

stress following those interactions, suggesting that tactile

stimulation may be driving the beneﬁts that we typically

attribute to the animals (e.g., Beetz et al., 2012; Kertes

et al., 2017). Indeed, the beneﬁts of tactile stimulation for

mental health are well-established (see Ardiel & Rankin,

2010, for a review). Of course, it is likely that interactions

with animals produce beneﬁts in part because of tactile

stimulation. However, the key question in terms of isolating

the effects of AAAs is whether the beneﬁts are attributable

entirely to the effects of tactile stimulation.

Identifying the active ingredient in AAAs and ruling out

alternative explanations is a crucial step in ensuring that

these programs are as effective and efﬁcient as possible

(Chambless & Hollon, 2012). This is particularly important

in light of the fact that these programs have already been

widely implemented. In addition, interventions that involve

interactions with animals introduce concerns about the wel-

fare of the animals, the possibility of exposing individuals

with allergies to and/or fears of the animals, and the risk of

contamination/disease transmission. These concerns under-

score the importance of ensuring that the animals are con-

tributing meaningfully to the process of change. As a

starting point, the ﬁeld needs carefully controlled

CROSSMAN ET AL.

laboratory studies to isolate the effects of the types of

child–animal interactions that compose AAAs and to rule

out alternative explanations for their effects. Such investi-

gations will provide a foundation of evidence on which to

build our understanding of the role of AAAs in children’s

mental health care. Speciﬁcally, these investigations will

inform efforts to develop guidelines for practice, identify

populations of children that may be especially likely to

beneﬁt, and determine the precise role(s) of AAAs in

clinical child psychology.

The present study evaluated the effects of brief, unstruc-

tured interactions with unfamiliar dogs on children’s anxi-

ety, affect, and physiological arousal, following exposure to

a stressful task. We compared the effects of these interac-

tions to the effects of tactile stimulation, as well as to no

intervention. A secondary goal was to evaluate whether

ﬁndings from prior studies, which have focused on whether

interactions with dogs buffer against the negative effects of

stress, would extend to interactions with dogs that occur

after stress exposure, when symptoms are already elevated.

We predicted that brief, unstructured interactions with unfa-

miliar dogs would increase children’s positive affect and

reduce negative affect, anxiety, and physiological arousal,

relative to tactile stimulation or no intervention. In addition,

we were interested in how children who did not have pets

of their own would respond to the interactions, based on

suggestions that prior experiences with animals may be

necessary to gain the full beneﬁts of AAAs (e.g., Stewart

& Strickland, 2013). We therefore explored the role of

children’s ownership of, experience with, and feelings and

behavior toward pets in their responses to the interactions.

METHOD

Human Participants

Participants were 78 children (43 [55.1%] female, 35

[44.9%] male), ranging in age from 10 to 13

(M =12.01,SD = 1.13) and drawn from the local com-

munity surrounding a university in the northeastern

United States. The county includes a mix of urban, sub-

urban, and rural towns and has a total population of

approximately 900,000 (U.S. Census Bureau, 2016).

The county is predominantly White, Non-Hispanic, and

the median household income as of 2015 was approxi-

mately $60,000, with a poverty rate of 13.6%. Per parent

report, 44 participants ( 56.4%) were White, 6 (7.7%)

were Hispanic/Latino, 5 (6.4%) were Asian, 4 (5.1%)

were Black/African American, 1 (1.3%) was Native

American/Alaska Native, and 18 (23.1%) were multira-

cial/multiethnic. One additional participant began the

study but did not complete the full procedure and was

excluded from all analyses. For this participant, the pro-

cedure was terminated during the Trier Social Stress Test

for Children (discussed below) to prevent excessive

stress. Detailed demographic and background informa-

tion is available in Table 1.

Participants were recruited from September 2015 through

September 2016 using a range of methods including online

postings, e-mail, advertisements, ﬂiers, and in-person at local

sites. Exclusion criteria included fears of dogs and allergies to

animal dander, saliva, or urine. Each child participant received a

$5 gift certiﬁcate, a small toy, and a certiﬁcate of appreciation;

parents also received a $15 gift certiﬁcate. Th is study was

reviewed and approved by the Institutional Review Board and

the Institutional Animal Care and Use Committee of Yale

University. Parents and child participants provided informed

consent and assent, respectively.

We elected to use an unselected community sample in light

of the need for intervention methods that can reach a broad

range of children. In addition, the goal of this investigation was

to contribute to efforts to isolate the effects of interactions with

dogs on clinically relevant symptoms in children rather than to

tablish the efﬁcacy of AAAs for any full syndrome disorder.

We selected this age group (10–13) because older children

making the transition to adolescence are particularly susceptible

to barriers to treatment such as perceived stigma, discomfort

talking about mental health problems, and wanting to cope

independently (Gulliver, Grifﬁths, & Christensen, 2010). In

addition, AAAs are already in widespread use for children of

this age group (Friesen, 2010).

Sample Estimate

We based our target sample of 75 participants (actual

N = 78) on a sample estimate for a repeated measures

analysis of variance (ANOVA) with a small effect

(f = 0.2), three groups, a correlation of .7 between

repeated measurements, and 95% power (Faul, Erdfelder,

Lang, & Buchner, 2007). The results suggested that a

sample of 63 participants was needed. We added four

participants per condition to account for dropout, experi-

menter error, and related issues. We used this method in

light of ongoing debate about how to calculate a sample

estimate for our particular analytic approach (analysis of

covariance [ANCOVA], adjusting for baseline s cores on

the outcome measures), as well as the clear consensus that

this approach increases power relative to the ANOVA

approach (Van Breukelen, 2006; Vickers & Altman,

2001; Zhang et al., 2014). In other words, our sample

estimate reﬂects a conservative approach.

Measures

Positive and Negative Affect Schedule for Children,

Short Form

We used the Positive and Negative Affect Schedule for

Children, Short Form (PANAS-C-S) to assess affect at

EFFECTS OF DOGS ON ANXIETY, AFFECT, AND AROUSAL 3

baseline and after the intervention. The PANAS-C-S

includes a ﬁve-item Positive Affect scale and a ﬁve-item

Negative Affect scale (Ebesutani et al., 2012). Children

were asked to indicate the extent to which they felt each

item “right now” using a 5-point Likert scale. For example,

children were asked to rate the extent to which they felt

“happy” and “miserable” for the Positive and Negative

Affect scales, respectively. The items are summed to yield

separate Positive and Negative Affect scales.

The Positive and Negative Affect scales in the PANAS-

C-S have higher inter-item correlations than their full-

length counterparts. In the present investigation,

Cronbach’s alpha was .90 for the Positive Affect scale at

baseline. However, Cronbach’s alpha for the Negative

Affect scale at baseline was .57, indicating that negative

affect was not assessed reliably. We examined the intra-

item correlations to evaluate whether there were items with

low correlations to the rest of the scale, but we did not ﬁnd

any such items and the Negative Affect scale was excluded

from further analysis. The PANAS-C-S has discriminant

validity comparable to that of the full-length measure.

Baseline scores on the Positive Affect scale were compar-

able to those reported among other nonclinical samples and

among children experiencing chronic physical illness (e.g.,

Hexdall & Huebner, 2008; Reddy, Palmer, Jackson, Farris,

& Alfano, 2017).

State/Trait Anxiety Inventory for Children

We used the State portion of the State/Trait Anxiety

Inventory for Children (STAI-C) to assess anxiety at base-

line and after the intervention period. The STAI-C consists

of 20 items, and children are asked to report how they feel

“right now, at this very moment.” For example, children are

asked to indicate whether they feel “very relaxed,”

“relaxed,” or “not relaxed” (Spielberger, Gorsuch,

Lushene, Vagg, & Jacobs, 1973). Responses on all 20

items are summed. Cronbach’s alpha at baseline was .85.

The STAI-C has demonstrated construct, concurrent, and

discriminant validity for children and adolescents (e.g.,

Seligman, Ollendick, Langley, & Baldacci, 2004;

Spielberger et al., 1973). Baseline STAI-C scores were

similar to those reported for other unselected samples but

were lower than those reported for samples of children with

clinically signiﬁcant anxiety disorders (e.g., Flannery-

Schroeder & Kendall, 2000; Kendall, 1994; Muris, Rapee,

Meesters, Schouten, & Geers, 2003; Reddy et al., 2017;

Spielberger et al., 1973).

TABLE 1

Participant Characteristics

Characteristic Full Sample (N = 78) Dog (n = 26) Tactile Stimulation Control (n = 26) Waiting Control (n = 26)

Age

Range 10.00–13.96 10.13–13.92 10.27–13.94 10.00–13.96

M (SD) 12.01 (1.13) 12.16 (1.18) 12.18 (1.17) 11.69 (1.02)

Sex (%)

Female 43 (55.10) 14 (53.80) 15 (57.70) 14 (53.80)

Male 35 (44.90) 12 (46.20) 11 (42.30) 12 (46.20)

Race/Ethnicity (%)

White, Non-Hispanic 44 (56.40) 16 (61.50) 11 (42.30) 17 (65.40)

Black 4 (5.10) 1 (3.80) 2 (7.70) 1 (3.80)

Hispanic/Latino 6 (7.70) 2 (7.70) 4 (15.40) 0 (0.00)

Asian 5 (6.40) 0 (0.00) 0 (0.00) 5 (19.20)

Native American/Alaska Native 1 (1.30) 0 (0.00) 1 (3.80) 0 (0.00)

Multiracial/Multiethnic 18 (23.10) 7 (26.90) 8 (30.80) 3 (11.50)

Pet Ownership (%)

Current 62 (79.50) 22 (84.60) 18 (69.20) 22 (84.60)

Past 60 (76.90) 21 (80.80) 18 (69.20) 21 (80.80)

Dog Ownership (%)

Current 43 (55.10) 16 (61.50) 16 (61.50) 11 (42.30)

Past 34 (43.60) 12 (46.20) 10 (38.50) 12 (46.20)

CABS 28.02 (5.23) 27.94 (5.75) 27.88 (4.00) 28.21 (5.88)

EDI

Positive 31.46 (5.64) 33.16 (5.60) 31.09 (5.70) 30.04 (5.38)

Negative 10.42 (2.88) 10.40 (2.43) 9.78 (2.52) 11.04 (3.53)

PANAS-C-P M (SD)

Positive 18.36 (3.66) 18.88 (3.79) 17.88 (3.76) 18.31 (3.51)

Negative 8.23 (3.13) 7.42 (1.90) 7.40 (2.45) 9.85 (4.05)

Note: CABS = Companion Animal Bonding Scale; EDI = Experiences with Dogs Inventory; PANAS-C-P = Positive and Negative Affect Schedule for

Children-Parent Report, Short Form.

4 CROSSMAN ET AL.

Salivary Cortisol

Salivary cortisol is a reliable measure of activation of

the Hypothalamic-Pituitary-Adrenal axis, which regulates

the body’s stress response, and has been extensively vali-

dated as a measure of physiological stress (see Kirschbaum

& Hellhammer, 1989, for a review). Salivary cortisol was

assessed as a physiological indicator of stress at ﬁve time

points: Baseline; immediately after the stress induction; and

at 5, 10, and 15 min into the intervention period. Thus,

accounting for the time to collect the cortisol samples, the

last cortisol sample was collected approximately 41 min

after stressor onset and 23 min after its conclusion.

Although a longer time line of assessment (45 to 60 min

post-exposure) is commonly used to capture complete cor-

tisol recovery, the timing of the cortisol assessments was

selected based on consideration of a number of factors.

Previous research examining the use of an intervention to

promote cortisol recovery following stress exposure

showed group differences at approximately 20 min post-

exposure (Khalfa, Bella, Roy, Peretz, & Lupien, 2003).

Studies of cortisol recovery in children and adolescence

additionally show recovery at approximately 20 min post-

exposure (Hankin, Badanes, Abela, & Watamura, 2010).

Finally, in previous pilot work, our research group found

that absent any intervention, child participants showed

nearly complete cortisol recovery using this time line. In

light of these ﬁndings and because our priority was to

detect differences in the rate of cortisol recovery between

groups, we chose to prioritize more frequent assessments to

maximize our chances of capturing between-group differ-

ences in the rate of cortisol recovery. In addition, we

selected the length of the interaction (15 min, not including

breaks for cortisol assessments) based on our expectation of

what would be maximally effective for the children and in

the best interests of the dogs. Extending the cortisol assess-

ment period would thus have meant continuing the assess-

ments beyond the conclusion of the interactions, which

might have diluted the between-group effects.

Participants were instructed not to consume any food in the

hour before coming to the lab (Kirschbaum & Hellhammer ,

1989). Salivary cortisol was collected using Saliva Bio Oral

Swabs (Salimetrics, Carlsbad, CA), which participants

chewed for 1 min per sample. After collection, samples

were frozen (below −20 °C) until assay. Samples were

assayed in duplicate according to the manufacturer’s recom-

mended protocol (#1–3002, Salimetrics, Carlsbad, CA) using

a highly sensitive enzyme immunoassay designed for analyz-

ing saliva samples and a sample test volume of 25 μLof

saliva per determination. The lower limit of sensitivity of the

assay is 0.007 μg/dL, and the standard curve ranges between

0.012 μg/dL to 3.0 μg/dL. The average intra-assay coefﬁcient

of variability (CV) was 7.92%, indicating that cortisol con-

centration was assessed reliably.

To evaluate cortisol output over the course of the study,

we computed the area under the curve with respect to

ground (AUC

; Pruessner, Kirschbaum, Meinlschmid, &

Hellhammer, 2003). AUC

provides a summary of the

overall cortisol output, allowing for straightforward analy-

sis of repeated measures data. AUC

takes into account

change over time with respect to baseline scores, as well as

the overall magnitude of the cortisol response (Khoury

et al., 2015; Pruessner et al., 2003).

Positive and Negative Affect Schedule for

Children–Parent Report, Short Form

We hypoth esized that children’s affect would change over

the course of the study. However, change in affect may be

inﬂuenced by more long-term moods. Accordingly, we

included the Positive and Negative Affect Schedule for

Children–Paren t Report, Short Form (PANAS-C-P) as a possi-

bleadjustmentvariable.ThePANAS-C-Pisthesameasthe

child-report version (see earlier), but in this case parents were

asked to respond based on how their children have felt “during

the past few weeks.” Cronbach’salphawas.86forthePositive

Affect scale and .81 for the Negative Affect scale. The

PANAS-C-P has demonstrated construct validity (Ebesutani,

Okamura, Higa-McMillan, & Chorpita, 201 1). Scores on the

PANAS-C-P in the present sample were similar to those

reported previously among a large, school-based sample

(Ebesutani et al., 2011).

Companion Animal Bonding Scale

We included the Companion Animal Bonding Scale

(CABS) to explore the role of children’s feelings and behavior

toward companion animals in their responses to the interactions

in our study. The CABS is a self-report measure that assesses

children’s day-to-day interac tions with and feelin gs toward their

companion animals. Children indicate the frequency of each of

eight behaviors and feelings relating to their bonding with their

pet (e.g., “How often do you feel that you have a close relation-

ship with your companion animal?”) on a 5-point Likert scale.

The CABS has established convergent validity and internal

consistency (Cohen, 2002; Poresky, Hendrix, Mosier , &

Samuelson, 1987; Triebenba cher, 1999). Cronbach’salphafor

the CABS was .70. Participants in all three condition s who had a

pet at the time of the study completed the CABS. Participants

who did not have a pet at the time of participation did not

complete the CABS.

Experiences with Dogs

We were interested in exploring the role of each child’s

general prior experiences with dogs, in addition to their level

of bonding with a particular companion animal. In addition, we

wished to quantify children’s general exposure to dogs, regard-

less of whether they themselves had a dog at the time of the

EFFECTS OF DOGS ON ANXIETY, AFFECT, AND AROUSAL 5

study. We therefore asked parents of child participants to report

on current and past pet ownership and to specify the species of

those pets. This allowed us to analyze the effects of pet owner-

ship in general, as well as dog ownership in particular. In

addition, we included an adapted version of the Experiences

with Dogs Inventory (EDI; Crossman, Kazdin, & Knudson,

2015). The EDI asked parents to rate how frequently their

childrenhadengagedineachof20experiencesinvolving

dogs, including 13 positive/neutral experiences (e.g., “My

child has touched/petted a dog”) and seven negative experi-

ences (e.g., “My child has been harmed by a dog”), which yield

separate Positive and Negative sum scores. Responses are made

using a Likert scale ranging from 1 (never)to5(very fre-

quently). Cronbach’s alpha for the Positive dimension of the

EDI was .69. W e therefore removed one item (“My child has

worked with dogs to earn money [e.g., dog walker, dog sitter],”

which had a small, negative correlation; r = −.01) with the rest

of the scale. Cronbach’s alpha for the resulting 12-item Positive

dimension was .70, and we therefore used this version.

Cronbach’s alpha for the Negative dimension was .72. The

EDI has established convergent validity (Crossman et al.,

2015).

Stress Task

We used the Trier Social Stress Test for Children (TSST-C)

to induce moderate psychosocial stress (Buske-Kirschbaum

et al., 1997). The TSST-C involves an anticipation phase, a

public speaking phase, and a mental arithmetic phase. The

TSST-C reliably elicits moderate psychosocial stress in

children ages 7 to 14 (Gunnar, Talge, & Herrera, 2009).

Selection and Protection of the Dogs

Eight dogs (four male, four female) participated in data collec-

tion. Two additional dogs participated but were not involved in

data collection for the experimental condition (i.e., only parti-

cipants in the control conditions interacted with these dogs, after

completing the study). Although it would have been preferable

to limit the number of dogs (and consequ ently the amount of

variability in the nature of the interactions), this number of dogs

was needed to facilitate participant scheduling. In addition, the

variability introduced by the eight dogs included in the present

study was less than is introduced in studies evaluating interac-

tions between children and their own dogs (e.g., Kerns et al.,

2017;Kertesetal.,2017). All dogs were certiﬁed or registered

as therapy dogs with leading therapy dog or ganizations (e.g.,

The Good Dog Foundation, Pet Partners) or had successfully

completed an equivalent behavioral evaluation. Veterinary

records were provided for all dogs to indicate that they were

up-to-date on vaccinations and free of parasites.

Before

enrollment, each dog completed an orientation visit at the lab,

which included time for acclimation to the facility, orientation

of the handler and dog to the study procedures, and evaluation

of the dog’s behavioral response to the procedures.

A key component of the hypothesis was that interaction

with a dog (rather than a dog and handler) would alleviate

stress and improve affect. In addition, a primary limitation

of past human–animal interaction research has been the

failure to disentangle the effects of interaction with the

dogs from interaction with the dogs’ handlers and other

program participants (Marino, 2012). Accordingly, it was

important that participants in the present study interact

independently with the dogs without extensive intervention

from human handlers. We achieved this goal by having a

member of the research team sit in the corner of the room

to supervise the interactions, intervening only as necessary

to protect the well-being of the dog and/or child. The dog’s

handler waited in a separate room.

To address concerns raised by separating dogs from their

handlers, our procedure included a number of protections for

both the dogs and the child participants. First, a central compo-

nent of the orientation visit was to evaluate the dog’s response to

being separated from the handler. After familiarization with the

facility and the members of the research team, each dog was left

with a member of the research team for a test period of

5–10 min. During this test period, the handler was free to

observe the dog through a two-way mirror or closed-circuit

television system. If dogs showed signs of stress, the test was

ended. Dogs that showed signs of stress or otherwise demon-

strated inappropria te behavior were not accepted into the study.

During the study, handlers were again able to watch all

interactions through a two-way mirror and had the option to

stop any interaction to protect their dog’s safety or well-

being. However, an interaction was terminated early (after

10 min) on only one occasion, due to experimenter con-

cerns about signs of stress (e.g., barking).

Finally, partici-

pants were given instructions for safe interactions with

dogs and were provided brief, scripted prompts to remind

them of these instructions as needed. All members of the

research team were trained to detect signs of stress and

Therapy dog registration/certiﬁcation was used as an indicator of

dogs’ behavioral suitability for interactions with unfamiliar children.

Because the interactions in this study were conducted without the handler

in the room, these interactions are not considered therapy dog visits/

interactions.

One dog that demonstrated appropriate behavior and did not show

signs of stress at the orientation visit later showed signs of stress during its

ﬁrst days of participation in the study. This dog was accordingly with-

drawn from participation following consideration of changes in her beha-

vior in and around the lab by the members of the research team and the

dog’s handler.

In that instance, the dog was reunited with its handler, and the child

waited for the remaining 5 min of the interaction period. Because the

majority of the interaction was completed, that participant was retained for

all analyses. However, we also conﬁrmed that the pattern of results

(including main and supplementary analyses) does not change when that

participant is removed from the analyses.

6 CROSSMAN ET AL.

aggression in dogs and received hands-on training for mon-

itoring child–dog interactions. To prevent fatigue, dogs

never participated in the study 2 days in a row and never

had more than two sessions per day, with breaks provided

between sessions.

Study Conditions

Assignment to the three study conditions was made using a

random number generator (https://www.randomizer.org/).

The experimental condition involved a 15-min interaction

with a dog, which was supervised by an experimenter.

Participants were permitted to interact freely with the

dogs so long as they maintained appropriate behavior

(e.g., no touching the dogs’ eyes/ears/mouths, no yelling)

and the dogs did not show signs of stress or aggression.

To establish whether the interactions conveyed beneﬁts

beyond the effects of tactile stimulation, we included a

tactile-stimulation control condition in which children

were given a soft blanket. The goal was to evaluate whether

something about the act of interacting with the dogs, rather

than just the tactile stimulation they provide or the presence

of any novel and comforting object, would reduce chil-

dren’s stress. We also included a waiting control condition

to evaluate whether interaction with a dog is more effective

than children’s independent coping skills and to rule out the

effects of factors such as completing the measures repeat-

edly, familiarization with the setting, and the simple pas-

sage of time as alternative explanations. As in the

experimental condition, the experimenter remained in the

corner of the room during both control conditions.

Procedure

Participant sessions were scheduled to begin between 1 and

6 p.m. After providing informed consent, parents of child

participants completed a background questionnaire about

their children, which included the PANAS-C-P. After pro-

viding assent, child participants provided baseline saliva

samples and then completed baseline self-report measures

(PANAS-C-S, STAI-C). The TSST-C was then completed,

followed immediately by collection of the second cortisol

sample. Participants then engaged in their respective con-

ditions for 15 min, with the third, fourth, and ﬁfth cortisol

samples collected at 5-min intervals. Participants then com-

pleted the posttest PANAS-C-S and STAI-C, and partici-

pants in both control conditions were given the chance to

interact with a dog. Participants were then debriefed and

provided their toys, gift certiﬁcates, and certiﬁcates of

participation.

Data Analytic Plan

Treatment of Missing Data

We computed sum scores for all self-report measures.

For cases in which individual items were missing, missing

data were prorated using the mean of the completed items.

Measures missing more than one fourth of the items were

excluded. The average number of missing items per mea-

sure per participant was 0.57.

Preliminary Analysis

We chec ked f or dif ferences on background and demo-

graphic variables by condition using chi-squares and one-way

ANOVAs. We used Pearson product–moment correlations to

evaluate the relations among baseline variables and to check for

redundancy of measures. We used a threshold of .71 (indicating

a shared variance of 50%) to evaluate redundancy of measures.

In case of correlations exceeding .71, we evaluated the correla-

tions between change scores on those measures, with the inten-

tion that we would combine measures where change scores

were also highly correlated. For participants in the experimental

condition, we used one-way ANOVAs to check for differences

in change scores for the STAI-C and the Positive Affect scale,

as well as for differences in AUC

, based on the dog with

which participants interacted.

Effects of the Interactions

We used a one-way ANCOVA to evaluate the effect of

condition on each self-report measure (the Positive Affect

scale and the STAI-C), adjusting for scores on that same

measure at baseline. We used this approach (rather than a

repeated-measures ANOVA) because baseline measures

were collected before the stress induction and the interven-

tion (rather than immediately before the intervention). In

cases of signiﬁcant effects, planned contrast were con-

ducted to evaluate the difference between the experimental

condition and each of the two control conditions. For

ANCOVAs, adjusted means are presented unless otherwise

noted. For planned contrasts, Cohen’s d was calculated

using adjusted means and the square root of the mean

square error. We used a one-way ANOVA to evaluate

differences in AUC

based on study condition.

Supplementary Analysis

We conducted exploratory analyses to examine the

importance of children’s current and past ownership of

pets and of dogs in particular, experiences with dogs, and

their self-reported feelings and behaviors toward their dogs.

For this analysis, we focused only on participants in the

After Participant 44, random assignment was split up by gender (i.e.,

there were separate random assignment lists for male and female partici-

pants) to ensure similar proportions of male and female participants in

each condition.

EFFECTS OF DOGS ON ANXIETY, AFFECT, AND AROUSAL

experimental condition. We used Pearson’s partial correla-

tions to examine the relations between the continuous ani-

mal experience variables (EDI, CABS) and each self-report

outcome (PANAS-C-S, STAI-C), adjusting for baseline

scores on the same self-report measure. We additionally

used Pearson product–moment correlations to evaluate the

relations between AUC

and EDI and CABS scores. For

the pet ownership variables (current and past pet owner-

ship, current and past dog ownership), we used one-way

ANCOVAs to evaluate the effects on posttest scores on

each self-report measure, adjusting for baseline scores on

the same self-report measure and one-way ANOVAs to

evaluate the effects of these pet ownership variables on

AUC

RESULTS

Preliminary Analysis

Participants in the three conditions did not differ in terms of

age, race/ethnicity, sex, current or past dog ownership, EDI

scores, or CABS scores (ns).

In light of the low number of

participants who did not have pets at the time of participa-

tion (n = 11) or in the past (n = 13), the sample size was

insufﬁcient to evaluate whether the conditions differed in

terms of current and/or past pet ownership of participants.

Frequencies of current and past pet ownership are presented

in Table 1. Participants in the different conditions did not

differ at baseline on the Positive Affect scale, the STAI-C,

or cortisol (ns). Parent reports of participants’ positive

affect “during the past few weeks” using the PANAS-C-P

also did not differ by condition (ns). However, parent

reports on the Negative Affect scale of the PANAS-C-P

did vary across conditions, F(2, 74) = 5.87, p = .004, η

= 0.14. Parent-reported Negative Affect scale scores were

higher for participants in the waiting control condition

(M = 9.85, SD = 4.05) than for participants in the experi-

mental condition ( M = 7.42, SD = 1.90), a mean difference

of 2.42 points, 95% CI [0.79, 4.05], p = .004, d = 0.77.

Scores for participants in the waiting control condition

were also higher than for those in the tactile-stimulation

control condition (M = 7.40, SD = 2.45), a mean difference

of 2.45 points, 95% CI [0.80, 4.09], p = .004, d = 0.73.

However, parent-reported Negative Affect scores were not

signiﬁcantly correlated with change in Positive Affect scale

scores, r(73) = .06, p = .623, change in STAI-C scores, r

(73) = −.05, p = .690, or AUC

, r(73) = .07, p = .582. As a

result, and in consideration of the fact that self-reports on

the Negative Affect scale were excluded as an outcome

measure, parent-reported Negative Affect was not included

in subsequent analyses.

In terms of the relations among baseline measures,

Positive Affect scores were negatively correlated with

STAI-C scores, r(75) = −.78, p < .001. The magnitude of

this correlation raised concerns about the possibility of

redundancy of measures. However, change in Positive

Affect scores and change in STAI-C scores from baseline

to posttest were only moderately negatively correlated, r

(74) = −.46, p < .001. We consequently retained the

Positive Affect scale and STAI-C as separate measures.

Baseline salivary cortisol was not signiﬁcantly correlated

with baseline Positive Affect scale scores, r(72) = −.01,

p = .926, or STAI-C scores, r(72) = .03, p = .800. Within

the experimental condition, there were no signiﬁcant differ-

ences in change in Positive Affect scores, F(7, 17) = 1.12,

p = .396, η

= 0.32; change in STAI-C scores, F(7,

17) = 0.96, p = .488, η

= 0.28; or AUC

, F(7, 17) = 0.79,

p = .603, η

= 0.25, based on the dog with which participants

interacted.

Effects of the Interactions

Baseline scores and unadjusted posttest scores on each of

the outcome measures are presented in Table 2. The

ANCOVA evaluating the effect of the interactions on

Positive Affect scores revealed a signiﬁcant effect of con-

dition, F(2, 72) = 4.37, p = .016, η

= 0.11. Planned

contrasts revealed that posttest Positive Affect scores for

participants in the experimental condition (M = 18.82,

SE = 0.60) were signiﬁcantly higher than those of partici-

pants in the waiting control condition (M = 16.90,

SE = 0.59), a mean difference of 1.92 points, 95% CI

[0.24, 3.60], p = .025, d = 0.65.

Positive Affect scores

for participants in the experimental condition were also

signiﬁcantly higher than those of participants in the tac-

tile-stimulation control condition (M = 16.48, SE = 0.59), a

mean difference of 2.35 points, 95% CI [0.66, 4.03],

p = .007, d = 0.79. Following exposure to a stressful task,

child participants who interacted with a dog showed higher

levels of positive affect than participants who received

tactile stimulation without any interaction or waited.

Detailed statistics for planned contrasts are presented in

Table 3.

Because the majority of the sample was White and cell sizes for all

ethnicities other than White were small (< 7), we collapsed across all

ethnicities other than White (Asian, Black/African American, Hispanic/

Latino, Native American/Alaska Native, and “other”) for the purposes of

this analysis. The total number of White participants was 44, and the total

number of non-White participants was 34.

Initially, we conducted the planned contrasts for the self-report out-

comes using a Bonferroni correction for multiple comparisons. However, a

reviewer raised the possibility that this approach might have been too

conservative, and we consequently removed the correction. When the

Bonferroni correction is applied, the comparison between participants in

the experimental condition and those in the waiting control condition is

not signiﬁcant for the Positive Affect scale (Bonferroni corrected thresh-

old = .025). The results for the STAI-C are unchanged.

8 CROSSMAN ET AL.

There was also a signiﬁcant effect of condition for

STAI-C scores, F(2, 72) = 4.69, p = .012, η

= 0.12.

Participants in the experimental condition (M = 26.47,

SE = 0.85) had signiﬁcantly lower scores than participants

in the waiting control condition (M = 30.09, SE = 0.83), a

mean difference of 3.63 points, 95% CI [1.25, 6.01],

p = .003, d = 0.86. However, the difference between

STAI-C scores in the experimental condition and those in

the tactile-stimulation control condition (M = 28.72,

SE = 0.85) was not signiﬁcant, p = .065, 95% CI [−0.14,

4.65], d = 0.53. These ﬁndings convey that interaction with

the dog did reduce anxiety relative to waiting without

intervention but raise questions about whether this anxio-

lytic effect is speciﬁc to interaction with a dog, or whether

tactile stimulation from any soothing object might convey a

similar beneﬁt.

The ANOVA for AUC

failed to reveal a signiﬁcant

effect of condition, F(2, 73) = 0.98, p =.381,η

= 0.03.

Participants who interacted with the dogs did not differ

signiﬁcantly from participants in the other conditions in

terms of overall physiological arousal. A visual depiction

of change in salivary cortisol concentration for each

condition is presented in Figure 1.

Supplementary Analysis

Among participants in the experimental condition, we

explored whether participants’ responses to the interactions

varied based on their EDI and CABS scores and based on

current and past pet ownership.

EDI

Among participants in the experimental condition, scores

on the Positive dimension of the EDI were signiﬁcantly

correlated with posttest Positive Affect scale scores, adjust-

ing for baseline Positive Affect scale scores, r(21) = .42,

p = .047. However, the Positive dimension of the EDI was

not signiﬁcantly correlated with either posttest STAI-C

scores (adjusting for baseline STAI-C scores), r(21) = −.23,

p = .288, or AUC

, r(22) = −.21, p = .329. Scores on the

negative dimension of the EDI were not associated with

posttest Positive Affect scores, r(21) = .16, p = .465, or

STAI-C scores (adjusting for baseline scores on the same

measures), r(21) = .05, p = .839. Surprisingly, however,

higher scores on the Negative dimension of the EDI were

associated with lower AUC

, r (22) = −.47, p =.021.

CABS

Adjusting for baseline scores, the CABS was not asso-

ciated with posttest Positive Affect scale scores, r(14) = .08,

p = .776, or STAI-C scores, r(14) = −.04, p = .897. The

CABS was also not associated with AUC

, r(14) = −.37,

p =.161.

Pet ownership

Participants who had pets at the time of participation

showed signiﬁcantly higher Positive Affect scale scores at

posttest (M = 20.45, SE = 0.79) than participants who did not

have pets at the time of participation (M =15.89,SE = 1.88),

when adjusting for baseline Positive Affect scale scores,

F(1, 22) = 4.85, p =.038,η

= 0.18. However, STAI-C

scores, F(1, 22) = 0.25, p =.622,η

= 0.01 and AUC

F(1, 23) = 0.93, p =.346,η

= 0.04, did not vary based on

TABLE 2

Unadjusted Means and Standard Deviations for Primary Outcome

Measures at Baseline and Posttest

Measures:

M (SD)

Full

Sample

(N = 78)

Dog

(n = 26)

Tactile Stimulation

Control (n = 26)

Waiting

Control

(n = 26)

Baseline

Scores

PANAS-

C-S:

Positive

17.58

(5.22)

18.69

(4.76)

17.40 (4.86) 16.65 (5.96)

STAI-C 28.08

(4.66)

27.15

(3.84)

28.32 (4.85) 28.77 (5.22)

Cortisol

(μg/dL)

0.10 (0.09) 0.11

(0.10)

0.07 (0.04) 0.12 (0.11)

Posttest

Scores

PANAS-

C-S:

Positive

17.39

(5.51)

19.72

(5.18)

16.36 (4.65) 16.15 (6.03)

STAI-C 28.64

(5.86)

25.80

(4.94)

29.42 (5.43) 30.58 (6.24)

Cortisol

(μg/dL)

0.19 (0.18) 0.19

(0.17)

0.16 (0.17) 0.22 (0.20)

Note. PANAS-C-S = Positive and Negative Affect Schedule for

Children, Short Form; STAI-C = State/Trait Anxiety Inventory for

Children (state portion).

TABLE 3

Planned Contrasts for Self-Report Outcomes

95% CI

Outcome Contrast

Difference LL UL p

Cohen’s

PANAS-C-S:

Positive

Dog vs. tactile

stimulation

2.35 0.66 4.03 .007 0.79

Dog vs. waiting 1.92 0.24 3.60 .025 0.65

STAI-C Dog vs. tactile

stimulation

2.25 −0.14 4.65 .065 0.53

Dog vs. waiting 3.63 1.25 6.01 .003 0.86

Note: Mean differences reﬂ ect differences in adjusted posttest means

(posttest scores, adjusted for baseline on the same measure). Cohen’s d

was calculated using adjusted means and the square root of the mean

square error. CI = conﬁdence interval; LL = lower limit; UL = upper limit;

PANAS-C-S = Positive and Negative Affect Schedule for Children, Short

Form; STAI-C = State/Trait Anxiety Inventory for Children (state portion).

EFFECTS OF DOGS ON ANXIETY, AFFECT, AND AROUSAL

current pet ownership. Neither Positive Affect scale scores,

F(1, 22) = 3.02, p =.096,η

= 0.12; nor STAI-C scores,

F(1, 22) = 0.14, p =.710,η

=0.01;norAUC

F(1, 23) = 3.26, p =.084,η

= 0.12, varied based on past

pet ownership. Positive Affect scale scores, F(1, 22) = 0.04,

p = .838, η

= 0.002; STAI-C scores, F(1, 22) = 0.04,

p =.844,η

= 0.002; and AUC

, F(1, 23) = 0.36, p = .553,

= 0.02, also did not vary based on current dog ownership.

Finally, Positive Affect scale scores, F(1, 22) = 0.08,

p = .787, η

= 0.003; STAI-C scores, F(1, 22) = 0.99,

p =.331,η

= 0.04; and AUC

, F(1, 23) = 0.06, p = .805,

= 0.003, did not vary based on past dog ownership.

DISCUSSION

We found that child participants who engaged in a brief,

unstructured interaction with an unfamiliar dog after expo-

sure to a moderate stressor showed higher positive affect,

relative to participants who received a soothing object or

waited for the same amount of time. Participants who inter-

acted with a dog also showed lower anxiety relative to those

who waited for the same amount of time but not relative to

those who received a soothing object. These effects on

positive mood and anxiety were medium in magnitude.

However, we did not detect any effect of the interactions

with the dogs on physiological arousal, as measured by

salivary cortisol. Overall, our ﬁndings support the effects

of brief, unstructured interactions with dogs on subjective

but not physiological indicators of affect and anxiety.

Interaction with a dog produced increases in positive

affect that exceeded the effects of tactile stimulation from a

soothing object or waiting for the same amount of time. This

ﬁnding suggests that the effects of interaction with a dog on

positive affect are not simply attributable to the effects of

tactile stimulation. Put another way, interaction with a dog

produced beneﬁts for positive affect that exceeded the effects

of another commonly used coping strategy. However, in the

case of anxiety, the effects of the experimental condition

exceeded those of waiting but not of tactile stimulation.

This suggests that although interaction with a dog reduces

anxiety relative to children’s independent coping abilities,

children may experience similar beneﬁts from a comforting

object that provides tactile stimulation. To summarize, we

found that interaction with a dog conveys some beneﬁts

beyond the effects of tactile stimulation from a soothing

object. However, our ﬁndings also highlight the importance

of continuing to question what it is about the interactions

that drives the beneﬁts (Marino, 2012).

We did not detect any effect of the interactions with the

dogs on salivary cortisol. There are two main possible inter-

pretations of this ﬁnding. First, it is possible that the inter-

actions did affect cortisol but that we failed to detect this

effect. We observed a high degree of variability in cortisol

output within each condition, which may have obscured any

effects of the interactions on cortisol. In addition, in this

study the ﬁnal cortisol sample was collected approximately

41 min after the onset of the stressor (and 23 min after the

conclusion of the stressor). At that ﬁnal time point, mean

cortisol levels had not returned to baseline, and it is possible

that we would have detected differences in the rate at which

cortisol returned to baseline with a longer period of interac-

tion and assessment. However, we view this possibility as

unlikely, as visual inspection of the cortisol data suggests

0.05

0.1

0.15

0.2

0.25

0.3

T1 T2 T3 T4 T5

Salivary Cortisol ( µg/dL )

Time Point

Dog

Tactile Stimulation

Waiting

FIGURE 1 Salivary cortisol concentration, broken down by study condition. Note: Error bars represent standard error.

As with the main analysis, we initially conducted the supplementary

analyses using a Bonferroni correction for multiple comparisons. However,

at the suggestion of a reviewer, we have reported the results of the supple-

mentary analyses without the correction for multiple comparisons in order

to reveal all possibly meaningful ﬁndings to inform future hypothesis

testing. When the Bonferroni correction is applied, none of the supplemen-

tary analyses are signiﬁcant (Bonferroni corrected threshold = .007).

10 CROSSMAN ET AL.

that the tactile-stimulation control condition (rather than the

experimental condition) was on track to recover most

quickly of the three conditions.

The second possible interpretation is that the interactions

in our investigation did not reduce overall cortisol output.

Findings from prior studies indicate that interaction with a

dog can buffer against increases in cortisol when the inter-

action occurs during stress exposure (e.g., Beetz et al.,

2012; Vagnoli et al., 2015). However, those stress-buffering

effects may not extend to postexposure cortisol reduction.

In other words, interactions with animals may affect corti-

sol reactivity but not cortisol recovery, as these are distinct

processes (Linden, Earle, Gerin, & Christenfeld, 1997;

Ramsay & Lewis, 2003). In addition, the fact that interac-

tions with dogs appeared to affect subjective but not phy-

siological indicators of stress in the context of our

investigation is not necessarily surprising, given that

many factors (e.g., cognitive appraisal, time course) may

contribute to the generally low covariance between salivary

cortisol and subjective reports of stress (Hellhammer, Wüst,

& Kudielka, 2009). It is plausible that the interactions with

the dogs caused participants to reappraise their experiences

of the stress task without altering their already elevated

levels of cortisol or their natural patterns of cortisol recov-

ery. Prior studies have also used different control conditions

(e.g., interaction with a person) than we used in this study;

it may be that the effects of interactions with dogs on

cortisol documented in prior studies are due primarily to

the tactile stimulation the dogs provide rather than to the

interactions with the dogs per se. Finally, consistent with

our ﬁndings, a number of other studies have also failed to

detect effects of interactions with dogs on cortisol (e.g.,

Handlin et al., 2011; Kaminski et al., 2002).

Among participants who interacted with the dogs in our

study, tho se who had more extensive histo ries of previous

interactions with pets, and with dogs in particular, seemed to

beneﬁtmost.Speciﬁcally, compared to children who did not

have pets, children who had pets at the time of participation

showed higher levels of posttest positive affect. Similarly,

children who were reported to have had more positive

experiences with dogs in the past had higher levels of

posttest positive affect. It may be that children who have

pets and/or have histories of positive experiences with dogs

are more prepared to beneﬁt from the interactions, or else

that these factors are simply indicators that a child has a

particular afﬁnity for pets. In addition, and somewhat sur -

prising, children who were reported to have had more nega-

tive experiences with dogs in the past showed lower overall

arousal in response to the interactions with the dogs. It is

possible that children who have had negative experiences

with dogs in the past show less excitement in response to

the interactions, or else that these children show less antici-

patory arousal precisely because they have already been

exposed to some adverse events related to dogs and have

developed a sense of mastery through those experiences.

However, none of the other measures of children’s feelings

toward and experiences with dogs were associated with

improvement on any of the three outcome measures. Thus,

it is possible that these ﬁndings represent false positives,

highlighting the need for further research to directly investi-

gate th e ch aracteristics o f children who are most likely to

beneﬁt from AAAs.

The present ﬁndings extend the literature on the effects of

interactions with dogs on children’s affect and anxiety in

three key ways. First, participants in our study interacted

with the dogs independently and without distractions, sug-

gesting that something about the interactions with the dogs

themselves produced the beneﬁts.Thisisnotablebecausea

failure to disentangle the effects of interactions with animals

from the effects of interactions with other people has been a

key limitation of previous studies and an important threat to

the construct validity of AAAs. Second, these ﬁndings

demonstrate that, at least for positive affect, the beneﬁts of

AAAs are not attributable entirely to the effects of tactile

stimulation. This is important in light of prior observational

ﬁndings that show associations between the amount of touch

that occurs during an AAA and the degree of improvement

shown by the child. In addition, this ﬁnding makes a small

but important contribution to efforts to establish the extent to

hich the effects of AAAs can really be attributed to the

interactions with the animals speciﬁcally. Finally, our ﬁnd-

ings demonstrate that AAAs can alleviate the negative

effects of stress on affect and anxiety after stress exposure,

in addition to preventing increases during exposure.

We demonstrated these effects in an unselected commu-

nity sample that was relatively diverse in terms of race,

gender, and prior experience with and ownership of pets

and that showed levels of baseline anxiety and affect that

were similar to those other community samples. As detailed

earlier , we elected to use a community sample for two

reasons. First, the primary goal of this investigation was to

provide a carefully controlled demonstration of the effects of

interactions with dogs for children. The goal was not to

establish the effects of these in teractions for any particular

disorder or population but rather to contribute to efforts to

begin to understand the basic processes involved. Second,

our interest in AAAs as a strategy for alleviating children’s

suffering comes in part from the possibility that AAAs may

be used to reach a broad range of children. Thus, demon-

strating the effects of AAAs among a community sample of

children is a strength, because it provides initial evidence for

the effects of AAAs when they are applied broadly.

Our study bears three important limitations. First, we eval-

uated only immediate changes in affect, anxiety, and arousal,

and we did not address whether any of the observed effects are

sustained in the long term. We focused on short-term changes

in light of the emerging state of this literature, as well as the

importance of even short-term symptom relief. As our

EFFECTS OF DOGS ON ANXIETY, AFFECT, AND AROUSAL 11

understanding of the strength and nature of these effects is

reﬁned, it will be important to establish how long they are

sustained and what impact they have on mental health and

well-being in the long term. Second, due to the logistical

factors described earlier, eight dogs participated in the present

study, and casual observations by experimenters suggested

that there were differences in the nature of the interactions

with the different dogs (e.g., some engaged in more active

play, whereas others were more likely to sit with the partici-

pant and be petted throughout the interaction). As a result, it is

possible that there were undetected differences in the strength

of the effects of the intervention, depending on the dog with

which participants interacted. Third, we evaluated the inﬂu-

ence of interactions with dogs among participants who volun-

teered to take part in a study involving dogs, introducing the

possibility of a selection effect. As a result, the beneﬁts of

interactions with dogs observed in this study, and especially

the ﬁndings relating to participants feelings toward and

experiences with pets, may not necessarily hold for partici-

pants who have highly negative views toward dogs or no

experience at all with dogs. However, children who partici-

pate in AAAs in applied settings are also likely to have

positive attitudes toward and experiences with dogs.

The present ﬁndings suggest two important avenues

for future r esearch. First, it will be important to continue

to establish how (i.e., through what processes) interac-

tions with dogs improve affect and anxiety symptoms.

Commonly proposed mechanisms of action include tac-

tile stimulation, social support, and emotional contagion.

Studies are needed that manipulate these various pro-

cesses or use established paradigms from the study of

interpersonal relationships to investigate them. This

research will help to clarify the speciﬁcbeneﬁts of inter-

actions with animals and inform stra tegies for maximiz-

ing those beneﬁts.

The second area for future research stems from observa-

tions by experimenters in our studies that the different dogs

appeared to behave differently. Speciﬁcally, it will be

important for future research to evaluate whether certain

characteristics of the dogs are more effective for improving

affect and anxiety. For example, are more active dogs who

initiate more contact and interaction more engaging and

enjoyable to interact with, or do more relaxed dogs promote

more relaxation in their human interaction partners? Other

characteristics of potential importance include the size and

age of the dogs, the amount of eye contact they naturally

initiate, their responsiveness to their names or other over-

tures, and the softness of their coats. Of course, it is not

necessarily the case that certain characteristics are better for

improving affect and anxiety, whereas others are worse;

instead, it may be that certain characteristics are more

effective for certain individuals, under different circum-

stances, or with different goals in mind. We can think of

this area of research as the canine version of long-standing

efforts in psychotherapy research to tailor treatments to

individual clients and situations.

SUMMARY AND CONCLUSION

We evaluated the inﬂuence of brief, unstructured interactions

with dogs on children’s affect, anxiety, and arousal, following

exposure to a stressful task. These interactions increased

children’s positive affect, relative to waiting without any

intervention or receiving tactile stimulation from a soothing

object, and reduced children’s state anxiety relative tactile

stimulation. We did not detect any effect on physiological

arousal. These ﬁndings convey that brief, unstructured inter-

actions with dogs have a moderate impact on children’s

subjective experiences of anxiety and affect, following expo-

sure to a stressor. However, additional research is needed to

establish whether the beneﬁts of interactions with animals

exceed the effects of other common strategies for improving

mental health and alleviating stress. The present ﬁndings

support the notion that AAAs may be an efﬁcient strategy

for improving children’s mental health but highlight the fact

that there is not yet sufﬁcient evidence to support their

already widespread prevalence in practice.

ACKNOWLEDGEMENT

The primary

ﬁndi

ngs reported in this article are based on data

also reported in the doctoral dissertation of the ﬁrst author . We

additionally thank the Good Dog Foundation for their assistance

with this study , as well as the dogs and handlers who generously

volunteered their time to participate , and without whom this

study would not have been possible. Dr. Rajita Sinha addition-

ally provided essential assistance with the analysis of the corti-

sol data.

FUNDING

This study, in whole or in part, was funded by the Morris

Animal Foundation exclusively from a partnership with the

Human-Animal Bond Research Institute (HABRI; D15HA-

025). Additional support for this study was provided by the

Laura J. Niles Foundation and the Humane Society of the

United States.

REFERENCES

Ardiel, E. L., & Rankin, C. H. (2010). The importance of touch in

development. Paediatrics & Child Health, 15, 153–156. doi:10.1093/

pch/15.3.153

Balzás, J., Miklósi, M., Keresztény, A., Hoven, C., Carli, V., Wasserman,

C., … Wasserman, D. (2013). Adolescent subthreshold-depression and

12 CROSSMAN ET AL.

anxiety: Psychopathology, functional impairment and increased suicide

risk. Journal of Child Psychology and Psychiatry, 54, 670–677.

doi:10.1111/jcpp.12016

Barker, S. B., Knisely, J. S., Schubert, C. M., Green, J. D., & Ameringer,

S. (2015). The effect of an animal-assisted intervention on anxiety and

pain in hospitalized children. Anthrozoös, 28, 101–112. doi:10.2752/

089279315X14129350722091

Beetz, A., Julius, H., Turner , D., & Kotrschal, K. (2012). Effects of social

support by a dog on stress modulation in male children with insecure

attachment. Frontiers in Psychology , 3,1–9. doi:10.3389/fpsyg.2012.00001

Beetz, A., Kotrschal, K., Turner, D. C., Hediger, K., Uvnäs-Moberg, K., &

Julius, H. (2011). The effect of a real dog, toy dog and friendly person on

insecurely attached children during a stressful task: An exploratory study.

Anthrozoös, 24,349–368. doi:10.2752/175303711X13159027359746

Braun, C., Stangler, T., Narveson, J., & Pettingell, S. (2009). Animal-

assisted therapy as a pain relief intervention for children.

Complementary Therapies in Clinical Practice, 15, 105–109.

doi:10.1016/j.ctcp.2009.02.008

Buske-Kirschbaum, A., Jobst, S., Wustmans, A., Kirschbaum, C., Rauh,

W., & Hellhammer, D. (1997). Attenuated free cortisol response to

psychosocial stress in children with atopic dermatitis. Psychosomatic

Medicine, 59, 419–426.

Centers for Disease Control. (2013). Mental health surveillance among

children—United States, 2005-2011. Morbidity and Mortality Weekly

Report: Supplement, 62,1–35.

Chambless, D. L., & Hollon, S. D. (2012). Treatment validity for inter-

vention studies. In H. Cooper (Ed.), APA Handbook of Research

Methods in Psychology: Vol 2. Quantitative, Neuropsychological, and

Biological (pp. 1–24). Washington, D.C.: American Psychological

Association.

Cohen, S. P. (2002). Can pets function as family members? We ster n Journal

of Nursing Research, 24,621–63

8. doi:10.1177/019394502320555386

Crossman, M. K. (2017). Effects of interactions with animals on human

psychological distress. Journal of Clinical Psychology, 73, 761–784.

doi:10.1002/jclp.22410

Crossman, M. K., & Kazdin, A. E. (2015). Animal visitation programs in

colleges and universities: An efﬁcient model for reducing student stress. In

Handbook on animal-assisted therapy: Foundations and guidelines for

animal-assisted interventions (4th ed., pp. 333–337). Waltham, MA:

Elsevier.

Crossman, M. K., Kazdin, A. E., & Knudson, K. (2015). Brief unstruc-

tured interaction with a dog reduces distress. Anthrozoös, 28, 649–659.

doi:10.1080/08927936.2015.1070008

Ebesutani, C., Okamura, K., Higa-McMillan, C., & Chorpita, B. F. (2011).

A psychometric analysis of the positive and negative affect schedule for

children–parent version in a school sample. Psychological Assessment,

23, 406–416. doi:10.1037/a0022057

Ebesutani, C., Regan, J., Smith, A., Reise, S., Higa-McMillan, C., &

Chorpita, B. F. (2012). The 10-item positive and negative affect sche-

dule for children, child and parent shortened versions: Application of

item response theory for more efﬁcient assessment. Journal of

Psychopathology and Behavioral Assessment, 34, 191–203.

doi:10.1007/s10862-011-9273-2

Erskine, H. E., Mofﬁtt, T. E., Copeland, W. E., Costello, E. J., Ferrari, A.

J., Patton, G., … Scott, J. G. (2015). A heavy burden on young minds:

The global burden of mental and substance use disorders in children and

youth. Psychological Medicine, 45, 1551–1563. doi:10.1017/

S0033291714002888

Faul, F., Erdfelder, E., Lang, A., & Buchner, A. (2007). G*Power 3: A

ﬂexible statistical power analysis program for the social, behavioral, and

biomedical sciences. Behavior Research Methods, 39, 175–191.

doi:10.3758/BF03193146

annery-Schroeder, E. C., & Kendall, P. C. (2000). Group and individual

cognitive-behavioral treatments for youth with anxiety disorders: A

randomized clinical trial. Cognitive Therapy and Research, 24,251–

278. doi:10.1023/A:1005500219286

Friesen, L. (2010). Exploring animal-assisted programs with children in

school and therapeutic contexts. Early Childhood Education Journal,

37, 261–267. doi:10.1007/s10643-009-0349-5

Gulliver, A., Grifﬁths, K. M., & Christensen, H. (2010). Perceived barriers

and facilitators to mental health help-seeking in young people: A systema-

tic review. BMC Psychiatry, 10,113–1 22. doi:10.1 186/1471-24 4X-10-113

Gunnar, M. R., Talge, N. M., & Herrera, A. (2009). Stressor paradigms in

developmental studies: What does and does not work to produce mean

increases in salivary cortisol. Psychoneuroendocrinology, 34, 953–967.

doi:10.1016/j.psyneuen.2009.02.010

Handlin, L., Hydbring-Sandberg, E., Nilsson, A., Ejdebäck, M., Jansson,

A., & Uvnäs-Moberg, K. (2011). Short-term interaction between dogs

and their owners: Effects on oxytocin, cortisol, insulin and heart rate—

An exploratory study. Anthrozoös, 24, 301–315. doi: 10.2752/

175303711X13045914865385

Hankin, B. L., Badanes, L. S., Abela, J. R. Z., & Watamura, S. E. (2010).

Hypothalamic-pituitary-adrenal axis dysregulation in dysphoric children

and adolescents: Cortisol reactivity to psychosocial stress from preschool

through middle adolescence. Biological Psychiatry, 68,484–490.

doi:10.1016/j.biopsych.2010.04.004

Hellhammer,D.H.,Wüst,S.,&Kudielka,B.M.(2009). Salivary cortisol as

a biomarker in stress research. Psychoneuroen docrinology, 34, 163–171.

doi:10.1016/j.psyneuen.2008.10.026

Hexdall, C. M., & Huebner, E. S. (2008). Subjective well-being in pedia-

tric oncology patients. Applied Research in Quality of Life, 2, 189–208.

doi:10.1007/s1

1482-008-9037-7

International Association of Human– Animal Interaction Organizations.

(2013). IAHAIO white paper: The IAHAIO deﬁnitions for animal

assisted intervention and animal assisted activity and guidelines for

wellness of animals involved. Chicago, IL: Author.

Kaminski, M., Pellino, T., & Wish, J. (2002). Play and pets: The physical

and emotional impact of child-life and pet therapy on hospitalized

children. Children’s Health Care, 31, 321–335. doi:10.1207/

S15326888CHC3104_5

Kataoka, S. H., Zhang, L., & Wells, K. B. (2002). Unmet need for mental

health care among U.S. children: Variation by ethnicity and insurance

status. American Journal of Psychiatry, 159, 1548–1555. doi:10.1176/

appi.ajp.159.9.1548

Kendall, P. C. (1994). Treating anxiety disorders in children: Results of a

randomized clinical trial. Journal of Consulting and Clinical

Psychology, 62, 100 –110.

Kerns, K. A., Stuart-Parrigon, K. L., Coifman, K. G., Van Dulmen, M. H.

M., & Koehn, A. (2017). Pet dogs: Does their presence inﬂuence

preadolescents’ emotional response to a social stressor? Social

Development. doi:10.1111/sode.12246

Kertes, D. A., Liu, J., Hall, N. J., Hadad, N. A., Wynne, C. D. L., & Bhatt,

S. S. (2017). Effect of pet dogs on children’s perceived stress and

cortisol stress response. Social Development, 26, 382–401.

doi:10.1111/sode.12203

Kessler, R. C., Berglund, P., Demler, O., Jin, R., Merikangas, K. R., &

Walters, E. E. (2005). Lifetime prevalence and age-of-onset distributions

of DSM-IV disorders in the National Comorbidity Replication. Archives

of General Psychiatry, 62,593–602. doi:10.1001/archpsyc.62.6.593

Khalfa, S., Bella, S. D., Roy, M., Peretz, I., & Lupien, S. J. (2003). Effects

of relaxing music on salivary cortisol level after psychological stress.

Annals of the New York Academy of Sciences, 999,

374–376.

Khoury, J. E., Gonzalez, A., Levitan, R. D., Pruessner, J. C., Chopra, K.,

Santo Basile, V., … Atkinson, L. (2015). Summary cortisol reactivity

indicators: Interrelations and meaning. Neurobiology of Stress, 2,34–43.

doi:10.1016/j.ynstr.2015.04.002

Kirschbaum, C., & Hellhammer, D. H. (1989). Salivary cortisol in psycho-

biological research: An overview. Neuropsychobiology, 22,150–169.

doi:10.1159/000118611

EFFECTS OF DOGS ON ANXIETY, AFFECT, AND AROUSAL

Krause-Parello, C. A., Thames, M., Ray, C. M., & Kolassa, J. (2018).

Examining the effects of a service-trained facility dog on stress in

children undergoing forensic interview for allegations of child sexual

abuse. Assessment and Interviewing, 27, 305–320.

Kruger , K. A., Trachtenber g, S. W., & Serpell, J. A. (2004). Can animals help

humans heal? Animal-assisted interventions in adolescent mental health.

Philadelphia, PA: University of Pennsylvania School of Veterinary Medicine.

Linden, W., Earle, T. L., Gerin, W., & Christenfeld, N. (1997).

Physiological stress reactivty and recovery: Conceptual siblings sepa-

rated at birth? Journal of Psychosomatic Research, 42,117–135.

Marino, L. (2012). Construct validity of animal-assisted therapy and

activities: How important is the animal in AAT? Anthrozoös, 25,

s139–s151. doi:10.2752/175303712X13353430377219

Muris, P., Rapee, R., Meesters, C., Schouten, E., & Geers, M. (2003).

Threat perception abnormalities in children: The role of anxiety dis-

orders symptoms, chronic anxiety, and state anxiety. Anxiety Disorders,

17, 271–287. doi:10.1016/S0887-6185(02)00199-8

Nimer, J., & Lundahl, B. (2007). Animal-assisted therapy: A meta-analy-

sis. Anthrozoös, 20, 225–238. doi:10.2752/089279307X224773

Poresky, R. H., Hendrix, C., Mosier, J. E., & Samuelson, M. L. (1987).

The CABS: Internal reliability and construct validity. Psychological

Reports, 60, 743–746. doi:10.2466/pr0.1987.60.3.743

Pruessner, J. C., Kirschbaum, C., Meinlschmid, G., & Hellhammer, D. H.

(2003). Two formulas for computation of the area under the curve

represent measures of total hormone concentration versus time-depen-

dent change. Psychoneuroendocrinology, 28, 916–931.

Ramsay, D., & Lewis, M. (2003). Reactivity and regulation in cortisol and

behavioral responses to stress. Child Development, 74, 456–464.

Reddy, R., Palmer, C. A., Jackson, C., Farris, S. G., & Alfano, C. A.

(2017).

Impact of sleep restriction versus idealized sleep on emotional

experience, reactivity and regulation in healthy adolescents. Journal of

Sleep Research, 26, 516–525. doi:10.1111/jsr.12484

Seligman, L. D., Ollendick, T. H., Langley, A. K., & Baldacci, H. B.

(2004). The utility of measures of child and adolescent anxiety: A meta-

analytic review of the revised children’s manifest anxiety scale, the

state-trait anxiety inventory for children, and the child behavior

checklist. Journal of Clinical Child & Adolescent Psychology, 33,

557–565. doi:10.1207/s15374424jccp3303_13

Souter, M. A., & Miller, M. D. (2007). Do animal-assisted activities

effectively treat depression? A meta-analysis. Anthrozoös, 20, 167–

180. doi:10.2752/175303707X207954

Spielberger, C. D., Gorsuch, R. L., Lushene, R., Vagg, P. R., & Jacobs, G.

A. (1973). State-trait anxiety inventory for children: Manual, instru-

ment, and scoring guide. Menlo Park, CA: Mind Garden, Inc.

Stewart, A., & Strickland, O. (2013). The companion animal in a work

situation: The roles of task difﬁculty and prior companion-animal guar-

dianship in state anxiety. Society & Animals, 21, 249–265. doi:10.1163/

15685306-12341287

Triebenbacher, S. L. (1999). Re-evaluation of the CABS. Anthrozoös, 12,

169–173. doi:10.2752/089279399787000200

Tsai, C., Friedmann, E., & Thomas, S. A. (2010). The effects of animal-

assisted therapy on stress responses in hospitalized children.

Anthrozoös, 23, 245–258. doi:10.2752/175303710X12750451258977

U.S. Census Bureau (2016). New Haven County, Connecticut; New Haven

city, Connecticut. Re trieved from https://www.census.gov/quickfacts/

fact/table/newhavencountyconnecticut,newhavencityconnecticut/

PST045216

Vagnoli, L., Caprilli, S., Vernucci, C., Zagni, S., Mugnai, F., & Messeri, A.

(2015). Can presence of a dog reduce pain and distress in children

during venipuncture? Pain Management Nursing, 16,8

9–95.

doi:10.1016/j.pmn.2014.04.004

Van Breukelen, G. J. P. (2006). ANCOVA versus change from baseline

had more power in randomized studies and more bias in nonrandomized

studies. Journal of Clinical Epidemiology, 59, 920 –925. doi:10.1016/j.

jclinepi.2006.02.007

Vickers, A. J., & Altman, D. G. (2001). Analysing controlled trials with

baseline and follow up measurements. BMJ, 323, 1123–1124.

Zhang, S., Paul, J., Nantha-Aree, M., Buckley, N., Shahzad, U., Cheng, J.,

… Thabane, L. (2014). Empirical comparison of four baseline covariate

adjustment methods in analysis of continuous outcomes in randomized

controlled trials. Clinical Epidemiology, 6, 227–235. doi:10.2147/CLEP.

S56554

14 CROSSMAN ET AL.