I need 2 page paper due in 3 hours

A

K

h h c d a t m i

f

1 d

Original article

Prevalence and Frequency of Internet Harassment Instigation: Implications for Adolescent Health

Michele L. Ybarra, M.P.H., Ph.D.a,*, and Kimberly J. Mitchell, Ph.D.b aInternet Solutions for Kids, Inc., Irvine, California

bCrimes against Children Research Center, Family Research Laboratory, University of New Hampshire, Durham, New Hampshire

Manuscript received September 18, 2006; manuscript accepted March 15, 2007

bstract Purpose: Youth psychosocial and behavioral characteristics are examined based upon varying frequency of Internet harassment perpetration online. Methods: Data are from the Second Youth Internet Safety Survey, a national telephone survey of youth between the ages of 10 and 17 years (N � 1,500). Interviews took place between March and June 2005. Results: In all, 6% of youth reported frequent Internet harassment perpetration, an additional 6% reported occasional perpetration, and 17% reported limited perpetration of Internet harassment in the previous year. In general, behavioral and psychosocial problems increased in prevalence as the intensity of harassing behavior increased. Rule-breaking problems were reported three times more frequently by occasional perpetrators (p � .002) and seven times more frequently by frequent perpetrators (p � .001) as compared to otherwise similar youth who never harassed others in the previous year. Aggression problems were associated with twofold increased odds of limited perpetration (p � .03) and ninefold increased odds of frequent perpetration of Internet harassment (p � .001) among otherwise similar youth. Girls were 50% more likely to be limited perpetrators (p � .02), whereas boys were three times more likely to be frequent perpetrators of harassment online (p � .001). Conclusions: A categorical definition of Internet harassment behavior reveals differences among youth who perpetrate online harassment at different frequencies. Findings reinforce previous research that youth who harass others online are likely facing concurrent behavioral and psychos- ocial challenges. Internet harassment perpetration may be a marker for a larger constellation of psychosocial problems. © 2007 Society for Adolescent Medicine. All rights reserved.

Journal of Adolescent Health 41 (2007) 189 –195

eywords: Harassment; Internet; Behavior problems; Bullying

c c l s o e b h o b a

Research suggests that bullies face multiple mental ealth problems and represent an important adolescent ealth issue. Bullies tend to have concurrent psychosocial hallenges, including alcohol and other substance use [1–3], epression [1,4 –7], and aggressive behavior [8 –10], as well s poor caregiver– child relationships [8]. Over the long erm, bullies are much more likely than other youth to anifest antisocial behavior [7,9], sometimes culminating

n a criminal conviction [8].

*Address reprint requests to: Dr. Michele L. Ybarra, Internet Solutions or Kids, Inc., 74 Ashford, Irvine, CA 92618.

iE-mail address: [email protected]

054-139X/07/$ – see front matter © 2007 Society for Adolescent Medicine. All oi:10.1016/j.jadohealth.2007.03.005

Although a relatively new field, research suggests that hildren and adolescents who harass others online are facing oncurrent mental health problems akin to traditional bul- ying, including a poor emotional bond with their caregiver, ubstance use, delinquent behavior, and peer victimization nline as well as offline [11]. Despite similarities, differ- nces between offline bullying and online harassment have een noted. Males and females are equally likely to report arassing others online [11], whereas most literature on ffline bullying suggests that boys are more likely to be ullies than girls [3,5,8,9,12,13]. One study has found that s age increases, the likelihood of being an online harasser

ncreases [11]. In comparison, the likelihood of being a

rights reserved.

b [ p i v f r

F

i t o b [ K b t a K b t a ( t

t D s s b a s s v [ w p i p t fl a s v i f

M

d s a

U a

S

U s h w m c

S

o p a o s A A n h h t f a c Y

M

H i t i t s y o fi t [ i r

T F

F

N 1 2 3

190 M.L. Ybarra and K.J. Mitchell / Journal of Adolescent Health 41 (2007) 189 –195

ully offline tends to decrease with age and grade 3,8,10,14], suggesting aspects of online harassment such as ower dynamics may differ from traditional bullying. With ncreasing numbers of youth online [15,16] and great ad- ances in Internet technology in the last 5 years [15–19], urther investigation of youth perpetration of Internet ha- assment is warranted.

requency of offline bullying

Probably because the common definition of bullying ndicates the behavior must be frequent and continue over ime [9], little is known about young people who bully thers infrequently. Nonetheless, studies suggest that most ullying behavior occurs infrequently rather than frequently 1,10,20]. For example, in a cross-national study, Smith- huri et al report that 24% of youth in the United States ully others infrequently (once or twice) during the school erm [10]. In contrast, 9% of youth bully others sometimes nd 6% bully others frequently (weekly or more often) [10]. altalia-Heino et al also report that among Finnish students, eing involved in infrequent bullying (either as the perpe- rator or the victim) is much more common (38% of girls nd 54% of boys) than being involved in frequent bullying 8% of girls and 17% of boys) during the current school erm [1].

It seems that as the frequency of involvement increases, he prevalence of mental health problems also increases. epressive symptoms, anxiety, excessive psychosomatic

ymptoms, excessive alcohol consumption, and use of other ubstances all increased in frequency as the involvement in ullying increased among participants in Kaltalia-Heino et l’s study [1]. Similarly, Solberg and Olweus report that ocial disintegration, negative self-evaluations, and depres- ive tendencies worsen as the frequency of bullying in- olvement increases among the 16,410 youth they surveyed 21]. How this pattern may apply to bullies in the online orld has yet to be reported. This knowledge is especially ertinent to our understanding of adolescent Internet health ssues as Internet use continues to increase among young eople [15,16]. It may be that Internet harassment perpetra- ion represents a range of behaviors, on the one hand re- ecting minor disagreements between friends typical in dolescence [22], and on the other hand reflecting aggres- ion of a greater frequency that fits our more traditional iew of bullying. Identifying where the line may be has mportant implications for prevention and intervention ef- orts.

ethods

The Second Youth Internet Safety Survey (YISS-2), con- ucted between March and June of 2005, was a national urvey of 1,500 households of youth between the ages of 10

nd 17 years (mean 14.2, SD 2.1) and one caregiver. The

�

niversity of New Hampshire Institutional Review Board pproved and supervised the protocol.

ampling method

A sample size of 1,500 households across the continental nited States was predetermined to ensure a maximum

ampling error of 2.5% at the .05 significance level. House- olds were identified via random digit dialing. Interviewers ere conducted by the national survey research firm Schul- an, Ronca, and Bucuvals, Inc. (SRBI). Using common

alculations [23], the response rate was 45%.

ample

Youth were required to have used the Internet at least nce a month for the past 6 months and to speak English. To romote a diverse sample, Internet access was allowed from nywhere (i.e., home access was not required). Half (51%) f youth respondents were female. Three quarters (76%) elf-identified as White, 13% Black, 2.5% Indian, 3% sian, and 1% “other” (3% declined to provide an answer). lmost one in 10 (9%) self-identified as of Hispanic eth- icity. Twenty-two percent of adults reported a household igh school education or less, and 35% reported an annual ousehold income of $50,000 or less. YISS-2 households ended to be more educated, have higher incomes, and more requently cited White race as compared to the national verage, as was expected base upon recent surveys of the urrent Internet population [15,16]. Further details about ISS-2 methodology are published elsewhere [24].

easures

arassment perpetration. Our main outcome for the current nvestigation was engagement in online harassment perpe- ration in the previous year. Two types of harassing behav- or were queried: (1) the number of times in the last year hat youth had used the Internet to harass or embarrass omeone they were mad at; and (2) the number of times that outh had made rude or nasty comments to someone else nline. Answers were coded as: never, once, twice, three to ve times, and six or more times (Table 1). Event charac-

eristics for both types of harassing behavior were similar 25], suggesting that they could be combined into a global ndicator of harassment. To do so while also capturing elative frequency of harassment, an algorithm was created:

able 1 requency of harassing behaviors (n � 1,500)

requency Rude or nasty comments Harassing or embarrassing someone

ever 72.3 (1084) 91.4 (1371) time 9.7 (145) 5.1 (76) times 6.3 (95) 1.9 (29)

–5 times 6.0 (90) 1.1 (16)

6 times 5.7 (86) 0.5 (8)

y o y t c i b q e i

P V d p t w a b ( i

n c p i a v ( t

y t b a e f

B h l a c i i S r b

I m t m c p u

a u w e

D w a r

S

c t P b i o i r t p h h a o t l o h

R

h O m e a o i I s ( l

I s p p i

191M.L. Ybarra and K.J. Mitchell / Journal of Adolescent Health 41 (2007) 189 –195

outh who reported engaging in either harassing behavior ne or two times were classified as “limited” perpetrators; outh who reported engaging in either harassing behavior hree to five times, or both behaviors two times each were lassified as “occasional” perpetrators; those who engaged n one of the two types of behaviors six or more times or in oth behaviors three to five times were classified as “fre- uent” perpetrators. Because only eight individuals reported ngaging in both behaviors six or more times, they were ncluded with the “frequent” harassers.

sychosocial problems. Six questions from the Juvenile ictimization Questionnaire [26] were asked to measure ifferent types of offline victimization. Sexual abuse and hysical abuse in the previous year (yes/no) were combined o ensure cell stability (i.e., sufficient numbers of youth ithin a category to allow statistical comparisons). Being

ttacked (yes/no), being hit or jumped by a gang (yes/no), eing hit by peers (yes/no), and being picked on by peers yes/no) were combined to reflect any interpersonal victim- zation (yes/no).

Using a four-point Likert scale (1 � all of the time, 4 � ever/rarely), the respondents rated how frequently their aregiver nagged them, yelled at them, and took away their rivileges. Based upon exploratory factor analysis suggest- ng a common latent factor (Eigenvalue 1.69; percent vari- nce 56.2), items were reverse coded and a summation ariable was created to measure global parent– child conflict mean 3.98, SD 1.43). This was dichotomized at 1 SD above he mean because of indications of nonlinearity.

Being a target of Internet harassment was indicated if outh responded positively to at least one of the following wo questions: (1) Did you ever feel worried or threatened ecause someone was bothering or harassing you online? nd (2) Did anyone ever use the Internet to threaten or mbarrass you by posting or sending messages about you or other people to see?

ehavior. The Youth Self-Report (YSR) of the Child Be- avior Check List [27] was used to measure behavior prob- ems. It is one of the most recognized instruments used to ssess problem behavior from the youth perspective. The urrent study includes two subscales measuring externaliz- ng problems (rule-breaking behavior and aggressive behav- or) and one internalizing subscale (withdrawn/depressed). ubscale scores were categorized according to Achenbach’s ecommendations and dichotomized to reflect youth in the orderline or clinical range versus nonclinical range.

nternet use. Four different aspects of Internet use were easured: high experience with the Internet, high impor-

ance of the Internet to oneself, and typically spending 4 or ore days per week or 2 or more hours per day online. A

ommon latent factor was suggested (Eigen value 1.71; ercent variance 42.9). A summation score was created

sing these four variables and was dichotomized at 1 SD i

bove the mean versus all others to reflect “high Internet se.” Activities that reflected interaction with others also ere queried and included in the current analysis (i.e.,

mailing, blogging, instant messaging, and chat room use).

emographic characteristics. The individual’s age and sex ere reported by caregivers, as were household education

nd annual household income. Race and ethnicity were eported by the youth themselves.

tatistical analyses

Using Stata 9 [28], missing and nonresponsive data were oded to the sample mode. In most cases, this affected less han 1% of data; income (8%, n � 123) being the exception. sychosocial and behavioral attributes were then compared ased upon reported frequency of Internet harassing behav- or. The Cuzick nonparametric test for trend across the four rdered groups was conducted for each bi-variate compar- son [28,29]. This test is an extension of the Wilcoxon ank-sum test and is useful when measuring trends across hree or more ordered, independent groups. Next, to com- are the estimates generated using the classic definition of arassment (i.e., frequent) versus a graduated definition of arassment perpetration (i.e., limited, occasional, frequent), logistic regression model was estimated to quantify the

dds of engaging in harassment perpetration six or more imes in the previous year versus fewer and multinomial ogistic regression then was used to estimate the conditional dds of engaging in any of the three frequencies of Internet arassment versus none.

esults

Almost one in three youth (29%, n � 435) reported arassing someone online at least once in the previous year. f the 416 youth who reported making rude or nasty com- ents to another person, 82% (n � 343) said that someone

lse made rude or nasty comments to the respondent first in t least one incident, and 25% (n � 106) said they were the ne to make rude or nasty comments first in at least one ncident. Similarly, of the 129 youth who reported using the nternet to harass or embarrass someone, 81% (n � 104) aid that someone else did it to the respondent first, and 31% n � 40) acknowledged doing it to someone else first at east once.

Of all youth respondents, 6% (n � 90) reported frequent nternet harassment perpetration, 6% (n � 97) reported occa- ional perpetration, and 17% (n � 248) reported limited per- etration of Internet harassment. As shown in Table 2, most sychosocial characteristics and behavioral problems increased n prevalence as the frequency of harassment perpetration

ncreased. Internet use also appeared to increase in intensity as

h o p h

C h

d t i y t p t w p

s p w p f h a t p

a b y f s s w s c

c o e a n a

I h G t . f p y �

T P

P

P

B

I

Y

H

)

192 M.L. Ybarra and K.J. Mitchell / Journal of Adolescent Health 41 (2007) 189 –195

arassment perpetration increased in frequency. As frequency f harassing behavior increased, youth age (p � .001), and the revalence of white youth (p � .001) increased, whereas ousehold income decreased (p � .001).

omparison of dichotomous versus increasing frequency arassment perpetration

Two regression models were estimated, one using a ichotomous definition of harassment perpetration, and he other using a categorical definition reflecting increas- ng frequency of perpetration. Each was a saturated model, ielding estimates adjusted for all other variables included in he model. As shown in Table 3, the associations between sychosocial and behavior problems with harassment perpe- ration for the dichotomous definition (“frequent harassers”) ere similar to estimates for youth who were “frequent erpetrators in the categorical definitions.

The likelihood of reporting behavioral problems and ome psychosocial problems increased as harassment per- etration increased. For example, rule-breaking problems ere reported three times more frequently by occasional erpetrators (p � .002) and seven times more frequently by requent perpetrators compared to otherwise similar, non- arassing youth (p � .001). Aggression problems were ssociated with twofold increased odds of limited perpetra- ion (p � .03) and ninefold increased odds of frequent

able 2 revalence rate of psychosocial characteristics and behavior problems by

ersonal characteristics Frequency of harassing behavior

None (71%, n � 1,065)

Limited (17%, n

sychosocial characteristics Interpersonal victimization 32.4 (345) 50.4 (12 High caregiver–child conflict 10.4 (111) 15.7 (39 Target of Internet harassment 4.9 (52) 14.9 (37 Physical/sexual victimization 2.1 (22) 5.2 (13

ehavior problems Aggressive 3.4 (36) 7.7 (19 Rule breaking 3.1 (33) 6.5 (16 Withdrawn/depressed 3.8 (41) 4.4 (11

nternet use High use 20.0 (213) 37.5 (93 Instant messaging 59.8 (637) 83.1 (20 E-mail 73.8 (786) 87.1 (21 Chat rooms 24.6 (262) 38.7 (96 Blogging 13.6 (145) 22.2 (55

outh demographic characteristics White race 73.6 (784) 79.4 (19 Male 49.7 (529) 38.7 (96 Age, years, mean (SD) 13.9 (2.1) 14.6 (1.8 Hispanic ethnicity 8.4 (89) 10.9 (27

ousehold characteristics Lower income (�$50,000) 37.5 (399) 30.2 (75 High school education or less 22.8 (243) 23.8 (59

erpetration (p � .001) compared to nonharassing youth f

fter adjusting for all other characteristics. The likelihood of eing a target of Internet harassment was elevated for all outh perpetrators of Internet harassment, but especially so or those who were frequent perpetrators among otherwise imilar youth. Offline interpersonal victimization was more trongly related to limited and occasional perpetration, hereas physical and sexual victimization was more

trongly related to frequent perpetration holding all other haracteristics equally.

High Internet use generally and instant messaging spe- ifically were both associated with significantly elevated dds of reporting harassment perpetration and this was specially true for frequent perpetrators after adjusting for ll other characteristics. On the other hand, blogging was ot significantly associated with harassment perpetration of ny frequency among otherwise similar youth.

Differences in demographic characteristics were noted. ncreasing age was associated with increasing frequency of arassment perpetration among otherwise similar youth. irls were significantly more likely to be limited perpetra-

ors (adjusted conditional odds ratio [ACOR] � 1.5, p � 02) whereas boys were significantly more likely to be requent perpetrators of online harassment (ACOR � 3.9, � .001) after adjusting for all other characteristics. White

outh (ACOR � 4.3, p � .001) and Hispanic youth (ACOR 4.1, p � .003) appeared to be especially likely to be

g behavior online (N � 1,500)

Statistical comparison

) Occasional (6%, n � 97)

Frequent (6%, n � 90)

Test for trend (z)

p

59.8 (58) 55.6 (50) 7.2 �.001 27.8 (27) 27.8 (25) 6.3 �.001 18.6 (18) 25.6 (23) 8.6 �.001

4.1 (4) 12.2 (11) 5.1 �.001

8.2 (8) 28.9 (26) 9.0 �.001 15.5 (15) 32.2 (29) 11.1 �.001

5.2 (5) 8.9 (8) 2.1 .04

42.3 (41) 66.7 (60) 10.9 �.001 90.7 (88) 95.6 (86) 10.1 �.001 95.9 (93) 93.3 (84) 6.9 �.001 45.4 (44) 55.6 (50) 7.8 �.001 23.7 (23) 22.2 (20) 3.7 �.001

84.5 (82) 86.7 (78) 3.8 �.001 54.6 (53) 66.7 (60) 1.9 0.06 15.2 (1.5) 15.8 (1.4) 10.0 �.001

5.2 (5) 13.3 (12) 1.0 0.30

32.0 (31) 25.6 (23) �2.8 .005 21.7 (21) 13.3 (12) �1.6 0.11

harassin

online

� 248

5) ) ) )

) ) )

) 6) 6) ) )

7) ) ) )

)

requent perpetrators of Internet harassment.

D

l r a o a s v t i

P

e f s h y t r t

[ f i o c

t b c r a f w o m t p

l m l

T C

P

P

B

I

D

H

a

e t harra

193M.L. Ybarra and K.J. Mitchell / Journal of Adolescent Health 41 (2007) 189 –195

iscussion

As predicted [1,10,20], psychosocial and behavior prob- ems increase in frequency as the frequency of online ha- assment perpetration increases. This is especially true for ggression and rule breaking problems, and being the target f Internet harassment for youth. Importantly, youth who re limited or occasional perpetrators report these psycho- ocial challenges at significantly higher rates than unin- olved youth. A graduated definition of harassment perpe- ration reveals psychosocial and behavior problems in less ntensively involved youth.

ower dynamics

Consistent with the previous study by Ybarra and Mitch- ll [11], as age increases so too does the likelihood and requency of Internet harassment perpetration. Observations uggesting that older youth are more likely to engage in arassment online as compared to studies reporting younger outh engage in bullying offline [3,8,10,14] may indicate hat the power dynamics online are different and for some eason(s) more attractive to older youth. Given the impor-

able 3 omparison of findings for a dichotomous versus categorical definition of

ersonal characteristics Dichotomous comparisona

(n � 1,151) Categori

Frequent harassers: 6� times (7%, n � 86)

Limited (17%, n

AOR 95% CI p ACOR

sychosocial characteristics Interpersonal victimization 1.7 0.9, 3.1 .13 2.2 High caregiver–child conflict 2.1 1.0, 4.4 .06 1.2 Target of Internet harassment 4.7 2.1, 10.7 �.001 2.4 Physical/sexual victimization 3.3 0.8, 14.4 .11 1.6

ehavior problems Aggressive 7.9 2.9, 21.7 �.001 2.1 Rule breaking 5.7 2.2, 14.8 �.001 1.3 Withdrawn/depressed 0.8 0.2, 2.9 .78 0.9

nternet use High Internet use 5.8 3.0, 11.0 �.001 1.7 Instant messaging 9.9 2.4, 41.2 .002 2.1 E-mail 0.8 0.3, 2.6 .74 1.0 Chat rooms 2.6 1.4, 4.8 .002 1.5 Blogging 0.8 0.4, 1.7 .53 1.0

emographic characteristics White race 5.2 2.0, 13.6 .001 1.5 Male 3.0 1.5, 5.9 .001 0.7 Age 1.7 1.4, 2.2 �.001 1.1 Hispanic ethnicity 4.4 1.5, 12.6 .006 1.8

ousehold characteristics Lower income (�$50,000) 0.7 0.3, 1.5 .35 0.7 High school education or less 0.4 0.2, 1.1 .09 1.2

Abbreviations: AOR � adjusted odds ratio (estimates are adjusted for a djusted conditional odds ratio (estimates are adjusted for all other variabl

a n � 1,151 Individuals who reported engaging in either harassment b ngaging in either behavior (n � 1,065). Youth who reported less frequen

ant social development that occurs in later adolescence b

30], the fact that older youth are more likely involved as requent perpetrators may indicate that instead of engaging n normative and important personal development, some lder youth are not developing the necessary skills to suc- eed in adulthood.

As suggested by Ybarra and Mitchell [11], it is possible hat some Internet harassers are acting in retaliation to ullying they received in-person or harassment they re- eived online. This is consistent with the finding that ha- assers in the current survey are significantly more likely lso to report offline interpersonal victimization, and is urther supported by the report from four in five harassers ho say their harassing behavior was in response to an nline harassment incident initiated by someone else. This ay suggest that one important prevention technique will be

o equip youth with effective means of more positive and roductive conflict management skills.

In the current investigation boys are significantly more ikely than girls to be frequent perpetrators in online harass- ent whereas girls are significantly more likely to report

imited perpetration. This is in contrast to expectations that

ing behavior

parisons (compared with nonharassers) (n � 1,500)

Occasional (6%, n � 97)

Frequent (6%, n � 90)

I p ACOR 95% CI p ACOR 95% CI p

9 �.001 3.0 1.9, 4.8 �.001 1.8 1.0, 3.1 .05 8 .50 2.0 1.2, 3.6 .01 1.3 0.7, 2.7 .39 9 �.001 2.8 1.5, 5.4 .001 4.6 2.3, 9.5 �.001 5 .26 1.0 0.3, 3.4 .97 3.2 1.0, 10.6 .06

1 .03 1.6 0.6, 4.0 .34 9.3 3.9, 21.8 �.001 6 .49 3.4 1.5, 7.3 .002 7.1 3.1, 16.1 �.001 9 .77 0.8 0.3, 2.2 .62 0.9 0.3, 2.7 .84

4 .001 1.9 1.2, 3.1 .006 6.4 3.6, 11.6 �.001 3 .001 3.0 1.4, 6.6 .005 11.3 3.1, 41.9 �.001 6 .92 3.0 1.0, 9.1 .05 1.0 0.3, 2.8 .93 0 .01 1.7 1.1, 2.6 .03 2.3 1.4, 4.0 .002 5 1.0 1.0 0.6, 1.7 .98 0.8 0.4, 1.6 .57

2 .05 2.1 1.1, 3.9 .03 4.3 1.9, 9.9 �.001 9 .02 1.5 1.0, 2.4 .08 3.9 2.1, 7.1 �.001 2 .008 1.3 1.1, 1.5 .001 1.8 1.5, 2.2 �.001 0 .03 0.9 0.3, 2.5 .83 4.1 1.6, 10.4 .003

0 .08 1.0 0.6, 1.6 .94 0.8 0.4, 1.5 .52 8 .26 1.1 0.6, 2.0 .75 0.5 0.3, 1.2 .13

variables in the model, i.e., all characteristics shown in table; ACOR � model, i.e., all characteristics shown in table); CI � confidence interval.

6 or more times (n � 86) are compared with those who reported never ssing behavior (n � 349) were not included in the model.

harass

cal com

� 248)

95% C

1.6, 2. 0.8, 1. 1.5, 3. 0.7, 3.

1.1, 4. 0.6, 2. 0.4, 1.

1.2, 2. 1.4, 3. 0.6, 1. 1.1, 2. 0.7, 1.

1.0, 2. 0.5, 0. 1.0, 1. 1.1, 3.

0.5, 1. 0.9, 1.

ll other es in the ehavior

oys and girls will be equally likely to report harassing

o d t b g p s o a t p B 2 I

p t r a e m e b t F

N m

h i f a I t m p q y s i t i t a

P I

i r c o p

a r g p b q l t g p u o s

S

i g s t a [ m p t a t n

C

y b p a G b h l

A

1 t f C P i d r

194 M.L. Ybarra and K.J. Mitchell / Journal of Adolescent Health 41 (2007) 189 –195

thers online [11]. Findings persist even after underlying ifferences in aggression and rule-breaking problems are aken into account. These sex differences are not explained y behavior problems that are more typical in boys than irls. Perhaps boys are more likely to retaliate than girls to erceived harassment. It may also be that boys experience a tronger power differential between themselves and the target f their harassment, which reinforces the behavior and leads to greater frequency of perpetration. The power dynamics be-

ween girl harassers and their targets may not be as strong, or ower may not generally be a goal for girls who harrass. ecause this sex difference appears to have emerged since 000 [31], it also is possible that there are underlying shifts in nternet culture that help to explain the findings as well.

Sex and age differences in online and offline harassment rovide intriguing clues about the power dynamics for fu- ure research. Examinations of how the expression of ha- assment may vary for girls versus boys online, with specific ttention to potential differences in the way boys and girls xpress emotional— or mental health—related difficulties, ight be of great use to future targeted interventions. For

xample, it is possible that because girls are more likely than oys to be limited perpetrators, occasional or frequent perpe- ration for girls denotes greater psychosocial problems overall. indings should be replicated in future studies.

ormative disagreements between friends, or something ore problematic?

Some may argue that youth who are engaging in Internet arassment less frequently (i.e., once or twice) are display- ng a normative disagreement with friends, as it is normal or conflict to occur between friends [22]. In addition, with lmost one in four youth reporting limited or occasional nternet harassment perpetration in the current investiga- ion, this lesser degree of perpetration is somewhat com- on. A clear distinction in the amount of behavior and

sychosocial problems is observed for those reporting fre- uent versus less frequent perpetration as well. Nonetheless, outh who report limited and occasional perpetration are ignificantly more likely to report psychosocial and behav- or problems as compared with nonharassing youth. Al- hough peer disagreements are normative, the current find- ngs suggest that the expression of this disagreement hrough Internet harassment may not be normative given its ssociation with general psychosocial problems.

sychosocial and behavior problems associated with nternet harassment

As expected, aggressive behavior [8 –10], peer victim- zation online and offline [11], and poor caregiver– child elationships [8,11] are each significantly related to in- reased odds of varying frequencies of harassing behavior nline holding all other characteristics equal. Rule-breaking

roblems also are observed for youth who are occasional m

nd frequent perpetrators of Internet harassment in the cur- ent investigation. These findings further support the sug- estion that youth who harass others online are likely ex- eriencing challenges on multiple fronts. The likelihood of ehavioral and psychosocial problems varies as the fre- uency of harassment perpetration increases. Youth who are imited or occasional perpetrators may represent an oppor- unity to intervene early with potential improvements in eneral psychosocial functioning. Those who are frequent erpetrators appear likely to require more intensive follow- p. Current findings also provide further justification that nline harassment prevention be integrated into traditional, chool-based anti-bullying programs.

tudy limitations

Our findings should be interpreted within the study lim- tations. First, the response rate (45%) is reflective of a eneral decline in response rates for national telephone urveys [32]. However national telephone surveys continue o obtain representative samples of the public and provide ccurate data about the views and experiences of Americans 33]. Second, as a cross-sectional study, we cannot deter- ine temporality. We cannot conclude that aggression

roblems cause a young person to harass others online, or hat alternatively, harassing others causes one to become ggressive. It likely is a bi-directional relationship that con- inues to reinforce both behaviors. Longitudinal studies are eeded to parse this out.

onclusion

In conclusion, current results support the findings that outh who harass others online are likely facing concurrent ehavioral and psychosocial challenges. As frequency of erpetration increases, so too do the odds of psychosocial nd behavior problems. Sex differences also are observed. irls are more likely to be limited perpetrators, whereas oys are more likely to be frequent perpetrators of Internet arassment. Harassment perpetration may be a marker for a arger constellation of psychosocial problems.

cknowledgment

For the purposes of compliance with Section 507 of PL 04-208 (the “Stevens Amendment”), readers are advised hat 100% of the funds for this program are derived from ederal sources (National Center for Missing & Exploited hildren and the Office of Juvenile Justice & Delinquency revention). The total amount of federal funding involved

n this project is $348,767. Points of view or opinions in this ocument are those of the authors and do not necessarily epresent the political position or policies of the U.S. Depart-

ent of Justice or the Department of Homeland Security.

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[ [

[

[

[

195M.L. Ybarra and K.J. Mitchell / Journal of Adolescent Health 41 (2007) 189 –195

eferences

[1] Kaltiala-Heino R, Rimpela M, Rantanen P, Rimpela A. Bullying at school—an indicator of adolescents at risk for mental disorders. J Adolesc 2000;23:661–74.

[2] Nansel T, Craig W, Overpeck M, et al, Health Behavior in School- aged Children Bullying Working Group. Cross-national consistency in the relationship between bullying behaviors and psychosocial ad- justment. Arch Pediatr Adolesc Med 2004;158:730 – 6.

[3] Nansel T, Overpeck M, Pilla RS, et al. Bullying behaviors among US young people: prevalence and association with psychosocial adjust- ment. JAMA 2001;285:2094 –100.

[4] Kaltiala-Heino R, Rimpela M, Martunen M, et al. Bullying, depres- sion, and suicidal ideation in Finnish adolescents: school survey. Br Med J 1999;319:348 –51.

[5] Sourander A, Helstela L, Helenius H, Piha J. Persistence of bullying from childhood to adolescence—a longitudinal 8-year follow-up study. Child Abuse Neglect 2000;24:873– 81.

[6] Saluja G, Iachan R, Scheidt P, et al. Prevalence of and risk factors for depressive symptoms among young adolescents. Arch Pediatr Ado- lesc Med 2004;158:765.

[7] Rigby K. Peer victimization at school and the health of secondary school students. Br J Educ Psychol 1999;69:95–104.

[8] Batsche GM, Knoff HM. Bullies and their victims: understanding a pervasive problem in the schools. School Psychol Rev 1994;23:165–74.

[9] Olweus D. Annotation: Bullying at school: basic facts and effects of a school based intervention program. J Child Psychol Psychiatry 1994;55:1171–90.

10] Smith-Khuri E, Iachan R, Scheidt PC, et al. A cross-national study of violence-related behavior in adolescents. Arch Pediatr Adolesc Med 2004;158:539 – 44.

11] Ybarra M, Mitchell K. Youth engaging in online harassment: asso- ciations with caregiver– child relationships, Internet use, and personal characteristics. J Adolesc 2004;27:319 –36.

12] Kumpulainen K, Rasanen E. Children involved in bullying at ele- mentary school age: their psychiatric symptoms and deviance in adolescence. Child Abuse Neglect 2000;24:1567–77.

13] Forero R, McLellan L, Rissel C, Bauman A. Bullying behavior and psychosocial health among school students in New South Wales, Australia: cross national survey. Br Med J 1999;319:344 – 8.

14] Smith PK, Madsen KC, Moody JC. What causes the age decline in reports of being bullied? Towards a developmental analysis of risks of being bullied. Educ Res 1999;41:267– 85.

15] Lenhart A, Madden, M, Hitlin P. Teens and technology: youth are leading the transition to a fully wired and mobile nation [report].

2005. Washington DC: Pew Internet and American Life.

16] USC Annenberg School Center for the Digital Future. Ten years, ten trends. Year 4. 2004 [report]. The Digital Future Report: Surveying the Digital Future.

17] Horrigan J, Rainie L. The broadband difference. How online Amer- icans’ behavior changes with high-speed Internet connections at home [report]. Washington, DC: Pew Internet Life Project; 2002.

18] Lenhart A, Horrigan J, Fallows D. Content creation online. 44% of U.S. internet users have contributed their thoughts and their files to the online world [report]. 2004. Washington DC: Pew American Life Project.

19] Rainie L. 16% of Internet users have viewed a remote person or place using a web cam. 2005. Washington DC: Pew Internet & American Life Project; 2005. [report]

20] Due P, Holstein BE, Lynch J, et al. Bullying and symptoms among school-aged children: international comparative cross sectional study in 28 countries. Eur J Public Health 2005;15:128 –32.

21] Solberg M, Olweus D. Prevalence estimation of school bullying with the Olweus Bully/Victim Questionnaire. Agress Behav 2003;29:239 – 68.

22] Laursen B. Conflict and social interaction in adolescent relationships. J Res Adolesc 1995;5:55–70.

23] American Association for Public Opinion Research. Standard defini- tions: final dispositions of case codes and outcome rates for surveys. AAPOR. Available at: http://www.aapor.org/pdfs/standarddefs_3.1. pdf [3rd]. 2-1-2005. Lenexa, KS. 11-15-2005.

24] Wolak J, Mitchell K, Finkelhor D. Online victimization of youth: 5 years later. National Center for Missing and Exploited Children. 2006. Available at: http://www.unh.edu/ccrc.

25] Ybarra M, Mitchell K, Finkelhor D, Wolak J. Internet prevention messages: are we targeting the right online behaviors: Arch Pediatr Adolesc Med 2007;161(2):138 – 45.

26] Finkelhor D, Hamby SL, Ormond R, Turner H. The Juvenile Victim- ization Questionnaire: reliability, validity, and national norms. Child Abuse Neglect 2005;29:383– 412.

27] Achenbach TM. Manual for the youth self-report and 1991 profile. Burlington, VT: Department of Psychiatry, University of Vermont; 1991.

28] StataCorp. Stata Statistical Software. Release 7.0 ed. College Station, TX: Stata Corporation; 2000.

29] Cuzick J. A Wilcoxon-type test for trend. Stat Med 1985;4:87–9. 30] Ponton LE, Judice S. Typical adolescent sexual development. Child

Adolesc Psychiatr Clin North Am 2004;13:497–511. 31] Finkelhor D, Mitchell K, Wolak J. Online victimization: A report on

the nation’s young people. 2000. National Center for Missing and Exploited Children. Available at: http://www.unh.edu/ccrc.

32] Curtin R, Presser S, Singer E. Changes in telephone survey nonresponse over the past quarter century. Public Opinion Q 2005;69:87–98.

33] Pew Research Center for the People and the Press. Polls face growing resistance, but still representative. 4-20-2004. Washington DC: The

Pew Research Center for the People and the Press.