Computers in Human Behavior 25 (2009) 749–760

Contents lists available at ScienceDirect

Computers in Human Behavior

journal homepage: www.elsevier.com/locate/comphumbeh

A randomised controlled trial of a self-guided internet intervention promoting

well-being
Joanna Mitchell a,*, Rosanna Stanimirovic b, Britt Klein c, Dianne Vella-Brodrick a
a
School of Psychology, Psychiatry and Psychological Medicine, Monash University, P.O. Box 197, Caulfield East, Vic. 3145, Australia
b
Department of Performance Psychology, Australian Institute of Sport, Bruce, ACT, Australia
c
National eTherapy Centre for Anxiety Disorders, Faculty of Life and Social Sciences, Swinburne University, Hawthorn, Vic., Australia

a r t i c l e i n f o a b s t r a c t

Article history: Positive psychology is paving the way for interventions that enduringly enhance well-being and the
Available online 16 March 2009 internet offers the potential to disseminate these interventions to a broad audience in an accessible
and sustainable manner. There is now sufficient evidence demonstrating the efficacy of internet interven-
Keywords: tions for mental illness treatment and prevention, but little is known about enhancing well-being. The
Subjective well-being current study examined the efficacy of a positive psychology internet-based intervention by adopting
Internet a randomised controlled trial design to compare a strengths intervention, a problem solving intervention
Positive psychology
and a placebo control. Participants (n = 160) completed measures of well-being (PWI-A, SWLS, PANAS,
Strengths
Cognitive-behavioural therapy
OTH) and mental illness (DASS-21) at pre-assessment, post-assessment and 3-month follow-up. Well-
Happiness being increased for the strengths group at post- and follow-up assessment on the PWI-A, but not the
Health promotion SWLS or PANAS. Significant changes were detected on the OTH subscales of engagement and pleasure.
No changes in mental illness were detected by group or time. Attrition from the study was 83% at 3-
month follow-up, with significant group differences in adherence to the intervention: strengths (34%),
problem solving (15.5%) and placebo control (42.6%). Although the results are mixed, it appears possible
to enhance the cognitive component of well-being via a self-guided internet intervention.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction complete picture of health is reflected in the World Health Organ-

isations definition of mental health as:
Enhancing well-being at a population level is explored in this
. . .a state of well-being in which the individual realizes his or
introduction in the context of two relatively young disciplines,
her own abilities, can cope with the normal stresses of life,
namely positive psychology and internet interventions. An over-
can work productively and fruitfully, and is able to make a con-
view of theory and research in positive psychology and then inter-
tribution to his or her own community. (WHO, 2001a, p. 1).
net interventions is presented as a rationale for the current study.
This definition encapsulates the idea that mental health is the
1.1. Positive psychology, mental health and well-being presence of well-being and not just the absence of mental illness.
To test a model of complete mental health and psychosocial func-
The positive psychology movement has helped create the re- tioning Keyes (2005) surveyed a nationally representative sample
search momentum necessary to broaden mental health knowledge of 3032 American adults. The results supported the theory that
and understanding beyond a focus on illness and its direct allevia- mental health and mental illness are independent but correlated
tion. Positive psychology is the scientific study of well-being and axes; and not merely opposite ends of a continuum. Moreover,
optimal functioning, focusing on positive emotions, character traits Keyes found that participants with no mental illness but low
and enabling institutions (Seligman & Csikszentmihalyi, 2000). The well-being (Keyes labels this languishing) had equivalently poor
proponents of this movement aim to bring together and develop psychosocial outcomes as the participants with a mental illness.
previously disparate lines of theory and research to provide a com- Consequently, promoting well-being and optimal psychosocial
plete picture of mental health (Duckworth, Steen, & Seligman, functioning is important in its own right, and not just an adjunct
2005; Seligman, Steen, Park, & Peterson, 2005). The notion of a to mental illness treatment and prevention.
Well-being, also referred to by some researchers as happiness
(these terms will be used interchangeably), is a complex construct
* Corresponding author. Tel.: +61 3 9214 5867; fax: +61 3 9214 5260. concerned with optimal experience and functioning (Ryan & Deci,
E-mail address: psychjm@gmail.com (J. Mitchell). 2001). There are two major conceptual approaches to defining and

0747-5632/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.chb.2009.02.003
750 J. Mitchell et al. / Computers in Human Behavior 25 (2009) 749–760

measuring well-being: eudaimonic and hedonic. Aristotle (384–322 84% of Australians, and 60% of Australian households (9.1 million
BC) first articulated the eudaimonic approach as being true to one’s people), have access to the internet (ABS, 2006; DCITA, 2005). These
inner self. In contemporary psychology this approach is best re- household access rates are similar to those reported for the United
flected by the concept of psychological well-being (PWB), which Kingdom (60.2%) and United States (62%) (ABS, 2006; Cheeseman
is broadly defined as the degree to which a person is fully function- Day, Janus, & Davis, 2005). People use the internet for a variety
ing and focuses on meaning and personal growth (Ryan & Deci, of purposes and there is a growing interest in wellness information
2001; Ryff, 1989). In contrast, the hedonic approach focuses on unrelated to symptoms of illness, a medical diagnosis or other
pleasure attainment and pain avoidance (Ryan & Deci, 2001) and health crisis (Evers, 2006; Fox, 2006). The internet has been
in contemporary psychology subjective well-being (SWB) best acknowledged by consumers, researchers, policy makers and clini-
encapsulates this approach. SWB is defined as how an individual cians as a valuable means of health promotion (Christensen, Grif-
evaluates his/her own life (Diener, 1984) and incorporates both fiths, & Evans, 2002; Evers, 2006; Korp, 2006).
affective (e.g., positive and negative moods or emotions) and cog- Obtaining health information via the web has taken a variety of
nitive (e.g., satisfaction judgements) components. There has been forms including static health educational sites, peer support
debate over the utility of the eudaimonic/hedonic divide and more groups, online health consultations and delivery of internet inter-
recently it has been proposed that these models are not mutually ventions. Ritterband et al. (2003) defined internet interventions
exclusive and can independently and in combination provide valu- for mental health as interventions that promote knowledge and
able insight about well-being measurement and underlying mech- behaviour change via web-based programs that are typically the-
anisms (Kashdan, Biswar-Diener, & King, 2008; Keyes, Shmotkin, & ory driven, self-paced, interactive, tailored to the user and utilise
Ryff, 2002; Ryan & Deci, 2001). Subsequently, some well-being the- the multimedia opportunities provided by the internet. These
orists have combined both SWB and PWB into unifying models of intervention websites are generally based on effective face-to-face
well-being, for example, the complete state model of mental health interventions that have been operationalised and transformed for
(Keyes, 2005; Keyes, 2007) and the orientations to happiness (Pet- internet delivery, for example, Panic Online – a treatment program
erson, Park, & Seligman, 2005). for panic disorder (Klein & Richards, 2001; Klein, Richards, & Aus-
While there is ample literature to suggest the pursuit of happi- tin, 2006).
ness is a worthwhile one (for a review see Lyubomirsky, King, & The number of internet interventions available for mental
Diener, 2005), there is less literature focussed on whether it can health treatment and prevention is growing rapidly, as are inter-
be sustained or enhanced at a population level. One model of ventions that promote health behaviour change (see Table 1).
enduring, or chronic, happiness proposes three key factors that These interventions have demonstrated efficacy (e.g., reduction
influence well-being: (1) a person’s genetically determined set in symptoms or number of people meeting clinical criteria for diag-
point, or set range, for happiness; (2) circumstantial factors (e.g., nosis of a disorder, for a range of mental health disorders) and the
income, location, education level and marital status) and; (3) majority are based on cognitive-behavioural approaches (Christen-
intentional cognitive, motivational, and behavioural activities that sen, Griffiths, Korten, Brittliffe, & Groves, 2004; Klein et al., 2006, in
can influence well-being (Lyubomirsky, Sheldon, & Schkade, 2005). pressb).
It is proposed that this last factor, with its focus on individual psy- In contrast to the growing internet-based treatment and pre-
chological processes, is most amenable to change. For example, vention literature, only one published randomised controlled trial
data from longitudinal studies have demonstrated that well-being was identified that focussed on well-being enhancement via the
can be enhanced via interventions that promote intentional activ- internet (Seligman et al., 2005). Seligman et al. (2005) used the
ity, such as practising gratitude, committing acts of kindness, visu- internet for participant recruitment, data collection and interven-
alizing best possible future selves, and processing positive life tion delivery. Five hundred and fifty-seven participants completed
experiences (Lyubomirsky, 2006; Lyubomirsky, Sheldon et al., the pre-assessment questionnaires with 166 participants (29%)
2005). The current study set out to determine whether well-being dropping out before the final 6-month assessment. Participants
could be enhanced by intentional activity and to extend previous were randomly assigned to one of six groups including five active
research by examining whether this type of intervention can be interventions and one placebo control. The five proposed happi-
delivered using the internet. ness interventions included: (1) a gratitude visit; (2) identifying
three good things in life; (3) identifying a time when you are at
1.2. The internet and mental health promotion your best; (4) identifying signature strengths; and (5) identifying
and using signature strengths in a new way. The placebo control
A key objective of mental health promotion is to deliver interven- involved writing about earliest memories. Participants completed
tions that have demonstrated efficacy and are accessible and sustain- a demographic survey and two questionnaires measuring depres-
able. Traditional forms of delivery such as mass media campaigns, or sion (Centre for Epidemiological Studies – Depression Scale) and
individual or group interventions that are offered through schools or happiness (Steen Happiness Index) that were repeated on six occa-
the work place, may demonstrate efficacy but are not always accessi- sions (pre-, post-assessment, 1-week, 1-, 3-, and 6-month follow-
ble (e.g., to rural communities or small businesses) or sustainable up); with reminder emails to complete the questionnaires sent at
(e.g., are costly to deliver). Mass media campaigns tend to address each time point. The 1-week intervention involved participants
only the most general determinants of a particular health issue or receiving instructions for their assigned activity via an email. Par-
behaviour (e.g., an Australian campaign run by VicHealth called ‘To- ticipants were encouraged to contact the researchers if they had
gether We Do Better’, which seeks to increase community awareness any questions about the activity. Adherence to the activity was
of the benefits of strong, connected and supportive communities), yet measured by a question requiring a ‘yes’ or ‘no’ response.
we are told that behaviour change is more likely if interventions are Using signature strengths in a new way and three good things pro-
targeted at the individual (de Vries & Brug, 1999). The internet has duced significant change in the expected direction on the happi-
the potential to address these issues of efficacy, accessibility, sustain- ness and depression outcome measures, with benefits apparent
ability and delivery at an individual level, therefore providing an at 6 months. The gratitude intervention was also effective in
adjunctive health promotion delivery framework (de Vries & Brug, improving happiness and depression ratings, however this change
1999; Evers, 2006; Mihalopoulos et al., 2005). lasted only 1 month. In addition, participants who reported contin-
Over the past 20 years the internet has become an integral part ued adherence to the happiness intervention beyond the required
of the lives of most Australians. A national survey indicated that 1 week, scored higher on happiness scores at all times points and
 J. Mitchell et al. / Computers in Human Behavior 25 (2009) 749–760 751

Table 1
Internet intervention controlled studies for the treatment and/or prevention of mental illness and health behaviour change.

Mental illness/health behaviour Examples of controlled studies

Alcohol use Kypri et al. (2004), Walters, Miller, and Chiauzzi (2005)
Anxiety Kenardy, McCafferty, and Rosa (2003), Proudfoot et al. (2003)
Bipolar disorder Proudfoot, Parker, Benoit, Manicavasagar, and Smith (2007), Proudfoot et al. (2007)
Depression Andersson et al. (2005), Christensen, Griffiths, and Jorm (2004), Christensen et al. (2002), Clarke et al. (2002), Patten (2003), Proudfoot
et al. (2004), Proudfoot, Ryden, and Goldberg (2005)
Eating and body image Celio et al. (2000), Winzelberg et al. (2000)
Encopresis Ritterband, Cox, et al. (2003), Ritterband et al. (2006)
Headaches Devineni and Blanchard (2005), Strom, Pettersson, and Andersson (2000)
Nutrition/dietary behaviour/ McKay, Glasgow, Feil, Boles, and Barrera (2002), Tate, Wing, and Winett (2001), Wantland, Portillo, Holzemer, Slaughter, and McGhee
weight loss (2004), Winett et al. (1999)
OCD Clark, Kirkby, Daniels, and Marks (1998)
Panic disorder Carlbring et al. (2006), Carlbring et al. (2005), Carlbring et al. (2001), Kiropoulos et al. (2008), Klein et al. (in pressa), Klein and Richards
(2001), Klein et al. (2006), Pier et al. (2008), Richards, Klein, and Austin (2006)
Phobias/panic Kenwright, Liness, and Marks (2001)
Physical activity Napolitano et al. (2003), Spittaels and De Bourdeaudhuij (2006), Vandelanotte, De Bourdeaudhuij, Sallis, Spittaels, and Brug (2005),
Wantland et al. (2004)
Posttraumatic stress Hirai and Clum (2005), Klein et al. (in pressb), Lange, van de Ven, and Shrieken (2003), Lange, van de Ven, Shrieken, and Emmelkamp
(2001), Litz, Engel, Bryant, and Papa (2007)
Smoking cessation Cobb, Graham, Bock, Papandonatos, and Abrams (2005), Etter (2005), Swartz, Noell, Schroeder, and Ary (2006), Walters, Wright, and
Shegog (2006)
Social phobia Andersson et al. (2006)
Tinnitus Abbott et al. (in press), Andersson, Strömgren, Ström, & Lyttkens (2002)

lower on depression scores at 1-month follow-up, compared to post- and follow-up assessment; (2) the problem solving group
those who did not continue to adhere. would demonstrate a decrease in mental illness at post- and fol-
A limitation of the study, from an internet interventions research low-up assessment and; (3) adherence would be greatest in the
perspective, was that the interventions were unlikely to qualify as a strengths intervention group.
true internet intervention according to the definition provided ear-
lier in this article (Ritterband et al., 2003). Although a website was 2. Methods
used to recruit participants and collect data, the interventions each
consisted of a single email with written instructions and did not uti- 2.1. Design
lise the interactive features of web-based technology.
In addition, the study was limited by a lack of clarity concerning A randomised controlled trial, 3 (group)  3 (time) design was
the amount of human contact that was provided to participants. used. The three groups included a positive psychology strengths
Participants were encouraged to contact the researchers if they intervention, a problem solving intervention and a placebo control
had any questions about the intervention, but no details were pub- group. Participants completed online assessments at pre-, post-,
lished about how much human support was provided, making it and 3-month follow-up, to evaluate the post-intervention out-
unclear whether the program was self-guided or partially sup- comes and durability of change over time.
ported for some or all of the participants. Human supported (e.g.,
via email, phone, face-to-face contact) internet interventions have 2.2. Measures
demonstrated larger effect sizes than pure self-help programs
(Spek et al., 2007). Despite these limitations, the study demon- The following measures were used to collect demographic
strated the potential for delivering mental health promotion inter- information and measure well-being, mental illness, and adher-
ventions to promote well-being via the internet. ence. The relevant Cronbach alpha coefficients for the current
study are reported in Table 2.
1.3. Aim and hypotheses
2.2.1. Personal Well-being Index – Adult (PWI-A) Scale
For the present study a positive psychology intervention, based The PWI-A (IWG, 2006) is a measure of subjective well-being
on the ‘using signature strengths in a new way’ intervention (see consisting of eight items of satisfaction, each one corresponding
Seligman et al., 2005), was developed and delivered via a purposely to a life domain (i.e., standard of living, health, achieving in life,
built, fully automated and interactive website. Seligman’s theory relationships, safety, community-connectedness, future security,
proposes that there are three orientations that promote happiness and spirituality/religion). The PWI-A has satisfactory validity and
(i.e., pleasure, engagement and meaning) and this intervention reliability and correlates .78 with the SWLS (IWG, 2006).
activates the engagement orientation to happiness by helping peo-
ple think about and use their personal strengths in a new way. The 2.2.2. Satisfaction with Life Scale (SWLS)
aim of this study was to test the efficacy of the internet interven- The SWLS (Diener, Emmons, Larsen, & Griffin, 1985) is a five
tion over time and in comparison to a cognitive-behavioural inter- item instrument designed to measure global cognitive judgments
vention (i.e., problem solving), as typically used in the treatment of one’s life. Respondents use a seven-point scale from 1 (strongly
and prevention literature, and a placebo control. It was hypothe- disagree) to 7 (strongly agree) to rate the extent of their agreement
sised that: (1) the strengths group would demonstrate an increase with five statements (e.g., ‘‘I am satisfied with my life”). The major-
in well-being and engagement and decrease in mental illness at ity of people obtain scores in the 23–28 range (slightly satisfied to
752 J. Mitchell et al. / Computers in Human Behavior 25 (2009) 749–760

Table 2
Cronbach alpha coefficients, means and standard deviations on dependant variables by group and time.

Measure Alpha Group Time 1 Time 2 Time 3

Mean SD Mean SD Mean SD
PWI-A .81 Strengths 71.25 15.61 72.65 14.16 73.63 14.45
Problem solving 70.20 12.65 70.22 12.86 70.25 12.72
Placebo control 71.69 13.75 70.08 15.67 69.90 15.95
SWLS .86 Strengths 23.54 6.96 24.21 6.98 24.42 6.95
Problem solving 23.38 6.08 23.50 6.12 23.34 6.07
Placebo control 24.65 5.75 24.96 6.28 25.19 6.09
PANAS Strengths 35.37 6.30 35.29 7.12 35.34 6.79
Positive affect .85 Problem solving 33.55 6.52 33.55 6.58 33.59 6.64
Placebo control 34.41 5.85 34.61 5.96 34.33 6.27
PANAS Strengths 15.52 4.63 15.06 4.45 14.72 4.18
Negative affect .86 Problem solving 15.93 5.71 16.02 5.76 15.95 5.77
Placebo control 16.17 4.66 15.65 4.51 16.13 5.46
DASS Strengths 6.33 5.52 6.19 5.74 5.75 5.29
Depression .77 Problem solving 5.79 4.55 5.52 4.56 5.52 4.62
Placebo control 5.07 5.09 5.28 5.16 5.54 5.52
DASS Strengths 4.00 4.83 3.35 4.09 3.67 4.37
Anxiety .74 Problem solving 3.74 3.55 3.71 3.62 3.57 3.70
Placebo control 4.19 5.10 3.56 4.20 3.89 4.55
DASS Strengths 10.25 7.97 10.48 8.49 10.67 8.18
Stress .81 Problem solving 10.76 6.03 10.83 6.04 11.00 5.85
Placebo control 10.93 5.90 10.74 5.57 11.07 6.03
OTH Strengths 3.32 .80 3.43 .92 3.39 .95
Pleasure .80 Problem solving 3.05 .73 3.05 .68 3.06 .72
Placebo control 2.94 .87 2.94 .87 2.86 .86
OTH Strengths 3.17 .74 3.23 .76 3.28 .72
Engagement .70 Problem solving 2.98 .67 2.99 .63 3.00 .66
Placebo control 2.88 .64 2.98 .63 2.97 .65
OTH Strengths 3.43 .84 3.44 .85 3.48 .87
Meaning .76 Problem solving 3.26 .75 3.28 .77 3.30 .80
Placebo control 3.45 .79 3.48 .82 3.52 .84

Note: Alpha scores in excess of .70 indicate adequate internal consistency (Nunnally, 1978).

satisfied) and the SWLS has demonstrated satisfactory validity and different happiness endorsements (pleasure, engagement, and
reliability (Diener et al., 1985; Pavot & Diener, 1993). meaning). Respondents used a five-point scale from 1 (not like
me at all) to 5 (very much like me) to rate the extent of their iden-
2.2.3. Positive and Negative Affect Schedule (PANAS) tification with each of the statements (e.g., ‘‘I seek out situations
The PANAS (Watson, Clark, & Tellegen, 1988) is a measure of po- that challenge my skills and abilities”). Internal consistencies of
sitive and negative affect, consisting of 10 positive emotions (inter- the three subscales were reported as .82 for pleasure, .72 for
ested, excited, strong, enthusiastic, proud, alert, inspired, engagement and .82 for meaning. Small to moderate correlations
determined, attentive, and active) and 10 negative emotions (dis- with the SWLS for each of the subscales are reported as .17 for
tressed, upset, guilty, scared, hostile, irritable, ashamed, nervous, pleasure, .30 for engagement and .26 for meaning (Peterson
jittery, and afraid). Participants rate items on a scale from 1 (very et al., 2005).
slightly or not at all) to 5 (extremely) based on the strength of emo- To establish the reliability of this scale in the current study, the
tion. The PANAS is commonly used in conjunction with the SWLS Cronbach alpha coefficients are reported (see Table 2) and a
to measure subjective well-being and has demonstrated satisfac- principal component analysis (PCA) was conducted (see Table 3).
tory validity and reliability (Watson et al., 1988). Bartlett’s Test of Sphericity was significant at p < .001, and the
Kaiser–Meyer–Olkin measure of sampling adequacy was .78;
2.2.4. Depression, Anxiety, Stress Scales (DASS-21) exceeding the recommended value of .6 (Tabachnick & Fidel,
The DASS-21 (Lovibond & Lovibond, 1995) is a short form of the 2007) supporting the factorability of the matrix. PCA revealed the
DASS and contains three self-report scales, each with 7-items, de- presence of six components with eigenvalues exceeding 1,
signed to measure the emotional states of anxiety, depression, and explaining 25.1%, 13.9%, 9.0%, 6.5%, 6.0%, and 5.6% of the variance,
stress. Respondents are asked to use a four-point severity/fre- respectively. An inspection of the scree plot revealed a clear break
quency scale from 0 (did not apply to me at all) to 3 (applied to after the third component and it was decided to retain three
me very much, or most of the time) to rate the extent that they components for further analysis.
had experienced each emotion over the last week (e.g., ‘‘I felt sad The three-component solution explained a total of 48.0% of the
and depressed”). The DASS-21 has satisfactory validity and reliabil- variance. Oblimin rotation was performed to aid in the interpreta-
ity (Antony, Bieling, Cox, Enns, & Swinson, 1998; Lovibond & Lovi- tion and indicated weak negative correlations between Compo-
bond, 1995). nents 1 and 2 (r = .14) and Components 1 and 3 (r = .29); and
a weak positive correlation between Components 2 and 3
2.2.5. Orientations to Happiness (OTH) (r = .25). Inspection of the matrix table showed a relatively clear
The OTH (Peterson et al., 2005) is a relatively new 18-item scale three-factor solution, and the interpretation was consistent with
consisting of three subscales (6-items per scale) measuring three previous research on the OTH scale, with meaning items loading
 J. Mitchell et al. / Computers in Human Behavior 25 (2009) 749–760 753

Table 3
Pattern matrix for PCA of three-factor solution of OTH items.

Number OTH items Pattern coefficients

Description Component 1 Component 2 Component 3
11 I have a responsibility to make the world a better place .854
14 What I do matters to society .790
12 My life has a lasting meaning .761
17 I have spent a lot of time thinking about what life means and how I fit into its big picture .614
2 My life serves a higher purpose .545
4 I seek out situations that challenge my skills and abilities .408
5 In choosing what to do, I always take into account whether it will benefit other people .399
18 For me, the good life is the pleasurable life .814
15 I agree with this statement: ‘‘Life is short-eat dessert first” .762
13 In choosing what to do, I always take into account whether it will be pleasurable .727
16 I love to do things that excite my senses .706
3 Life is too short to postpone the pleasures it can provide .599
8 I go out of my way to feel euphoric .433 .415
7 I am always very absorbed in what I do .710
1 Regardless of what I am doing, time passes very quickly .707
6 Whether at work or play, I am usually ‘‘in a zone” and not conscious of myself .616
9 In choosing what to do, I always take into account whether I can lose myself in it .582
10 I am rarely distracted by what is going on around me .565

on Component 1, pleasure items on Component 2 and engagement Participants were given 3 weeks to complete the intervention
items on Component 3. There were two exceptions to this model, and at completion were prompted via the website to answer the
with item 8 (‘I go out of my way to feel euphoric.’) crossloading same five mental health and well-being questionnaires and a pro-
on pleasure (.43) and engagement (.41); and Item 4 (‘I seek out sit- gram evaluation (Time 2). Three months later participants were
uations that challenge my skills and abilities’) loading on meaning sent an email request to login to the website and complete the five
and not on engagement as intended. On inspection of the items mental health and well-being questionnaires for a final time (Time
that make up the OTH-engagement subscale, five items (see Table 3). An unintended outcome of the website design was that only
3, items 1, 6, 7, 9, 10) appear to tap into the engagement experi- participants who completed the whole online intervention were
ence, akin to flow (Csikszentmihalyi, 1990), as a theorised outcome prompted to proceed to the post- and follow-up assessment phases
of using your strengths. In contrast item 4 appears to measure of the study (i.e., post- and follow-up assessment data was not col-
seeking out situations that may create flow rather than the flow lected for non- or partial-intervention completion).
experience itself, and this maybe why this item has loaded on
the meaning factor in this Australian sample. With some noted 2.4. The intervention groups
exceptions these results support the three-factor model of the
OTH proposed by the authors. The three programs were based on established protocols that
were operationalised and transformed for delivery on the internet.
2.2.6. Demographics The two active interventions (strengths and problem solving) were
Demographic data were collected and included 10 questions text and graphics based (no audio, animation or video) and used
about age, gender, income, education, employment, marital status, interactive features to engage the user in an active learning process
number of children, physical health, AIS athlete status, and resi- (e.g., participants were asked to type their responses to questions,
dential postcode, were included in the study. to click and drag objects around the page and provided with feed-
back based on the PWI-A questionnaire). The three programs were
2.2.7. Adherence delivered over three sessions, with a recommended 1-week break
Adherence to the intervention groups was manually recorded as between sessions, and automated weekly email reminders to com-
a dichotomous value (yes/no) depending on whether participants plete the next session.
had completed all three modules of the intervention (i.e., partial The strengths intervention was based on a positive psychology
or non-completers were categorised as ‘no’). intervention that involved identifying and using your strengths
(Seligman et al., 2005). In the first session, participants identified
2.3. Procedure and prioritised their perceived strengths from a list of 24 signature
strengths (Peterson & Park, 2004). At the end of the session they
Ethics approval for the study was provided by the relevant were assigned an offline activity, or homework task, asking them
Monash University and AIS Ethics Committees. Australian adults to share with a friend what they had learnt about identifying per-
were recruited through advertising sent via Monash University sonal strengths. In session two, participants provided feedback on
and the Australian Sports Commission’s online networks (e.g., their progress with the previous session’s offline activity and then
websites, eNewsletters, and email distribution lists). Participants selected three of their top 10 strengths to further develop in their
self-registered online for the study, completed an online informed daily life. Participants were asked to practice using their identified
consent process and were assigned a personal username and pass- strengths during the week and were provided examples and an on-
word for access to the intervention website. When participants line, downloadable diary to help them record their progress. The fi-
first logged-in they were asked to complete a demographic survey nal session reviewed participant progress, summarised the
and five mental health and well-being questionnaires (Time 1). On information provided to date, and directed participants to the
completion of the questionnaires all eligible participants were ran- post-intervention questionnaires. Once the questionnaires were
domly allocated to one of three groups via an automated com- completed, participants could view a graph with their scores on
puter-based random number generator built into the web-based the SWLS at pre- and post-assessment. Three months later, after
program by the web developers (www.janison.com). completing the follow-up assessment questionnaires, participants
754 J. Mitchell et al. / Computers in Human Behavior 25 (2009) 749–760

could view their SWLS graph again with the additional follow-up 3. Results
assessment data included.
The problem solving intervention was based on a cognitive-behav- 3.1. Participants
ioural approach to problem solving and was chosen because it is a life
skill that could be applied across clinical and non-clinical popula- Participants (n = 160) were included in the study if they were
tions. Problem solving is typically included in cognitive-behavioural Australian residents and at least 18 years old. For duty of care rea-
treatment and prevention programs for stress management and sons participants were excluded and referred to support services if
depression. In the first session, participants were introduced to three their DASS subscale scores were in the ‘severe’ range (n = 9), indi-
steps of a six-step approach to problem solving. The six steps are (1) cating the possibility of a mood or anxiety disorder. The participant
identify the problem; (2) generate possible solutions; (3) evaluate attrition rate for the study was 69% at post-assessment and 83% at
the alternatives; (4) decide on a solution; (5) implement the solu- 3-month follow-up. Participant flow through the study from
tion; and (6) evaluate and review progress. At the end of the first ses- recruitment to data analysis is summarised (see Fig. 1).
sion, participants were assigned an offline activity asking them to The mean age of participants was 37 years (range: 18–62;
share what they had learnt about problem solving with a friend/fam- SD = 11.2) and most were female (83%). The majority of partici-
ily member. In the second session, participants were asked to pro- pants were employed (80%) or students (16%); had completed
vide feedback on their progress with the offline activity from the an undergraduate or postgraduate degree (76%); were married
previous session. Next, participants were introduced to steps four or in a defacto relationship (57%); had no children (58%) or 1–2
and five of the problem solving model and to apply this information children (27%); and had a gross yearly income of $40,000 to
to a real life problem. They were asked to practice using their prob- $79,000 (48%) or less than $40,000 (36%). Most participants self-
lem solving skills during the week and were provided an online, rated their physical health as above average (57%) or average
downloadable diary to help them record their progress. In the final (32%) and there was one AIS scholarship holders (i.e., elite level
session, participants were asked to provide feedback on their offline athletes) (<1%).
activity, were introduced to step six of the model, given a summary
of the whole six-step model, and then directed to the post-interven- 3.2. PWI-A
tion questionnaires. As per the strengths intervention, participants
could view a graph of their SWLS scores at three time points. A repeated measures ANOVA showed a significant interaction
The placebo control was an abbreviated version of the problem between intervention group and time on the PWI-A, Wilks’ Lamb-
solving intervention but without utilising any of the interactive da = .93, F(4, 312) = 2.81, p = .02, partial eta squared = .03. The
web features (i.e., it is like reading an electronic book). Unlike strengths group showed an increase in PWI from pre- to post-
the problem solving group, participants were not asked to apply assessment and to 3-month follow-up. The problem solving group
the problem solving information to a real life problem, nor to com- showed no change over time and the placebo control group
plete any offline tasks. showed a decrease in PWI from pre- to post-assessment and then
no change to follow-up. These results are summarised in Fig. 2.
2.5. Statistical procedures and analyses
3.3. SWLS
Statistical analyses were conducted using SPSS version 14 and
10. Normality tests were performed on the data prior to running A repeated measures ANOVA showed no significant interaction
the analysis. To fulfil normality requirements outliers on three between intervention group and time on the SWLS, Wilks’ Lamb-
subscales (DASS subscales of depression and anxiety; and PANAS da = .97, F(4, 312) = 1.27, p = .28, partial eta squared = .02. The main
subscale of negative affect) had their raw scores truncated to be effect for time was not significant, Wilks; Lambda = .96,
one unit larger than the next most extreme score in the distribu- F(2, 156) = 2.90, p = .058, partial eta squared = .04. The main effect
tion (Tabachnick & Fidel, 2007). To confirm random assignment to for group was not significant, F(2, 157) = .84, p = .43, partial eta
the three conditions, one-way ANOVAs and Chi-square tests were squared = .01.
conducted on all pre-treatment measures and no significant dif-
ferences were found. Data analysis involved intention-to-treat 3.4. PANAS
analyses, with pre-assessment scores for participants who discon-
tinued their involvement at any stage (i.e., after they have been A repeated measures MANOVA was performed to investigate
randomised to one of the three conditions) carried forward and group differences on the subscales of the PANAS (i.e., positive affect
used in both the post- and follow-up assessments. Intention-to- and negative affect) over time. No significant differences were
treat analysis is an accepted strategy to address the problem of found for the main effects of group, F(4, 310) = .62, p = .65; Wilks’
attrition and missing data (Gross & Fogg, 2004; Lachin, 2000) Lambda = .98; partial eta squared = .01; or time, F(4, 153) = 1.23,
and has become common practice in internet-based treatment re- p = .27; Wilks’ Lambda = .97; partial eta squared = .03. The interac-
search (Andersson, Strömgren, Ström, & Lyttkens, 2002; Carlbring, tion effect between time and group was not significant,
Westling, Ljungstrand, Ekselius, & Andersson, 2001; Klein et al., F(8, 306) = 1.09, p = .37; Wilks’ Lambda = .94; partial eta
2006; Winzelberg et al., 2000). The means and standard devia- squared = .03.
tions for the dependant variables at all three time points are
shown in Table 2. 3.5. DASS-21
Preliminary assumption testing was conducted with no serious
violations noted. Repeated measures MANOVAs were performed to A repeated measures MANOVA was performed to investigate
investigate differences in mental illness and well-being on the group differences on the subscales of the DASS-21 (i.e., depression,
measures with more than one subscale (i.e., DASS, PANAS, and anxiety and stress) over time. No significant differences were found
OTH). Repeated measures ANOVAs were conducted to test for for the main effects of group, F(6, 310) = .29, p = .94; Wilks’ Lamb-
any group differences on participants’ well-being (i.e., SWLS and da = .99; partial eta squared = .01; or time, F(6, 152) = 1.42, p = .21;
PWI-A) over time. Type I error rate was set at .05. Finally, a Chi- Wilks’ Lambda = .95; partial eta squared = .05. The interaction effect
square test for independence was conducted to examine group dif- between time and group was not significant, F(12, 304) = .77, p = .68;
ferences on adherence to the interventions. Wilks’ Lambda = .94; partial eta squared = .03.
 J. Mitchell et al. / Computers in Human Behavior 25 (2009) 749–760 755

Self-registered online
(n = 169)

Excluded (n=9)
Eligibility Assessed DASS-21 scores in the
‘severe’ or higher range

Random allocation (n=160)

Intervention (3 weeks)

Strengths (n=48) Problem Solving (n=58) Placebo Control (n=54)

Completed (n=17) Completed (n=9) Completed (n=23)

Did not complete (n=31) Did not complete (n=49) Did not complete (n=31)

Post-assessment

Completed (n=17) Completed (n=9) Completed (n=23)

Did not complete (n=31) Did not complete (n=49) Did not complete (n=31)

Follow-up assessment (3-month)

Completed (n=11) Completed (n=5) Completed (n=11)

Did not complete (n=37) Did not complete (n=53) Did not complete (n=43)

Analysis (using ITT)

Analysed (n=48) Analysed (n=58) Analysed (n=54)

Fig. 1. Participant flow through the study from registration to data analysis.

74

73

72
Strengths
PWI-A score

71 Problem solving
Placebo control
70

69

68
Time 1 Time 2 Time 3

Fig. 2. PWI-A means by group at time 1 (pre), time 2 (post) and time 3 (3-month follow-up).

3.6. OTH ks’ Lambda = .92; partial eta squared = .04; and time, F(6, 152) =
2.48, p = .026; Wilks’ Lambda = .95; partial eta squared = .09. The
A repeated measures MANOVA was performed to investigate interaction effect between time and group was not significant,
group differences on the subscales of the OTH (i.e., pleasure, F(12, 304) = .1.36, p = .186; Wilks’ Lambda = .90; partial eta
engagement and meaning) over time. Significant differences were squared = .05. A review of the univariate data indicated a signifi-
found for the main effects of group, F(6, 310) = .2.20, p = .043; Wil- cant time effect for engagement, F(2) = 5.13, p = .006, partial eta
756 J. Mitchell et al. / Computers in Human Behavior 25 (2009) 749–760

squared = .03; and a significant group effect for pleasure the strengths group from pre- to post-assessment and 3-month fol-
F(2) = 4.40, p = .014, partial eta squared = .05. There was no signif- low-up. The effect size for this change was small, compared to the
icant time or group effect for meaning. moderate effect size reported in the email intervention of Seligman
Posthoc analysis using paired samples t-tests showed a et al. (2005). As hypothesised, the problem solving group demon-
significant increase in engagement scores (see Fig. 3) for the strated no change in well-being from baseline to post- or follow-
placebo control group from: Time 1 to Time 2, t(53) = 2.97, up assessment; and the placebo control group showed a slight de-
p < .01 (two-tailed) and; Time 1 to Time 3, t(53) = 2.75, p < .01 crease in well-being from baseline to post-, and then remained sta-
(two-tailed). Posthoc comparisons using the Tukey HSD test indi- ble to follow-up assessment.
cated that pleasure scores (see Fig. 4) were significantly greater Interestingly the SWLS, which like the PWI-A is a cognitive
for the strengths group compared to the placebo control group at measure of subjective well-being, followed the hypothesised up-
Time 2 (p = .01) and Time 3 (p < .01). ward trajectory for the strengths group but this result was not sta-
tistically significant. The difference in results between these two
3.7. Adherence measures of cognitive well-being may be accounted for by the glo-
bal versus domain specific approach used by the SWLS and PWI-A,
Adherence to the intervention was 42.6% (23/54) for the respectively. The PWI-A deconstructs the global cognitive satisfac-
placebo control group; 34.0% (16/47) for the strengths group; tion judgements into targeted life domains, providing a more spe-
15.5% (9/58) for the problem solving group; and overall adherence cific reference point to base participants’ satisfaction judgements,
was 30.2% (48/160). A Chi-square test for independence indicated potentially making it a more sensitive measure of well-being than
a significant association between group and adherence, the SWLS and so more able to detect changes in subjective well-
v2(1, N = 160) = 10.39, p > .01), with a small to moderate effect size being.
(Cramer’s V = .255). The third measure of SWB, the PANAS, addressed the affective
component of well-being and no significant changes were de-
4. Discussion tected. This result indicates that the strengths intervention has
the desired impact on the cognitive component of well-being but
4.1. Well-being not the affective component. Alternatively, the lack of significant
affective change may be because the PANAS is limited to only mea-
The PWI-A results support the first hypothesis with a significant suring activated emotions (e.g., excited, enthusiastic, distressed,
increase in the cognitive component of subjective well-being for guilty) not deactivated emotions (e.g., contented, calm, bored,

3.4

3.3
OTH - engagement scor e

3.2

3.1
Strengths
3 Problem solving
Placebo control
2.9

2.8

2.7

2.6
Time 1 Time 2 Time 3

Fig. 3. OTH-engagement subscale means by group at time 1 (pre), time 2 (post) and time 3 (3-month follow-up).

3.5
3.4
3.3
OTH - Pleasure score

3.2
3.1 Strengths
3 Problem solving
2.9 Placebo control
2.8
2.7
2.6
2.5
Time 1 Time 2 Time 3

Fig. 4. OTH-pleasure subscale means by group at time 1 (pre), time 2 (post) and time 3 (3-month follow-up).
 J. Mitchell et al. / Computers in Human Behavior 25 (2009) 749–760 757

sad) as described in the circumplex model (Russell, 1980). A recent becoming available, such as the Warwick–Edinburgh Mental
online longitudinal study found changes in deactivated, but not Well-being Scales (Tennant et al., 2007).
activated, positive emotions, as a result of a gratitude intervention
(Iyer, 2008). Future research would benefit from measuring both 4.2. Mental illness
activated and deactivated emotions.
The OTH was included in this study for exploratory purposes as The hypothesised reduction in mental illness, as measured by
there is no known longitudinal intervention data for this measure; the DASS-21, was not supported by the results, with no change
it has previously been used as a predictor of life satisfaction rather in depression, anxiety or stress scores over time or by group. This
than as an outcome variable. It was hypothesised that engagement does not support the findings of Seligman et al. (2005) who found
would increase for the strengths intervention group only, however decreases in depression symptoms using the CES-D for their
while there was a trend in the predicted direction, the result was strengths intervention. Examining the mean depression, stress
not significant. The results suggest that engagement, as measured and anxiety scores at baseline may provide an explanation as the
by the OTH subscale, is trait-like rather than state-like, and as a re- mean score for each subscale is at the low end of the normal range,
sult may be less amenable to change. As previously mentioned, the which may be creating a floor effect. This low mean score at base-
engagement subscale of the OTH appears to measure the experi- line would have been exacerbated by excluding participants from
ence of the flow state as a result of using one’s strengths, which re- the study with DASS-21 scores in the ‘severe’ or higher range. To
quires not just cognitive change but behavioural change. The be able to detect change in symptoms of mental illness a less
strengths intervention focuses on cognitive change in the first ses- healthy sample may be required, as was the case in the Seligman
sion and behavioural change in the second session, allowing little et al. study that reported baseline scores on the CES-D of mild
time for participants to experience a change in flow experiences depression.
prior to post-assessment. It is possible that change could be ob-
served at 3-month follow-up, but the small sample size and high 4.3. Adherence
attrition rate made it difficult to detect significant change. There
was an unexpected increase in engagement for placebo control The average adherence to the intervention groups was 31%,
group, which is likely to be a result of a low pre-assessment score with a significant between groups difference indicating that people
for the placebo control relative to the other two groups resulting in were more likely to adhere to the placebo control and strengths
regression toward the mean. The result is statistically significant intervention than to the problem solving intervention. Although
but unlikely to be a meaningful finding. the reason for these differences from the current study cannot be
Significant changes in participants’ orientation to pleasure were conclusively ascertained, one possible explanation is that partici-
recorded for the strengths group when compared to the placebo pants were more likely to complete the placebo control interven-
control at post- and follow-up assessment. It is plausible that an tion because it required less effort (e.g., reading information
increase in pleasure orientation is a by-product of the engagement online) and time compared to the other two groups. The strengths
intervention, but it is surprising not to also observe the hypothe- and problem solving groups required participants to put in more
sised change in engagement. It may be that it takes longer to see and equivalent amounts of effort (e.g., reading, writing, manipulat-
a shift in engagement then it does in pleasure, or that the OTH is ing date on screen and offline tasks).
more sensitive to shifts in pleasure orientation. An alternative pos- The difference in adherence between the two active interven-
sibility is that the results are merely a result of a low placebo con- tions could be accounted for by the focus of the intervention con-
trol mean and high strengths mean score for engagement at tent. The strengths intervention focussed on identifying what
baseline as the difference at baseline approaches significance participants did well and doing more of it; while the problem solv-
(p = .056). If the results are treated as meaningful, then according ing intervention focussed on problems in participants’ lives and
to theory the pleasure orientation is equated to SWB, and so these how to resolve them. Intuitively, it would seem more enjoyable
results support the aforementioned changed on the PWI-A for the and novel to do the strengths intervention than the problem solv-
strengths group. ing intervention. It has been identified that enjoyment is an impor-
Overall, the theory and operationalisation of the OTH requires tant mediator of intervention effectiveness (Lyubomirsky,
greater clarity to gain insight into exactly what is being measured. Dickerhoof, Boehm, & Sheldon, submitted for publication) and
The authors of the OTH note that it elicits people’s endorsement of leads to higher motivation (e.g., Sheldon’s self-concordant motiva-
ways to be happy, rather than actual behaviour (Peterson et al., tion theory). Perhaps health promotion could benefit from the ad-
2005). This result suggests that the OTH is more trait than state- junct of positive psychology interventions to traditional cognitive-
like. However, the OTH results should be treated with caution at behavioural interventions (e.g., problem solving, challenging nega-
this stage. tive thinking) not only because they are effective, but because they
The well-being results of the current study are not as definitive are enjoyable and so likely to increase adherence. It should be
as the results from Seligman et al. (2005) which measured well- noted that although the hypothesised group difference in adher-
being using the Steen Happiness Index (SHI). The SHI was a newly ence was supported by the data, there is not enough information
created, purposely built questionnaire designed to be a combined from this study to determine the exact reason for this difference.
measure of hedonic (SWB) and eudaimonic (PWB) aspects of
well-being. The current study used well established questionnaires 4.4. Attrition
measuring SWB, as well as the OTH as an emerging measure of
SWB and PWB combined. The strengths intervention in the current The attrition rate for internet interventions tends to be varied
study was intended to develop engagement, which is in theory (6–95%) and this study’s overall attrition rate, of 83% at 3-month
more proximal to PWB than SWB. While SWB and PWB are moder- follow-up assessment, is in the high end of the range. Attrition
ately correlated they measure different aspects of well-being, and from internet intervention studies tends to be higher for programs
for this study a clearer picture may have emerged by including that are automated; the implication being that human interaction
an established measure of PWB as a more proximal measure of (e.g., via email, telephone, face-to-face contact) reduces attrition,
well-being. The theory and measurement of well-being is a rapidly although this is still to be tested empirically. The current study
developing area and a number of brief, but valid and reliable pop- used a fully automated internet intervention which would have
ulation level measures that combine SWB and PWB are now contributed to the high attrition rate. In comparison, the reported
758 J. Mitchell et al. / Computers in Human Behavior 25 (2009) 749–760

attrition rate by Seligman et al. (2005) from their positive psychol- internet intervention research offers an immense opportunity to
ogy internet intervention was 29%. As discussed earlier, it is un- develop the field of health promotion world wide.
clear how much of a role human contact played in their study, or
if the mere expectation of support was enough to reduce attrition. 4.6. Limitations
An unforeseen technical factor that is likely to have contributed
to attrition was the website’s tunnel design (Danaher, McKay, & This study had a number of limitations that made it difficult to
Seely, 2005). The tunnel design meant participants needed to detect significant change, most notably: the small sample size;
sequentially complete each stage of the program and participants high attrition; and the low levels of mental illness and high levels
who only partially completed the intervention were actively ex- of well-being at baseline creating floor and ceiling effects. The sam-
cluded from completing the post- and/or follow-up assessment. ple was also largely female, tertiary educated and employed, thus
Allowing participants to complete the post- and follow-up ques- limiting the generalisability of the findings.
tionnaires regardless of adherence to the program would have de- While the benefit of conducting a longitudinal study is to dem-
creased attrition. While this approach ensured intervention onstrate enduring change overtime, the current study only went
fidelity, partial completion may have been enough to create change for a 3-month period. It would be ideal to assess change over years
considering that the majority of the content was in the first two rather than months, especially as some theories of well-being sug-
sessions. gest that change will only ever be temporary and that most people
A consideration for the current study was that high attrition can return to their set point of happiness (Cummins, Gullone, & Lau,
disrupt the randomisation process, which in turn can compromise 2002; Headey, 2008). Finally, as mentioned earlier, the current
the accuracy of the outcome results. To address this issue inten- study may have benefited from the inclusion of a measure of
tion-to-treat (ITT) analysis was used. As noted earlier, ITT analysis PWB which is theoretically more proximal to engagement then
is common practice in internet intervention research and is consid- SWB.
ered to be a conservative statistical approach (i.e., is likely to
underestimate the probability of significant change). As the field
5. Conclusions
of internet intervention research develops, so too are the statistical
techniques being applied to address attrition issues and research-
The results, with some caveats, lend support to the body of lit-
ers should remain aware of and open to advances in this area.
erature indicating that well-being can be enhanced through inten-
While the upside of the internet is that it can reach a large audi-
tional activity (i.e., identifying and using personal strengths) and
ence, the down side is that attrition from these interventions can
that these changes continue on an upward trajectory for at least
be high, as in the case of this study. Internet research is still in
3 months. In this study it is the cognitive, not the affective, compo-
its infancy and issues that impact on attrition, such as website de-
nent of subjective well-being that was amenable to change,
sign and human interaction, need greater exploration.
although it is unknown if this was a reflection of the measures
used, the intervention or both. While no changes in mental illness
4.5. Broad research implications
outcomes were found, no definitive conclusions can be made until
the interventions are tested on a less mentally healthy sample.
Two main points that emerge from this research are support for
Although the results do lend support for mental illness and mental
the theory that (a) it is possible to enduringly enhance well-being,
health as separate constructs, rather than being opposite ends of
and (b) well-being interventions can be effectively delivered via
the same continuum.
the internet. While being cautious about overstating the findings
The results demonstrate that the internet is an effective means
of this particular study, there are a number of health promotion
of disseminating well-being interventions, reflecting the findings
implications stemming from these two points.
of internet research for prevention and treatment of mental illness.
Keyes (2005) research indicated that only 20% of the adult pop-
The fact that the intervention was a fully automated internet-
ulation have high well-being (i.e., flourishing), leaving 80% with
based program, without any need for human contact, increases
low or moderate levels of well-being. As discussed previously,
its sustainability and accessibility in the real world. While high
the broad benefits of high well-being are better physical health, en-
attrition is an issue, delivery via the internet has the potential to
hanced social relationships and enhanced performance at work,
reach a large audience and even if a small percentage complete
school and home; which in turn help create healthy, flourishing
the intervention (e.g., 31% for this study), many more people can
communities. It makes sense to invest in health promotion strate-
be reached compared to traditional modes of dissemination.
gies that improve the well-being of individuals and communities.
Further research is needed to harness the full potential of posi-
Health promotion, however, is often the poor cousin to illness
tive psychology interventions via the internet and address issues of
treatment; perhaps as a result of the long held belief that if illness
adherence, attrition and effect size. However, this study indicates it
is eliminated then well-being will ensue. Overtime this biomedical
is possible to effectively deliver well-being enhancing interven-
approach has demonstrated that it is not sufficiently effective in
tions over the internet with some benefit to participants. These
stemming the growing burden of mental illness (Vaillant, 2003).
findings create exciting possibilities for the future direction of
An alternative option is to place greater focus on enhancing well-
health promotion delivery and the possibility of reaching a mass
being, both as an independent outcome and as an adjunct to men-
audience while creating change at an individual level.
tal illness treatment and prevention. Using such an approach may
serve the dual purpose of creating more flourishing individuals and
reducing the incidence of mental illness. Acknowledgements
Another recognised barrier to effective health promotion has
been the reliance on traditional delivery mechanisms (e.g., face- This research was funded via a Sport and Physical Activity Re-
to-face group programs; media campaigns). Internet delivery of search Network (SPARN) grant obtained by Dr. Michael Martin,
well-being interventions addresses many of the limitations of tra- Head of Performance Psychology, AIS and Dr. Graeme Hyman, Se-
ditional approaches, in particular the internet provides a more nior Lecturer, School of Psychology, Psychiatry and Psychological
accessible, sustainable, and personalised approach to health pro- Medicine, Monash University. The website was developed by Jani-
motion than has previously been possible. Combining what is son. The website content was written by the first author and edited
known from positive psychology and well-being research with by the second author.
 J. Mitchell et al. / Computers in Human Behavior 25 (2009) 749–760 759

References Evers, G. G. (2006). EHealth promotion: The use of the internet for health
promotion. American Journal of Health Promotion, 20(4), 1–7.
Fox, S. (2006). Online Health Search 2006 [Electronic Version]. Pew Internet and
Abbott, J., Kaldo, V., Klein, B., Austin, D., Piterman, L., & Andersson, G. (in press). A
American Life Project, pp. 1–15. Retrieved 12 January 2007 from http://
randomised controlled trial of an internet-based intervention program for
www.pewinternet.org/pdfs/PIP_Online_Health_2006.pdf.
tinnitus distress in an industrial setting. Cognitive Behaviour Therapy.
Gross, D., & Fogg, L. (2004). A critical analysis of the intent-to-treat principle in
ABS. (2006). Household use of information technology (No. 8146.0). Australian Bureau
prevention research. The Journal of Primary Prevention, 25, 475–489.
of Statistics.
Headey, B. (2008). The set-point theory of well-being: Negative results and
Andersson, G., Bergstrom, J., Hollandare, F., Carlbring, P., Kaldo, V., & Ekselius, L.
consequent revisions. Social Indicators Research, 85, 389–403.
(2005). Internet-based self-help for depression: Randomised controlled trial.
Hirai, M., & Clum, G. A. (2005). An internet-based self-change program for traumatic
British Journal of Psychiatry, 187, 456–461.
event related fear, distress, and maladaptive coping. Journal of Traumatic Stress,
Andersson, G., Carlbring, P., Holmstrom, A., Sparthan, E., Furmark, T., Nilsson-Ihrfelt,
18, 631–636.
E., et al. (2006). Internet-based self-help with therapist feedback and in vivo
IWG. (2006). Personal Wellbeing Index. Melbourne: International Wellbeing Group.
group exposure for social phobia: A randomized controlled trial. Journal of
Australian Centre on Quality of Life, Deakin University.
Consulting and Clinical Psychology, 74(4), 677–686.
Iyer, R. (2008). An empirical investigation into the ideal online gratitude
Andersson, G., Strömgren, T., Ström, L., & Lyttkens, L. (2002). Randomized controlled
intervention. Paper presented at the Fourth European Conference on Positive
trial of internet-based cognitive behavior therapy for distress associated with
Psychology, Opatija, Croatia.
tinnitus. Psychosomatic Medicine, 64, 810–816.
Kashdan, T. B., Biswar-Diener, R., & King, L. A. (2008). Reconsidering happiness: The
Antony, M. M., Bieling, P. J., Cox, B. J., Enns, M. W., & Swinson, R. P. (1998).
cost of distinguishing between hedonics and eudaimonia. Journal of Positive
Psychometric properties of the 42-item and 21-item versions of the depression
Psychology, 3(4), 219–233.
anxiety stress scales (DASS) in clinical groups and a community sample.
Kenardy, J., McCafferty, K., & Rosa, V. (2003). Internet-delivered indicated
Psychological Assessment, 10(2), 176–181.
prevention for anxiety disorders: A randomized controlled trial. Behavioural
Carlbring, P., Bohman, S., Brunt, S., Buhrman, M., Westling, B. E., Ekselius, L., et al.
and Cognitive Psychotherapy, 31, 279–289.
(2006). Remote treatment of panic disorder: A randomized trial of internet-
Kenwright, M., Liness, S., & Marks, I. (2001). Reducing demands on clinicians by
based cognitive behavior therapy supplemented with telephone calls. American
offering computer-aided self-help for phobia panic: A feasibility study. British
Journal of Psychiatry, 163(12), 2119–2125.
Journal of Psychiatry, 179, 456–459.
Carlbring, P., Nilsson-Ihrfelt, E., Waara, J., Kollenstam, C., Buhrman, M., Kaldo, V.,
Keyes, C. L. (2005). Mental illness and/or mental health? Investigating axioms of the
et al. (2005). Treatment of panic disorder: Live therapy vs. self-help via the
complete state model of health. Journal of Consulting and Clinical Psychology,
internet. Behaviour Research and Therapy, 43, 1321–1333.
73(3), 539–548.
Carlbring, P., Westling, B. E., Ljungstrand, P., Ekselius, L., & Andersson, G. (2001).
Keyes, C. L. (2007). Promoting and protecting mental health as flourishing: A
Treatment of panic disorder via the internet: A randomized trial of a self-help
complementary strategy for improving national mental health. American
program. Behavior Therapy, 32(4), 751–764.
Psychologist, 62(2), 95–108.
Celio, A., Winzelberg, A. J., Wilfley, D. E., Eppstein-Herald, D., Springer, E. A., & Dev, P.
Keyes, C. L., Shmotkin, D., & Ryff, C. D. (2002). Optimizing well-being: The empirical
(2000). Reducing risk factors for eating disorders: Comparison of an internet-
encounter of two traditions. Journal of Personality and Social Psychology, 82,
and a classroom-delivered psycho-educational program. Journal of Consulting
1007–1022.
and Clinical Psychology, 68, 650–657.
Kiropoulos, L., Klein, B., Austin, D. W., Gilson, K., Pier, C., Mitchell, J., et al. (2008). Is
Cheeseman Day, J., Janus, A., & Davis, J. (2005). Computer and internet use in the
internet-based CBT for panic disorder and agrophobia as effective as face-to-
United States: 2003. Washington: U.S. Census Bureau.
face. Journal of Anxiety Disorders, 22(8), 1273–1284.
Christensen, H., Griffiths, K. M., & Evans, K. (2002). Health in Australia: Implications of
Klein, B., Austin, D., Pier, C., Kiropoulos, L., Shandley, K., Mitchell, J. et al. (in pressa).
the internet and related technologies for policy. ISC Discussion Paper No. 3.
Frequency of email therapist contact and internet-based treatment for panic
Canberra: Commonwealth Department of Health and Ageing.
disorder: Does it make a difference? Cognitive Behaviour Therapy.
Christensen, H., Griffiths, K. M., & Jorm, A. F. (2004). Delivering interventions for
Klein, B., Mitchell, J., Gilson, K., Shandley, K., Austin, D., Kiropoulos, L. et al. (in
depression by using the internet: Randomised controlled trial. British Medical
pressb). A therapist-assisted internet-based CBT intervention for post-traumatic
Journal, 328, 265.
stress disorder: Preliminary results. Cognitive Behaviour Therapy.
Christensen, H., Griffiths, K. M., & Korten, A. (2002). Web-based cognitive behavior
Klein, B., & Richards, J. C. (2001). A brief Internet-based treatment for panic disorder.
therapy: Analysis of site usage and changes in depression and anxiety scores.
Behavioral Cognitive Psychotherapy, 29, 113–117.
Journal of Medical Internet Research, 4, e3.
Klein, B., Richards, J. C., & Austin, D. A. (2006). Efficacy of internet therapy for panic
Christensen, H., Griffiths, K. M., Korten, A. E., Brittliffe, K., & Groves, C. (2004). A
disorder. Journal of Behaviour Therapy and Experimental Psychiatry, 37, 213–238.
comparison of changes in anxiety and depression symptoms of spontaneous
Korp, R. R. (2006). Health on the internet: Implications for health promotion. Health
users and trial participants of a cognitive behavior therapy website. Journal of
Education Research, 21(1), 78–86.
Medical Internet Research, 6(4), e46.
Kypri, K., Saunders, J. B., Williams, S. M., McGee, R. O., Langley, J. D., Cashell-Smith,
Clark, A., Kirkby, K. C., Daniels, B. A., & Marks, I. M. (1998). A pilot study of
M. L., et al. (2004). Web-based screening and brief intervention for hazardous
computer-aided vicarious exposure for obsessive–compulsive disorder.
drinking: A double-blind randomized controlled trial. Society for the Study of
Australian and New Zealand Journal of Psychiatry, 32, 268–275.
Addiction, 99, 1410–1417.
Clarke, G., Reid, E., Eubanks, D., O’Connor, E., DeBar, L. L., Kelleher, C., et al. (2002).
Lachin, J. M. (2000). Statistical considerations in the intent-to-treat principle.
Overcoming depression on the internet (ODIN): A randomized trial of an
Controlled Clinical Trials, 21, 167–189.
internet depression skills intervention program. Journal of Medical Internet
Lange, A., van de Ven, J. P., & Shrieken, B. (2003). Interapy treatment of post-
Research, 4(3), e14.
traumatic stress via the internet. Cognitive Behaviour Therapy, 32, 110–124.
Cobb, N. K., Graham, A. L., Bock, B. C., Papandonatos, G., & Abrams, D. B. (2005).
Lange, A., van de Ven, J. P., Shrieken, B., & Emmelkamp, P. (2001). Interapy treatment
Initial evaluation of a real-world internet smoking cessation system. Nicotine &
of post-traumatic stress through the internet: A controlled trial. Journal of
Tobacco Research, 7(2), 207–216.
Behavior Therapy and Experimental Psychiatry, 32(2), 73–90.
Csikszentmihalyi, M. (1990). Flow: The psychology of optimal experience. New York:
Litz, B., Engel, C., Bryant, R., & Papa, A. (2007). A randomized, controlled proof-of-
HarperCollins.
concept trial of an internet-based, therapist-assisted self-management
Cummins, R. A., Gullone, E., & Lau, A. L. D. (2002). A model of subjective wellbeing
treatment for posttraumatic stress disorder. American Journal of Psychiatry,
homeostasis: The role of personality. In E. Gullone & R. A. Cummins (Eds.), The
164, 1676–1684.
universality of subjective wellbeing indicators: Social indicators research series
Lovibond, S. H., & Lovibond, P. F. (1995). Manual for the depression anxiety stress
(pp. 7–46). Dordrecht: Kluwer.
scales (2nd ed.). Sydney: Psychology Foundation.
Danaher, B. G., McKay, H. G., & Seely, J. R. (2005). The information architecture
Lyubomirsky, S. (2006). The costs and benefits of writing, talking, and thinking
of behaviour change websites. Journal of Medical Internet Research, 7(2),
about life’s triumphs and defeats. Journal of Personality and Social Psychology,
e12.
90(4), 692–708.
DCITA. (2005). Australia Online: 3rd quarter 2004 statistics (PDF). Canberra:
Lyubomirsky, S., Dickerhoof, R., Boehm, J. K., & Sheldon, K. M. (submitted for
Department of Communications, Information Technology and the Arts.
publication). Becoming happier takes both a will and a proper way: Two
de Vries, H., & Brug, J. (1999). Computer-tailored interventions motivating people to
experimental longitudinal interventions to boost well-being.
adopt health promoting behaviours: Introduction to a new approach. Patient
Lyubomirsky, S., King, L., & Diener, E. (2005). The benefits of frequent positive affect:
Education and Counseling, 36(2), 99–105.
Does happiness lead to success? Psychological Bulletin, 131(6), 803–855.
Devineni, T., & Blanchard, E. B. (2005). A randomized controlled trial of an internet-
Lyubomirsky, S., Sheldon, K. M., & Schkade, D. (2005). Pursuing happiness: The
based treatment for chronic headache. Behaviour Research and Therapy, 43(3),
architecture of sustainable change. Review of General Psychology, 9(2),
277–292.
111–131.
Diener, E. (1984). Subjective well-being. Psychological Bulletin, 95(3), 542–575.
McKay, G. H., Glasgow, R. E., Feil, E. G., Boles, S. M., & Barrera, M. (2002). Internet-
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The satisfaction with life
based diabetes self-management and support: Initial outcomes from the
scale. Journal of Personality Assessment, 49, 71–75.
diabetes network project. Rehabilitation Psychology, 47(1), 31–48.
Duckworth, A. L., Steen, T. A., & Seligman, M. E. P. (2005). Positive psychology in
Mihalopoulos, C., Kiropoulos, L., Shih, S. T.-F., Gunn, J., Blashki, G., & Meadows, G.
clinical practice. Annual Review of Clinical Psychology, 1, 629–651.
(2005). Exploratory economic analyses of two primary care mental health
Etter, J. (2005). Comparing the efficacy of two internet-based, computer-tailored
projects: Implications for sustainability. Medical Journal of Australia, 183(10),
smoking cessation programs: a randomized trial. Journal of Medical Internet
S73–S76.
Research, 7(1), e2.
760 J. Mitchell et al. / Computers in Human Behavior 25 (2009) 749–760

Napolitano, M. A., Fotheringham, M., Tate, D., Sciamanna, C., Leslie, E., Owen, N., Ryff, C. D. (1989). Happiness is everything, or is it? Explorations on the meaning of
et al. (2003). Evaluation of an internet-based physical activity interntion: A psychological well-being. Journal of Personality and Social Psychology, 57(6),
preliminary intervention. Annals of Behavioral Medicine, 225, 92–99. 1069–1081.
Nunnally, J. C. (1978). Psychometric theory (2nd ed.). New York: McGraw-Hill. Seligman, M. E. P., & Csikszentmihalyi, M. (2000). Positive psychology: An
Patten, S. B. (2003). Prevention of depressive symptoms through the use of distance introduction. American Psychologist, 55(1), 5–14.
technologies. Psychiatric Services, 54, 396–398. Seligman, M. E. P., Steen, T. A., Park, N., & Peterson, C. (2005). Positive psychology
Pavot, W., & Diener, E. (1993). Review of the satisfaction with life scale. Psychological progress: Empirical validation of interventions. American Psychologist, 60(5),
Assessment, 5(2), 164–172. 410–421.
Peterson, C., & Park, N. (2004). Classification and measurement of character Spek, V., Cuijpers, P., Nyklicek, I., Riper, H., Keyzer, J., & Pop, V. (2007). Internet-
strengths: Implications for practice. In P. A. Linley & S. Joseph (Eds.), Positive based cognitive behaviour therapy for symptoms of depression and anxiety: A
psychology in practice (pp. 433–446). Hoboken, NJ: John Wiley & Sons. meta-analysis. Psychological Medicine, 37, 319–328.
Peterson, C., Park, N., & Seligman, M. E. P. (2005). Orientations to happiness and life Spittaels, H., & de Bourdeaudhuij, I. (2006). Implementation of an online tailored
satisfaction: The full life versus the empty life. Journal of Happiness Studies, 6(1), physical activity intervention for adults in Belgium. Health Promotion
25–41. International, 221(4), 311–319.
Pier, C., Austin, D., Klein, B., Mitchell, J., Schattner, P., Ciechomski, L., et al. (2008). Strom, L., Pettersson, R., & Andersson, G. (2000). A controlled trial of self-help
Evaluation of internet-based behavioural therapy for panic disorder in general treatment of recurrent headache conducted via the internet. Journal of
medical practice. Mental Health in Family Medicine, 5, 29–39. Consulting and Clinical Psychology, 68(4), 722–727.
Proudfoot, J., Goldberg, D., Mann, A., Everitt, B., Marks, I., & Gray, J. (2003). Swartz, L. H. G., Noell, J. W., Schroeder, S. W., & Ary, D. V. (2006). A randomised
Computerized, interactive, multimedia cognitive-behavioural program for control study of a fully automated internet based smoking cessation
anxiety and depression in general practice. Psychological Medicine, 33, 217–227. programme. Tobacco Control, 15, 7–12.
Proudfoot, J., Parker, G., Benoit, M., Manicavasagar, V., & Smith, M. (2007). Helping Tabachnick, B. G., & Fidel, L. S. (2007). Using multivariate statistics (5th ed.). Boston:
patients adjust to the diagnosis of bipolar disorder: the role of an online Pearson Education.
psychoeducation program. Australian and New Zealand Journal of Psychiatry, Tate, D. F., Wing, R. R., & Winett, R. A. (2001). Using internet technology to deliver a
41(Suppl. 2), A498. behavioral weight loss program. Journal of the American Medical Association,
Proudfoot, J., Parker, G., Hyett, M., Manicavasagar, V., Smith, M., Grdovic, S., et al. 285(9), 1172–1177.
(2007). The next generation of self-management education: A web-based Tennant, R., Hiller, L., Fishwick, R., Platt, S., Joseph, S., Weich, S., et al. (2007). The
bipolar disorder program. Australian and New Zealand Journal of Psychiatry, 41, Warwick–Edinburgh mental well-being scale (WEMWBS): Development and
903–909. UK validation. Health and Quality of Life Outcomes, 5(63).
Proudfoot, J., Ryden, C., Everitt, B., Shapiro, D. A., Goldberg, D., Mann, A., et al. (2004). Vaillant, G. E. (2003). Mental health. American Journal of Psychiatry, 160(8),
Clinical efficacy of computerised cognitive behavioural therapy for anxiety and 1373–1384.
depression in primary care. British Journal of Psychiatry, 185, 46–54. Vandelanotte, C., De Bourdeaudhuij, I., Sallis, J. F., Spittaels, H., & Brug, J. (2005).
Proudfoot, J., Ryden, C., & Goldberg, D. (2005). ‘Beating the Blues’: Computer CBT Efficacy of sequential or simultaneous interactive computer-tailored
program for anxiety and depression. Australian Journal of Psychology, 57(Suppl.), interventions for increasing physical activity and decreasing fat intake. Annals
245. of Behavioral Medicine, 29, 138–146.
Richards, J. C., Klein, B., & Austin, D. W. (2006). Internet cognitive behavioural Walters, S. T., Miller, J. E., & Chiauzzi, E. (2005). Wired for wellness: e-Interventions
therapy for panic disorder: Does the inclusion of stress management for addressing college drinking. Journal of Substance Abuse Treatment, 29,
information improve end-state functioning? Clinical Psychologist, 10, 2–15. 139–145.
Ritterband, L. M., Cox, D. J., Gordon, T., Borowitz, S. M., Kovatchev, B., Walker, L. S., Walters, S. T., Wright, J. A., & Shegog, R. (2006). A review of computer and internet-
et al. (2006). Examining the added value of audio, graphics, and interactivity in based interventions for smoking behavior. Addictive Behaviors, 31(2), 264–277.
an internet intervention for pediatric encopresis. Children’s Health Care, 35, Wantland, D. J., Portillo, C. J., Holzemer, W. L., Slaughter, R., & McGhee, E. M. (2004). The
47–59. effectiveness of web-based vs. non-web-based interventions: A meta-analysis of
Ritterband, L. M., Cox, D. C., Walker, L., Kovatchev, B., McKnight, L., & Patel, K. behavioral change outcomes. Journal of Medical Internet Research, 6(4), e40.
(2003). A web-based intervention as adjunctive therapy for pediatric Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief
encopresis. Journal of Consulting and Clinical Psychology, 71, 910–917. measures of positive and negative affect: The PANAS scales. Journal of
Ritterband, L. M., Gonder-Frederick, L. A., Cox, J. C., Clifton, A. D., West, R. W., & Personality and Social Psychology, 54(6), 1063–1070.
Borowitz, S. M. (2003). Internet interventions: In review, in use, and into the Winett, R. A., Roodman, A. A., Winett, S. G., Bajzek, W., Rovniak, L. S., & Whiteley, J. A.
future. Professional Psychology Research and Practice, 34(5), 527–534. (1999). The effects of the Eat4Life internet-based health behavior program on
Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social the nutrition and activity practices of high school girls. Journal of Gender, Culture
Psychology, 39(6), 1161–1178. and Health, 4(3), 239–254.
Ryan, R. M., & Deci, E. L. (2001). On happiness and human potentials: A review of Winzelberg, A. J., Eppstein, D., Eldredge, K. L., Wifley, D., Dasmahapatra, R., Dev, P.,
research on hedonic and eudaimonic well-being. Annual Review of Psychology, et al. (2000). Effectiveness of an internet-based program for reducing risk factors
52, 141–166. for eating disorders. Journal of Consulting and Clinical Psychology, 68, 346–350.