Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: This study tested an updated cognitive-behavioral model of generalized problematic Internet use and
Available online 3 April 2010 reports results of a conrmatory analysis of the Generalized Problematic Internet Use Scale 2 (GPIUS2).
Overall, the results indicated that a preference for online social interaction and use of the Internet for
Keywords: mood regulation, predict decient self-regulation of Internet use (i.e., compulsive Internet use and a cog-
Problematic Internet use nitive preoccupation with the Internet). In turn, decient self-regulation was a signicant predictor of the
Computer-mediated communication extent to which ones Internet use led to negative outcomes. Results indicated the model t the data well
CMC
and variables in the model accounted for 27% of the variance in mood regulation scores, 65% of variance
Internet addiction
Preference for online social interaction
in participants decient self-regulation scores, and 61% of variance in the negative outcome scores.
Decient self-regulation 2010 Elsevier Ltd. All rights reserved.
GPIUS2
Measurement
2
* Tel.: +1 302 831 2958; fax: +1 302 831 1892. Although Davis et al. (2002) did conduct a conrmatory analysis of a GPIU
E-mail address: caplan@udel.edu measure, the Online Cognition Scale, the instrument did not assess negative
1
Symptoms of generalized PIU include cognitions and behaviors related to Internet outcomes. Pratarelli and Browne (2002) conducted a conrmatory analysis of a scale
use that are not linked to any specic content (i.e., gambling, shopping, etc.). Rather, measuring Internet use and addiction, but the instrument did not assess GPIU as
generalized PIU occurs when an individual develops problems due to the interper- described in the current paper. Similarly, Meerkerk et al. (2009) employed
sonal contexts available online (Caplan, 2002, 2003; Davis, 2001; Davis, Flett, & conrmatory analysis to validate their compulsive Internet use scale, but that
Besser, 2002). instrument did not assess other aspects of GPIUS.
0747-5632/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.chb.2010.03.012
1090 S.E. Caplan / Computers in Human Behavior 26 (2010) 10891097
elements of Davis (2001) cognitive-behavioral theory of GPIU, Ca- (Caplan, 2002, 2007; LaRose et al., 2003). In an early study, Caplan
plans (2003, 2005) work with preference for online social interac- (2002) found that mood regulation was a signicant cognitive pre-
tion, and the socio-cognitive model of unregulated Internet use dictor of negative outcomes associated with Internet use. In a later
advanced by LaRose and his colleagues (e.g., Kim, LaRose, & Peng, study, Caplan (2007) argued that socially anxious individuals may
2009; LaRose, Lin, & Eastin, 2003). Subsequent sections of this arti- prefer online interaction because it represents a way to mitigate
cle then report a conrmatory analysis of a new measurement their anxiety about self-presentation in interpersonal situations.
model (the GPIUS2 instrument) and results of a formal test of the LaRose and colleagues work on the socio-cognitive model of
hypothesized conceptual model. Before presenting the hypothe- unregulated Internet use also emphasizes the role of mood regula-
sized model in detail, the following section briey elaborates the tion in the development of decient self-regulation.
models key constructs.
1.4. Decient self-regulation
1.1. Key constructs in conceptual model
The literature also suggests that GPIU involves decient self-
Davis (2001; Davis et al., 2002) introduced a cognitive-behav- regulation (Kubey, Lavin, & Barrows, 2001; LaRose, 2001; LaRose,
ioral theory of generalized PIU; asserting that psychosocial prob- Eastin, & Gregg, 2001; LaRose, Mastro, & Eastin, 2001; LaRose
lems such as loneliness or depression predispose individuals to et al., 2003). The current model adopts La Rose and colleagues
develop maladaptive Internet-related cognitions and behaviors view that decient self-regulation of Internet use is a state in
that ultimately result in negative outcomes. According to Davis which conscious self-control is relatively diminished (p. 232).
(2001), GPIU cognitions and behaviors, and the negative outcomes More specically, decient self-regulation refers to a failure to
they lead to, are consequences, rather than causes, of broader psy- adequately monitor ones use, judge ones usage behaviors, and
chosocial problems. From this perspective, psychosocial problems adjust ones pattern of use (Bandura, 1986, 1991). As a conse-
predispose individuals to develop maladaptive cognitions that lead quence, decient self-regulation of ones Internet use may eventu-
to difculties with behavioral impulse control, ultimately resulting ally result in difculties in ones personal relationships, at work,
in negative outcomes associated with their Internet use. The re- and in school (e.g., Kubey et al., 2001). Decient self-regulation
search published since Davis introduced the model has identied is conceptually useful for explaining GPIU because it represents
several specic cognitive and behavioral constructs associated a higher-order construct that reects the interplay between com-
with negative outcomes of Internet use: preference for online so- pulsive behavioral symptoms and obsessive cognitive symptoms
cial interaction, mood alteration, cognitive preoccupation, and of GPIU.
compulsive behavior. The model proposed below makes an important distinction be-
tween the cognitive and behavioral aspects of decient self-regula-
1.2. Preference for online social interaction (POSI) tion in GPIU. Here, the cognitive manifestation of decient self-
regulation involves a preoccupation with the Internet. On the other
Research published subsequent to the introduction of the cogni- hand, compulsive Internet use is a behavioral aspect of decient
tive-behavioral model indicates that one important cognitive self-regulation. As the hypotheses presented below explain, if cog-
symptom of GPIU is a preference for online social interaction (Ca- nitive symptoms of GPIU are salient enough, they lead to behav-
plan, 2003, 2005, 2007; Kim & Davis, 2009; Kim et al., 2009). ioral symptoms that ultimately result in negative outcomes.
According to Caplan (2003) preference for online social interac-
tion is a cognitive individual-difference construct characterized 1.4.1. Cognitive preoccupation
by beliefs that one is safer, more efcacious, more condent, and Cognitive preoccupation refers to obsessive thought patterns
more comfortable with online interpersonal interactions and rela- involving the Internet use (i.e., I cant stop thinking about going
tionships than with traditional FtF social activities (p. 629). Caplan online or when I am off line, I cant stop wondering what is
argues that individuals who are lonely, socially anxious, or have happening online). Shapira and colleagues (2003) proposed that
decient social skills may also develop a heightened POSI because PIU entails a combination of both cognitive preoccupation and a
they perceive online interaction to be less face threatening and problematic pattern of use of the Internet. Similarly, Caplan and
they perceive themselves to be more socially efcacious when High (2007) found that the association between Internet use and
interacting with others online (Caplan, 2003, 2005, 2007). negative outcomes is more pronounced when cognitive preoccu-
Studies support the claim that people who exhibit interpersonal pation is present. In other words, the literature suggests that
problems such as social anxiety and low social skill report greater how individuals think about the Internet may help explain
levels of POSI. Both Caplan (2003) and Kim et al. (2009) found that whether they experience negative outcomes associated with
loneliness was a signicant predictor of POSI. In a 2005 study, Ca- their use.
plan found that individuals self-presentational condence (an as-
pect of social skill) predicted their level of POSIparticipants who 1.4.2. Compulsive internet use
reported lower self-presentational skill were especially likely to At the behavioral level, decient self-regulation takes the form
perceive online social interaction more favorably FtF communica- of compulsive Internet use (Kim & Davis, 2009; Kim et al., 2009). A
tion. Similarly, Kim et al. (2009) also found that decient social number of early publications emphasized the compulsive nature of
skill predicted POSI. In addition, Caplan (2007) found that social PIU (e.g., Grifths, 2000; Young, 1998; Young & Rogers, 1998). In
anxiety was a signicant predictor of POSI. Taken together, these one review, Shapira et al. (2003) concluded that, based on the cur-
results suggest that POSI is an important component of GPIU that rent limited empirical evidence, problematic Internet use may best
may help explain why certain individuals exhibit other indicators be classied as an impulse control disorder (p. 207). More recent
of problematic use such as going online for mood regulation and studies have produced empirical evidence supporting the claim
having decient self-regulation. that compulsive use is a central component of PIU (Caplan, 2005;
Kim et al., 2009; Meerkerk et al., 2009; van den Eijnden et al.,
1.3. Mood regulation 2008; van der Aa et al., 2009). Kim and Davis (2009) argue there
is no serious question that a behavior such as Internet use is prob-
Another cognitive symptom of GPIU, suggested by the extant lematic if the activities become compulsive, if they interfere with
literature, is a motivation to use the Internet for mood regulation normal activities of daily living, and if the person can no longer
S.E. Caplan / Computers in Human Behavior 26 (2010) 10891097 1091
control his/her own use (p. 490). In all, the extant literature sug- mechanism to relieve stress, loneliness, depression, or anxiety
gests that decient self-regulation, in the forms of both cognitive (p. 231). Moreover, these researchers contend that using the
preoccupation and compulsive behavior, plays an important role Internet to alleviate dysphoric moods fosters self-reactive out-
in GPIU. come expectations that Internet use will relieve ones negative
Now that the key constructs have been dened, the next section feelings. In other words, as LaRose et al. (2001) explain expecta-
presents the hypothesized conceptual model suggested by the ex- tions about the positive outcomes of Internet use. . .should in-
tant literature and tested in the current study. Fig. 1 summarizes crease Internet usage (p. 398). Such thinking is consistent with
the conceptual model and the following paragraphs elaborate on Banduras (1991) argument that anticipative affective reactions
the model and present the direct- and indirect-effect hypotheses to ones own behavior play a central role in the self-reactive pro-
tested in the current study. cess that governs self-regulation (p. 256). Thus, the next hypoth-
esis advanced in Fig. 1 is that using the Internet for mood
1.5. Direct effect hypotheses regulation predicts decient self-regulation (i.e., compulsive use
and cognitive preoccupation).
The model advanced here argues that, when individuals prefer
online interaction, they are more likely to use computer-mediated H1: Using the Internet for mood regulation is a direct positive
communication to regulate their mood. Indeed, one reason why predictor of decient self-regulation of Internet use.
some people might prefer online social interaction is because it
helps alleviate the stresses associated with face-to-face meetings. Davis (2001) cognitive-behavioral theory proposes that the cog-
For example, people with high POSI may seek to mitigate the stress nitive and behavioral processes described thus far work together to
they experience in traditional FtF situations with online social foster negative outcomes associated with ones Internet usage.
interaction because, online, features of interpersonal communica- Although mood regulation and increased feelings of social compe-
tion differ in ways that give a person greater self-presentational tence may be very helpful for some individuals, those whose Internet
control and condence (Caplan, 2005, 2007). Additionally, individ- use becomes compulsive and obsessive are likely to experience neg-
uals who have a high POSI may be especially likely to seek out ative outcomes in other domains (Caplan, 2005; Kim et al., 2009).
computer-mediated sources of social support to alleviate affective
distress. In other words, POSI may lead an individual to use the H1: Decient self-regulation is a direct positive predictor of neg-
Internet, rather than traditional FtF behavior when they seek com- ative outcomes arising from ones Internet use.
forting and companionship from members of their support net-
work. The current study proposed and tested the following 1.6. Indirect-effect hypotheses
hypothesis:
In addition to testing direct effects, researchers also need to
H1: Preference for online social interaction is a direct positive empirically assess mediated or indirect effects (see Holbert & Ste-
predictor of using the Internet for mood regulation. phenson, 2002, 2003; McLeod, Kosicki, & MeLeod, 2002). When
studies only test direct effects, they run the risk of under repre-
In addition to predicting mood regulation, the model in Fig. 1 senting total effects. Indeed, as Raykov and Marcoulides (2000) ex-
proposes that POSI also predicts decient self-regulation. To date, plain, if an indirect effect does not receive proper attention, the
studies have partially tested this hypothesis Caplan (2005) and relationship between two variables of interest may not be fully
Kim et al. (2009) both found that a POSI predicted individuals considered (p. 7). Consequently, the current study hypothesized
level of compulsive Internet use (one indicator of decient self- and systematically tested the following mediated relationships
regulation). In light of the previous discussion which described (illustrated in the model in Fig. 1):
compulsive use along with cognitive preoccupation as indicators
of decient self-regulation, the following hypothesis was H1: There is a positive indirect relationship between POSI and
advanced: decient self-regulation that is mediated by using the Inter-
net for mood regulation.
H1: Preference for online social interaction is a direct positive
H2: There is a positive indirect relationship between POSI and
predictor of decient self-regulation of Internet use.
negative outcomes that is mediated by decient self-
regulation.
LaRose et al. (2003) argue that using the Internet for mood
regulation can lead to decient self-regulation (also see Lee & H3: There is a positive indirect relationship between mood reg-
Perry, 2004). They note that the transition to problematic usage ulation and negative outcomes that is mediated by decient
can begin if the behavior acts as an important or exclusive self-regulation.
+
Deficient
POSI Self Regulation
+
+ +
Negative
Outcomes
Mood
Regulation
The next section describes a study that sought to test the tied in the previous section. The GPIUS2 is a revised and updated
hypotheses proposed above. version of the Generalized Problematic Internet Use Scale (GPIUS;
Caplan, 2002).
Originally, the GPIUS was an exploratory instrument designed
2. Method to identify both cognitive and behavioral dimensions of PIU, along
with their negative outcomes. Caplan (2002) conducted an explor-
2.1. Participants atory principal-axis factor analysis of the original GPIUS items and
found it tapped seven dimensions: mood alteration (an individual
Participants were recruited in two ways. First, 424 undergradu- using the Internet in order to facilitate some change in negative
ate students from a variety of different majors enrolled in a large affective states), social benets (the perceived social benets of
introductory communication course participated in the study for Internet use), social control (an individuals perceived degree of
extra credit. Next, to obtain a more representative sample (i.e., old- control over self-presentation when interacting with others on-
er and non-student), students in a different large undergraduate line), withdrawal (a preoccupation with the Internet), compulsive
class recruited 361 non-student participants who were at least use (an inability to control or regulate ones online behavior),
30 years old. In all, 785 people (543 men and 242 women) ranging excessive time online (the degree to which an individual felt that
in age from 18 to 70 (M = 33.14 years old, Median = 22, SD = 15.25) he or she spends an excessive amount of time online), and nega-
participated in the study. tive outcomes (the extent to which an individual experienced per-
sonal, social, and professional problems resulting from Internet
2.2. Measures use). Although the original GPIUS measure was exploratory, a
number of the dimensions signicantly predicted negative
2.2.1. The GPIUS2 outcomes.
The current study employed the Generalized Problematic Inter- Since its publication in 2002, subsequent research has yielded
net Use Scale 2 (GPIUS2) to operationalize the key variables iden- additional information that may be helpful in improving the GPIUS.
POSI1 .83
.80
POSI2 POSI
.71
POSI3
.66
Neg. Out 1 .79
.52 .69
.75 Negative
Neg. Out 2
.82 Outcomes
Neg. Out 3
.77
.47
Mood Reg.1 .71
Compulsive 1 .83
.90 Compulsive
Compulsive 2
Use
.77 .88
Compulsive 3 Deficient
Self-Regulation
Cog. Preoc. 1 .91
.82
.50 Cognitive
Cog. Preoc. 2 Preoccupation
into a single subscale operationalizing POSI. Additionally, as noted Subscale (Cronbachs a) Item wording M (SD)
earlier, LaRose and colleagues work suggests that compulsive Preference for online 1. I prefer online social interaction 1.88 (1.52)
Internet use and cognitive preoccupation are both symptoms of social interaction over face-to-face communication
decient self-regulation (LaRose, 2001; LaRose et al., 2001, 2003). (POSI) (a = .82) 2. Online social interaction is more 2.17 (1.69)
At this point, researchers need to update and rene the GPIUS mea- comfortable for me than face-to-face
interaction
sure, conrm the new factor structure, assess its construct validity, 3. I prefer communicating with 2.21 (1.73)
and conduct a formal assessment of the structural relationships people online rather than face-to-
hypothesized earlier. The study reported here sought to accom- face
plish all three goals. Mood regulation 1. I have used the Internet to talk 2.91 (2.22)
The GPIUS2 includes two new factors, preference for online so- (a = .86) with others when I was feeling
cial interaction (POSI) and decient self-regulation. In the original isolated
2. I have used the Internet to make 2.77 (2.10)
GPIUS, social benets and social control were separate factors; new
myself feel better when I was down
GPIUS2 items were written to operationalize POSI as a single con- 3. I have used the Internet to make 2.79 (2.12)
struct, consistent with Caplans (2003) conceptualization. Another myself feel better when Ive felt
important change is that the GPIUS2 operationalizes decient self- upset
regulation as a higher-order factor that inuences both cognitive Cognitive 1. When I havent been online for 1.97 (1.67)
preoccupation and compulsive Internet use subscale scores. Fig. 2 preoccupation some time, I become preoccupied
illustrates the proposed hierarchical factor structure in the mea- (a = .86) with the thought of going online
2. I would feel lost if I was unable to 2.97 (2.15)
surement model. go online
In addition, the original scales excessive Internet use subscale 3. I think obsessively about going 1.53 (1.21)
was omitted from the new scale. Frequency of Internet use, in online when I am ofine
and of itself, is not necessarily indicative of problematic use (Ca- Compulsive internet use 1. I have difculty controlling the 2.75 (2.13)
plan, 2005). Although people may consider the amount of time (a = .87) amount of time I spend online
they spend online to be excessive, their excessive use may have 2. I nd it difcult to control my 2.27 (2.13)
Internet use
more to do with how they rely on the Internet to do their work
3. When ofine, I have a hard time 2.25 (1.85)
(i.e., functional rather than dysfunctional purposes) than with trying to resist the urge to go online
their psychosocial well-being. People rely on the Internet to man-
Negative outcomes 1. My internet use has made it 1.48 (1.15)
age a number of aspects of their lives (e.g., nances, travel plans, (a = .83) difcult for me to manage my life
research, etc.) and, consequently, they may nd themselves 2. I have missed social engagements 1.44 (1.79)
spending excessive time online in order to attain positive out- or activities because of my Internet
comes at work and at home. Indeed, many people would probably use
3. My Internet use has created 1.56 (1.26)
report that they use the Internet excessively, but out of practical
problems for me in my life
necessity for doing their work. Such uses are more indicative of
*
functional use rather than of an impulse control disorder. On When all items above are used together to create an overall GPIUS2 composite
score, the scales reliability estimate a = .91.
the other hand, compulsive use, which was retained as a subscale
in the new measure, clearly reects a problem with impulse
control.
US2 index score may be computed by summing the 15 items,
Finally, the names of two of the original constructs were revised
which results in possible scores ranging from 15 to 120 (in the cur-
to improve their conceptual clarity in the GPIUS2. Specically, the
rent sample, M = 33.00, SD = 17.67). Or, a composite score may be
mood alteration factor in the original measure was renamed mood
created by averaging a participants scores on all 15 items, which
regulation to emphasize the motivation to use the Internet to alle-
results in possible scores ranging from 1 to 8 (in the current sam-
viate distressing feelings. Also, the withdrawal construct was re-
ple, M = 2.20, SD = 1.18). The next section describes the scales
named cognitive preoccupation to more clearly reect its
validity.
emphasis on obsessive thinking about the online world. The
remaining factors, negative outcomes and compulsive use retained
their names. 2.3. Employing the two-step approach: analyses and results
In sum, the GPIUS2 was written to tap four constructs: (1) POSI,
(2) mood regulation, (3) decient self-regulation (consisting of a As noted earlier, Anderson and Gerbing (1988) suggest a two-
compulsive use subscale and a cognitive preoccupation subscale), step method for using structural equation modeling (SEM) to test
and (4) negative outcomes. Table 1 presents the item wording for and develop theories. Specically, they contend that the model-
the GPIUS2 and identies which items belong to the ve sub- building task can be thought of as the analysis of two conceptually
scales: POSI, mood alteration, cognitive preoccupation, compulsive distinct models. . .a conrmatory measurement, or factor analysis,
use, and negative outcomes. In the current study, participants were model species the relations of the observed measures to their
instructed to rate the extent to which they agreed with each of the posited underlying constructs, with the constructs allowed to
15 GIPUS2 items on a scale ranging from 1 (denitely disagree) intercorrelate freely. A conrmatory structural model then speci-
to 8 (denitely agree). es the causal relations of the constructs to one another as posited
Table 2 presents the zero-order Pearson correlation coefcients by some theory (p. 411).
among all items in the GPIUS2. The GPIUS2 scale can be used in Anderson and Gerbing (1988) note that conrmation of the
two different ways, as a set of separate sub-scales or as an overall measurement model assesses construct validity in general, (Ben-
composite index of GPIUS. Among the current sample, the individ- tler, 1978), and convergent and discriminant validity in particular
ual GPIUS2 sub-scales demonstrated good reliability (Table 1 pre- (Campbell & Fiske, 1959). Next, assuming that the measurement
sents each scales item descriptive statistics and Cronbachs a model ts, the second step tests the structural model and can for-
coefcients). The overall reliability of the composite GPIUS2 mally evaluate nomological validity (Campbell, 1960; Chronbach &
(where all 15 scale items are included) was a = .91. An overall GPI- Meehl, 1955).
1094 S.E. Caplan / Computers in Human Behavior 26 (2010) 10891097
Table 2
Zero-order Pearson correlation coefcients among GPIUS2 items.
Item 1 2 3 4 5 6 7 8 9 10 11 12 13 14
(1) Compulsive use 1
(2) Compulsive use 2 .77*
(3) Compulsive use 3 .61* .67*
(4) Mood regulation 1 .48* .44* .45*
(5) Mood regulation 2 .48* .47* .42*
(6) Mood regulation 3 .54* .54* .49* .66*
(7) Negative outcomes 1 .39* .51* .44* .58* .81*
(8) Negative outcomes 2 .35* .43* .37* .29* .28* .33*
(9) Negative outcomes 3 .42* .48* .39* .31* .28* .31* .57*
(10) POSI 1 .36* .40* .42* .34* .35* .38* .66* .62* .46*
(11) POSI 2 .37* .42* .40* .39* .32* .36* .42* .46* .42* .66*
(12) POSI 3 .30* .35* .37* .46* .40* .40* .39* .44* .33* .61* .56*
(13) Cognitive preoccupation 1 .48* .56* .60* .34* .28* .31* .37* .37* .47* .47* .44* .41*
(14) Cognitive preoccupation 2 .30* .35* .43* .42* .39* .42* .47* .44* .24* .25* .30* .28* .45*
(15) Cognitive preoccupation 3 .47* .56* .63* .31* .32* .30* .26* .22* .47* .45* .39* .39* .65* .35*
*
p < .001.
2.4. Step 1: Testing the measurement model with a higher-order well (t statistics appear in Fig. 3). In addition, the analysis sup-
conrmatory factor analysis ported all seven hypotheses.
Cognitive Compulsive
Preoccupation Use
.92**
.86**
.52**
Deficient
POSI Self Regulation
R2= .65
.78**
.52** .40** Negative
Outcomes
R2=.61
Mood
Regulation
R2= .27
decient self-regulation scores, and 61% of variance in the negative in Table 2 are consistently higher within dimensions than across
outcome scores. dimensions). The results from the higher-order CFA support the
proposed second-order factor intended to tap decient self-regula-
3. Discussion tion. In all, the GPIUS2 produced results that were more than suf-
cient for proceeding with testing the conceptual model in Step 2.
The overall purpose of the study reported here was update the
cognitive-behavioral model of GPIU by integrating ideas and re- 3.2. Step 2: Structural model
sults from more recent research. The study employed Anderson
and Gerbings (1988) two-step approach to theory development Results from the SEM analysis provided support for the overall
and validation. As such, the study had two specic goals. First conceptual model presented in Fig. 1. Broadly, the model combined
the study validated a measurement instrument, the GPIUS2, de- Davis cognitive-behavioral model of GPIUS, Caplans work on POSI,
signed to operationalize key constructs necessary for testing the and LaRoses work on decient self-regulation of Internet use. In
conceptual model. With the measurement model in place, the sec- addition to supporting the overall structural model, the analysis
ond goal of the current study was to empirically test the hypothe- also yielded support for each of the hypotheses advanced earlier.
sized conceptual model. The results presented above demonstrate The rst four hypotheses offered predictions of direct effects. Con-
that both goals were accomplished. Moreover, as the R2 values re- sistent with H1, POSI was a direct positive predictor of the extent
ported in Fig. 3 suggest, the structural model explained a substan- to which individuals used the Internet to regulate their moods. In
tial proportion of variance in decient regulation and negative line with H2, POSI was also a direct positive predictor of decient
outcome scores. Such ndings are particularly encouraging given self-regulation of Internet use. Next, supporting H3, using the
that previous studies have yielded much smaller effects; Caplan Internet for mood regulation was a direct positive predictor of de-
(2003) accounted for 30% of negative outcome scores and Caplan cient self-regulation. With regard to H4, as predicted, decient
(2005) accounted for 10%. The following sections summarize the self-regulation was a direct positive predictor of negative out-
ndings in Steps 1 and 2. comes associated with ones Internet use.
Additionally, the results supported all three indirect-effect
3.1. Step 1: Measurement model hypotheses. As predicted by H5, mood regulation was a signicant
mediator of the association between POSI and decient self-regula-
The GPIUS2 represents an improvement over the earlier GPIUS tion. Consistent with H6, decient self-regulation was a signicant
measure. First, the GPIUS2 was designed from more advanced liter- mediator of the association between POSI and negative outcomes.
ature than was available at the time the GPIUS was written. For in- Finally, as H7 predicted, decient self-regulation was a signicant
stance, the new measure includes a subscale that operationalizes mediator of the association between mood regulation and negative
POSI and also includes two rst-order sub-scales that, together, outcomes.
constitute a second-order factor, decient self-regulation. Thus, Overall, these results suggest that POSI is a key cognitive com-
the new measure is more consistent with research published since ponent of GPIUS that predicts how individuals use the Internet,
the exploratory GPIUS measure was introduced. Results from the along with the outcomes of such use. As such, GPIUS clearly repre-
current study indicate that the GPIUS2 sub-scales are reliable sents a phenomenon worthy of scholarly attention by communica-
and that the scale has good construct validity. The results reported tion researchers, particularly those who study interpersonal
above also suggest the scale has and good discriminant and con- phenomena. Previous studies have shown that loneliness and so-
vergent validity among its sub-scales (i.e., correlations presented cial skills predict individuals degree of POSI. Thus, POSI represents
1096 S.E. Caplan / Computers in Human Behavior 26 (2010) 10891097
an important link in the association between psychosocial well- different interpersonal functions of the Internet. Future research
being and decient self-regulation of Internet use. might improve the model and the measure by considering if differ-
Additionally, the results lend further support to the argument ent types of interpersonal Internet use (i.e., instant messaging, chat
that decient self-regulation and mood regulation are central as- rooms, email) are more or less strongly associated with POSI, mood
pects of GPIUS. Mood regulation explains how POSI might be re- regulation, decient self-regulation and negative outcomes.
lated to decient self-regulation. Decient self-regulation helps The current study represents a step toward a better understand-
explain how and why Internet use can result in negative outcomes. ing of GPIU, conceptually, and better tools to study it with. Based
Together, POSI, mood regulation, and decient self-regulation play on the results of the conrmatory analysis conducted in the cur-
important roles in GIPU. Taken together, these variables accounted rent study, the GPIUS2 appears to be a reliable and valid measure
for a substantial proportion of variance in negative outcome scores. of GPIU cognitions, behaviors, and outcomes. Moreover, the test of
the conceptual model offers a clearer picture of the associations
3.3. Limitations of the current study among the variables that constitute GPIUS.
LaRose, R., Eastin, M. S., Gregg, J. (2001). Reformulating the Internet paradox: Social Morahan-Martin, J. (2008). Internet abuse: Emerging trends and lingering
cognitive explanations of Internet use and depression. Journal of Online Behavior, questions. In A. Barak (Ed.), Psychological aspects of Cyberspace. Theory,
1(2). <http://www.behavior.net/JOB/v1n2/paradox.html> Retrieved 06.05.05. research and applications (pp. 3269). Cambridge, UK: Cambridge University
LaRose, R., Lin, C. A., & Eastin, M. S. (2003). Unregulated Internet usage: Addiction, Press.
habit, or decient self-regulation? Media Psychology, 5, 225253. Pratarelli, M. E., & Browne, B. L. (2002). Conrmatory factor analysis of Internet use
LaRose, R., Mastro, D., & Eastin, M. S. (2001). Understanding Internet usage: A social- and addiction. CyberPsychology & Behavior, 5(1), 5364.
cognitive approach to uses and gratications. Social Science Computer Review, Raykov, T., & Marcoulides, G. A. (2000). A method for comparing completely
19, 395413. standardized solutions in multiple groups. Structural Equation Modeling, 7,
Lee, K. C., & Perry, S. D. (2004). Student instant message use in a ubiquitous 292308.
computing environment: Effects of decient self-regulation. Journal of Shapira, N. A., Lessig, M. C., Goldsmith, T. D., Szabo, S. T., Lazoritz, M., Gold, M. S.,
Broadcasting & Electronic Media, 48(3), 399420. et al. (2003). Problematic internet use: Proposed classication and diagnostic
MacKinnon, D. P., Lockwood, C., & Hoffman, J. (1998). A new method to test for criteria. Depression and Anxiety, 17(4), 207216.
mediation. Paper presented at the annual meeting of the Society for Prevention Shapiro, M. A. (2002). Generalizability in communication research. Human
Research, Park City, UT. Communication Research. Special Issue: Statistical and Methodological Issues in
MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002). A Communication Research, 28(4), 491500.
comparison of methods to test mediation and other intervening variable effects. van den Eijnden, R. J. J. M., Meerkerk, G., Vermulst, A. A., Spijkerman, R., & Engels, R.
Psychological Methods, 7(1), 83104. C. M. E. (2008). Online communication, compulsive internet use, and
McLeod, D. M., Kosicki, G. M., & MeLeod, J. M. (2002). Resurveying the boundaries of psychosocial well-being among adolescents: A longitudinal study.
political communication effects. In J. Bryant & D. Zillmann (Eds.), Media effects: Developmental Psychology, 44, 655665.
Advances in theory and research (2nd ed., pp. 215267). Mahwah, NJ, US: van der Aa, N., Overbeek, G., Engels, R. C. M. E., Scholte, R. H. J., Meerkerk, G., Van den
Lawrence Erlbaum Associates. Eijnden, R. J. J. M., et al. (2009). Daily and compulsive internet use and well-
Meerkerk, G., Van Den Eijnden, R. J. J. M., Vermulst, A. A., & Garretsen, H. F. L. (2009). being in adolescence. A diathesis-stress model based on big ve personality
The compulsive internet use scale (CIUS): Some psychometric properties. traits. Journal of Youth and Adolescence, 38(6), 765776.
CyberPsychology & Behavior, 12, 16. Young, K. S. (1998). Caught in the net: How to recognize the signs of Internet addiction
Morahan-Martin, J. (2007). Internet use and abuse and psychological problems. In J. and a winning strategy for recovery. New York: John Wiley & Sons.
Joinson, K. McKenna, T. Postmes, & U. Reips (Eds.), Oxford handbook of internet Young, K. S., & Rogers, R. C. (1998). The relationship between depression and
psychology (pp. 331345). Oxford, UK: Oxford University Press. Internet addiction. CyberPsychology & Behavior, 1, 2528.