This article was downloaded by: [Newcastle University]
On: 16 July 2012, At: 01:14
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Psychology & Health
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/gpsh20
Affective and health-related outcome
expectancies for physical activity in
older adults
a
b
Paul Gellert , Jochen P. Ziegelmann & Ralf Schwarzer
a c
a
Deparment of Health Psychology, Freie Universität Berlin, Berlin
14195, Germany
b
German Centre of Gerontology, Berlin, Germany
c
Warsaw School of Social Sciences and Humanities, Wroclaw,
Poland
Version of record first published: 25 Aug 2011
To cite this article: Paul Gellert, Jochen P. Ziegelmann & Ralf Schwarzer (2012): Affective and
health-related outcome expectancies for physical activity in older adults, Psychology & Health,
27:7, 816-828
To link to this article: http://dx.doi.org/10.1080/08870446.2011.607236
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-andconditions
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation
that the contents will be complete or accurate or up to date. The accuracy of any
instructions, formulae, and drug doses should be independently verified with primary
sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand, or costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.
Psychology and Health
Vol. 27, No. 7, July 2012, 816–828
Affective and health-related outcome expectancies for physical
activity in older adults
Paul Gellerta*, Jochen P. Ziegelmannb and Ralf Schwarzerac
a
Deparment of Health Psychology, Freie Universität Berlin, Berlin 14195, Germany;
German Centre of Gerontology, Berlin, Germany; cWarsaw School of Social Sciences
and Humanities, Wroclaw, Poland
b
Downloaded by [Newcastle University] at 01:14 16 July 2012
(Received 30 March 2011; final version received 18 July 2011)
This study tests the effects of affective and health-related outcome
expectancies on physical exercise, assuming stronger direct and indirect
(via intention) effects from affective outcome expectancy to physical
exercise than from health-related outcome expectancy to exercise.
Physical exercise and social cognitive variables were assessed at baseline,
and 6- and 12-month follow-up in 335 older adults (60–95 years of age).
Applying structural equation modelling, there was a direct effect from
affective, but not from health-related outcome expectancy on intentions
and behaviour. Also, the indirect effect from self-efficacy on physical
exercise via affective outcome expectancy was significant, whereas the
mediation via health-related outcome expectancy was not. These findings
emphasise the relative importance of affective versus health-related
outcome expectancies in predicting intentions and physical exercise in
older adults and highlight the importance to separate these facets at a
conceptual level to enhance both theory development and health
promotion.
Keywords: outcome expectancy; affective; physical activity; physical
exercise; intention; self-efficacy
Introduction
In social cognitive theory (SCT), outcome expectancy describes a contingency
between a given behaviour of an individual and the consequences following that
behaviour (Bandura, 1997). The construct of outcome expectancy is common in
physical activity research in older adults (Rotter, 1954; Williams, Anderson, &
Winett, 2005). Health behaviour change begins with intention formation, which is
facilitated by a favourable balance of positive and negative expectations regarding
outcomes of the respective behaviour. While health behaviour change messages
typically entail information of more distant health benefits, the more immediate
payoffs of engaging in health behaviours, such as affective responses, are typically not
part of health behaviour change messages. A growing body of recent studies has
highlighted the importance of affective components of outcome expectancy
*Corresponding author. Email: [email protected]
ISSN 0887–0446 print/ISSN 1476–8321 online
ß 2012 Taylor & Francis
http://dx.doi.org/10.1080/08870446.2011.607236
http://www.tandfonline.com
Downloaded by [Newcastle University] at 01:14 16 July 2012
Psychology and Health
817
(e.g. Conner, Rhodes, Morris, McEachan, & Lawton, 2011). Thus, in this study, we
examine the relative effects of affective and health-related outcome expectancies on
physical activity in older adults. We hypothesise a stronger direct and indirect effects
(via intentions) of the affective outcome expectancy component as opposed to the
health-related component.
Despite slightly different labels and conceptualisations of outcome expectancy
within different theoretical approaches, such as the transtheoretical model (TTM;
Prochaska & DiClemente, 1992), the theory of planned behaviour (TPB; Fishbein &
Ajzen, 1975) or the SCT (Bandura, 1997), these conceptualisations have a strong
overlap, as they all involve expected outcomes of physical activity. Within the TTM
(stages of change), outcome expectancy corresponds to the decisional balance
construct, which involves the weighing of perceived pros and cons of a given
behaviour. A positive balance is associated with stage progress and a greater
likelihood to perform the behaviour (Prochaska & DiClemente, 1992). The TPB
conceptualises outcome expectancies as behavioural beliefs underlying the attitude
construct, indicating that a behaviour will lead to a given outcome (Ajzen, 1991). In
SCT, outcome expectancy describes a contingency between an individual’s behaviour
and the consequences following that behaviour (Bandura, 1997).
In sum, outcome expectancy is generally defined as an expectation that a specific
outcome will follow a given behaviour (Williams et al., 2005). However, subdividing
the conceptualisation in terms of different facets of the expected outcomes may
increase the explanatory power of the construct. Affective (e.g. expecting to feel good
after exercise) versus health-related (e.g. expecting better health through exercise)
outcome expectancy has been shown to be a useful distinction (Lawton, Conner, &
Parker, 2007; Trafimow & Sheeran, 1998; van den Berg, Manstead, van der Pligt, &
Wigboldus, 2005), especially given the finding that older adults have a preference for
emotionally meaningful goals (Carstensen & Mikels, 2005).
Affective and health-related components
Anticipated affective responses to physical activity represent an understudied class of
outcome expectancies, as research has focused mainly on health benefits of physical
activity (Richard, van der Pligt, & de Vries, 1996; Williams et al., 2005). There is a
growing body of evidence that shows that affective outcome expectancies and healthrelated (synonymous to cognitive or instrumental; Lawton, Conner, & McEachan,
2009) outcome expectancies are distinct from each other (Trafimow & Sheeran, 1998;
Trafimow et al., 2004). In a large community-based sample, it was found that
affective, but not health-related outcome expectancies, predicted physical activity
over a 6-month period (Lowe, Eves, & Carroll, 2002). In a similar fashion, affective
outcome expectancy (attitudes in TPB) was a more powerful predictor of physical
activity (d ¼ 0.34) and the intention to be physically active (d ¼ 0.46) than healthrelated outcome expectancy (Lawton et al., 2009). A similar pattern was shown for
eight other health behaviours. The findings underscore the importance of affect in
the performance of health-enhancing behaviours and suggest that interventions
could target the affective consequences of engaging in these behaviours. Conner et al.
(2011) investigated the differential impact of affective and health-related outcome
expectancies on self-reported exercise in an experimental design. They compared an
affective message condition, a health-related message condition and a no-message
Downloaded by [Newcastle University] at 01:14 16 July 2012
818
P. Gellert et al.
control condition. Affective messages consistently produced greater increases in
physical exercise than in the other conditions.
There is evidence (Carstensen & Mikels, 2005; Scheibe & Carstensen, 2010) that
older persons apparently encode, hold in mind and decode emotionally positive
information better in comparison to negative information, which also has implications for expected behavioural consequences. Older adults as opposed to younger
ones rated emotionally positive messages as more informative than negative
messages, and they remembered a higher proportion of positive to negative messages
(Shamaskin, Mikels, & Reed, 2010). However, older adults falsely remembered
negative messages as being positive. Hence, valenced health care messages influence
motivational processes and behaviour (e.g. Conner et al., 2011), but beyond that the
processing of valenced information changes with age (Shamaskin et al., 2010). As
shown by Freund, Hennecke, and Riediger (2010), older adults prefer a process focus
in terms of direct outcomes (e.g. enjoying exercise), rather than a focus on healthrelated outcomes (e.g. weight control). Furthermore, only process focus was related
to emotional status. This could mean that especially for older adults there is a
preference of emotionally gratifying outcomes, and that affective outcome expectancy is a stronger predictor of behaviour.
Outcome expectancy in SCT
In SCT (Bandura, 1989, 1997), outcome expectancy is assumed to influence
behaviour not only indirectly through behavioural intentions, but also directly,
whereas in the TPB (Fishbein & Ajzen, 1975) only an indirect influence is assumed.
In SCT, self-efficacy as the perceived capability to perform behaviour is regarded as
the most influential predictor of human behaviour. Thereby, self-efficacy influences
outcome expectancy (for a debate on the direction of the influence, see Williams,
2010), intention and behaviour. Some studies with older adults have shown that
generic outcome expectancy is related to self-efficacy (r ¼ 0.240.70), and that
outcome expectancy accounts for at least some variation in physical activity beyond
that accounted for by self-efficacy (for a review, see Williams et al.,2005). According
to SCT, affective outcome expectancy should also influence physical activity (van der
Pligt & de Vries, 1998), although little research exists to support this (Williams et al.,
2005). Studies based on the TPB have consistently shown that generic outcome
expectancy (i.e. behavioural beliefs/attitudes) has a small to moderate association
with physical activity intention (r ¼ 0.210.50, see Williams et al., 2005). Lawton
et al. (2009) investigated in a TPB study the role of affective outcome expectancies
(attitudes in TPB) for a number of health behaviours. They found that, for at least 9
out of 14 health behaviours, except for physical activity, there was an effect
(if intention was not included in the model) of affective outcome expectancies on
behaviour. Contrary to other health behaviours, the prediction by affective outcome
expectancy diminished for physical activity when including intention into the model.
This refers to an indirect effect from affective outcome expectancies via intention to
physical activity (mediation). As affective outcome expectancies are more potent
predictors of both intentions and behaviour than health-related ones, at least for
some behaviours (Lawton et al., 2007, 2009), it still remains questionable whether
physical exercise can be predicted directly (not mediated via intention) or indirectly
when mediated via intentions.
Psychology and Health
819
Aims and hypotheses
Downloaded by [Newcastle University] at 01:14 16 July 2012
In this study, we examine the effects of affective and health-related outcome
expectancies on physical exercise within the social cognitive model. In addition to all
paths outlined by this model (Figure 1), we specify affective and health-related
outcome expectancies as two simultaneous mediators and compare their putative
mediator roles.
Hypothesis 1: First, we hypothesise a stronger direct effect of affective outcome
expectancy in comparison to health-related outcome expectancy on physical exercise.
Hypothesis 2: Moreover, we hypothesise a stronger indirect effect (mediation)
from affective outcome expectancy via intention to physical exercise than from
health-related outcome expectancy via intention to exercise.
Hypothesis 3: Further, we hypothesise that there is a stronger indirect effect from
self-efficacy via affective outcome expectancy to physical exercise than from
self-efficacy via health-related outcome expectancy on physical exercise.
Hypothesis 4: Also, the indirect effect from self-efficacy via affective outcome
expectancy and via intention to physical exercise (four-variables chain) is expected to
be stronger than the indirect effect from self-efficacy via health-related outcome
expectancy and via intention to physical exercise.
Finally, we examine whether there is a stronger direct or indirect (via intention)
effect from both outcome expectancies to exercise. Thus, the question is whether the
two components of outcome expectancies operate directly on exercise or in an
indirect manner (mediated via intention on physical exercise).
Design
Participants and procedure
This study focused on the adoption and maintenance of physical exercise in older
adults over a time span of 12 months, with three measurement points in time. Ethical
Figure 1. Structural model for predicting physical exercise (N ¼ 335).
Notes: Bold arrows indicate significant mediation paths on physical exercise.
*p 5 0.05; **p 5 0.01; ***p 5 0.001.
820
P. Gellert et al.
Downloaded by [Newcastle University] at 01:14 16 July 2012
guidelines were followed, and clearance from the ethics committee was obtained.
Inclusion criteria were being older than 60 years and not having a medical
contraindication to perform physical exercise. Participants were recruited via
newspaper announcements. After sending back the informed consent form,
participants received questionnaires to measure social-cognitive variables as well as
physical exercise levels via mail at baseline, 6 months later and again 12 months later.
Participants who completed all three questionnaires could choose between two books
on the topic of physical activity in old age. At baseline (Time 1 [T1]), 418 participants
completed the questionnaire. The second questionnaire (Time 2 [T2], 6 months later)
was answered by 340 participants, whereas 335 participants completed the third
questionnaire (Time 3 [T3], 12 months later). The final longitudinal sample
comprised 159 women and 176 men (80% of baseline) with a mean age of 66.7
years, SD ¼ 4.93, ranging from 60 to 95 years.
Measures
Outcome expectancy was measured with a six-item scale, with a six-point response
format, divided into affective and health-related outcome expectancies. Affective
outcome expectancy consisted of two items asking participants ‘If I would exercise on
a regular basis, then I would feel less disheartened and gloomily’, and ‘[. . .] then I
would feel well-balanced in my daily life’. The other four items measured healthrelated outcome expectancy, referring to the short- and long-term benefits of being
physically active. An example is ‘If I would exercise on a regular basis, then I would
be doing something good for my health in future years’, or ‘[. . .] then there is also a
short-term benefit for my health’. Factor analysis suggested a solution with two
factors for the six items, distinguishing between affective (factor loadings between
0.91 and 0.92) and health-related (factor loadings between 0.61 and 0.76)
components of outcome expectancy. Item examples are translations from German
and are adapted from Schwarzer, Luszczynska, Ziegelmann, Scholz, and Lippke
(2008).
Perceived self-efficacy was measured with a six-item scale (Scholz, Sniehotta, &
Schwarzer, 2005), with a six-point response format. Examples were ‘I am confident
that I can adopt a physically active lifestyle’, or ‘I am confident that I can resume a
physically active lifestyle, even if I have relapsed several times’. Factor analysis
suggested a solution with one factor for all six items (factor loadings between 0.72
and 0.88).
Intentions consisted of two single-item indicators (Schwarzer et al., 2008),
measured with a six-point response format. An example item is ‘I intend to exercise
on a regular basis’.
Physical exercise as the outcome measure was assessed at baseline (T1)
and 12 months later (T3). Based on the PAQ-50þ (adopted from Huy &
Schneider, 2008), participants were asked ‘Within the last 7 days, how often did
you engage on average in sports/exercise (e.g. swimming, ball games, running,
walking, water sports)?’ Then, they were asked to report the frequency (days)
and duration of physical exercise/sports (hours and minutes) over the last 7 days.
For the analyses, a composite score (frequency duration) was formed.
Physical exercise at T1 averaged 0.53 h a day, SD ¼ 0.57 (at T3: M ¼ 0.58,
SD ¼ 0.64) (Table 1).
821
Psychology and Health
Table 1. Estimated means, standard deviations, intercorrelations, covariances and variances
for latent social-cognitive variables and physical exercise.
1
1.
2.
3.
4.
5.
6.
Outcome expectancy (affective)
Outcome expectancy (health-related)
Self-efficacy
Intention
Physical exercise T1a
Physical exercise T3a
Downloaded by [Newcastle University] at 01:14 16 July 2012
Factor loadings for two
manifest indicators
Mean (M)b
Standard deviations (SD)b
Cronbach’s alphab
2
3
4
1.05
0.49***
0.36***
0.40***
0.12**
0.21***
0.39
0.61
0.35***
0.42***
0.12**
0.11*
0.37
0.27
1.01
0.54***
0.34***
0.23***
0.34
0.27
0.45
0.69
0.18***
0.23***
0.98/0.77
0.99/0.64
0.96/0.94
0.93/0.87
5.13
1.19
0.75c
5.08
0.88
0.74
5.02
1.01
0.91
5.48
0.93
0.82c
5a
6a
0.07
0.05
0.19
0.09
0.33
0.45***
0.14
0.05
0.14
0.12
0.17
0.41
–
–
0.53
0.57
–
0.58
0.64
–
Notes: Intercorrelations are presented below the diagonal, covariances are presented above the
diagonal and variances are presented in the diagonal and in boldface. N ¼ 335. T1 ¼ Time 1,
T2 ¼ Time 2, T3 ¼ Time 3.
a
Hours per day.
b
Manifest scale means, standard deviations and internal consistency.
c
Pearson correlation (two-item indicator).
*p 5 0.05, **p 5 0.01, ***p 5 0.001 (two-tailed).
Assessment of potential covariates
There were no significant effects of gender ( ¼ 0.001, p ¼ 0.98) and age ( ¼ 0.07,
p ¼ 0.15) on physical exercise outcome. The values of the hypothesised path model
were similar with and without gender and age, whereas the fit indices were less
optimal (model with gender, age and SF-12: RMSEA ¼ 0.04 (90%CI ¼ 0.01, 0.05),
SRMR ¼ 0.04, CFI ¼ 0.99, TLI ¼ 0.98, 2(46) ¼ 66.08, p ¼ 0.03). Therefore, gender
and age were excluded as covariates in the final model, and only baseline physical
exercise served as the covariate. To regress physical exercise on the hypothesised SCT
predictor variables, we controlled for baseline exercise levels. It should be noted,
however, that by controlling for baseline exercise level in this way it is possible that
the effects of social cognitive variables on exercise might be underestimated
(Weinstein, 2007).
Analytical procedure
The data analysis specified the SCT within a structural equation modelling
framework. In order to test the social cognitive model, direct and indirect paths
were estimated (Figure 1). The model was evaluated based on recommended fit
indices using Mplus Version 5.21 (L.K. Muthén & B.O. Muthén, 1998–2007). We
report fully standardised beta coefficients reflecting the variances of the continuous
variables as well as the variances of the background and outcome variables for
standardisation.
In order to reduce the number of parameters needed to define a construct,
increase the stability of the parameter estimates and improve the normality of the
distributions, parcelling was applied. A parcel can be defined as an aggregate-level
indicator comprising the average of two or more items (Little, Cunningham, Shahar,
& Widaman, 2002). Thereby, we created two manifest indicators per latent construct.
822
P. Gellert et al.
Downloaded by [Newcastle University] at 01:14 16 July 2012
We have spread items with maximal similar item content across parcels to generate
two parcels per latent construct that were equivalent in terms of mean, SD and
skewness.
Behaviour scores larger than three SDs were truncated. Missing data (55%) were
treated using the full information maximum likelihood (FIML) algorithm
(L.K. Muthén & B.O. Muthén, 1998–2007). Under ignorable missing data conditions
(missing completely at random and missing at random), FIML estimates are
regarded as unbiased and efficient in terms of convergence, parameter estimate bias,
parameter estimate efficiency and model goodness of fit (Enders & Bandalos, 2001).
Results
Drop out analyses yielded no baseline differences between participants in the
longitudinal sample and those who dropped out in terms of gender, age, marital
status, educational level, social-cognitive variables and physical exercise. A selective
dropout occurred for self-efficacy only (t(416) ¼ 0.24, p ¼ 0.02). The participants
who dropped out at T3 had lower baseline values (T1) for self-efficacy (M ¼ 4.72,
SD ¼ 1.22) than participants who remained in the longitudinal sample (M ¼ 5.03,
SD ¼ 1.02) with the difference corresponding to a small to medium effect size
(d ¼ 0.27).
Predicting physical exercise
The hypothesised model fit the data well (RMSEA 5 0.001 [90%CI ¼ 0.00, 0.04],
SRMR ¼ 0.026, CFI ¼ 1.00, TLI ¼ 1.00, 2(25) ¼ 23.23, p ¼ 0.56; 2/df ¼ 0.93).
Figure 1 displays the standardised parameter estimates for the tested model. A
baseline model that collapses affective and cognitive outcome expectancies – rather
than dividing them into two separate constructs – fits the data equally well
(RMSEA 5 0.001, [90%CI ¼ 0.00, 0.05], SRMR ¼ 0.022, CFI ¼ 1.00, TLI ¼ 1.00,
2(14) ¼ 13.46, p ¼ 0.49; 2/df ¼ 0.96). As these models are non-nested, a descriptive
comparison slightly favours the hypothesised model (AIC ¼ 7036, BIC ¼ 7181)
compared to the baseline model (AIC ¼ 7166, BIC ¼ 7295) when the number of
manifest variables is held constant. Just one general outcome expectancy factor
(baseline model) would mask the differential prediction pattern of the two
components of outcome expectancy (hypothesised model). Therefore, we further
focus on the hypothesised model with two separate outcome expectancy facets. Of
the physical exercise variance, 24% was accounted for by the hypothesised model.
With an interest in the direct and indirect (via intention) effects of the outcome
expectancy facets on exercise behaviour, the total effect can be decomposed into
direct and indirect components (Hayes, 2009). The total effect – that is without
including intention as a proposed mediator into the model – of affective outcome
expectancy on behaviour is ¼ 0.16, ([95%CI: 0.03, 0.18], p ¼ 0.007), whereas for
health-related outcome expectancy it is ¼ 0.04, ([95%CI: 0.13, 0.07], p ¼ 0.55).
For testing the hypotheses, intention is included into the model partitioning the
effect of affective and health-related outcome expectancies into the direct and
indirect effect components. For Hypothesis 1, there was a significant direct effect on
physical exercise by affective outcome expectancy ( ¼ 0.14, [95%CI: 0.02, 0.26],
p ¼ 0.02), whereas the direct effect from health-related outcome expectancy
( ¼ 0.06, [95%CI: 0.17, 0.05], p ¼ 0.29; Hypothesis 1) on exercise
Psychology and Health
823
Downloaded by [Newcastle University] at 01:14 16 July 2012
was nonsignificant. For Hypothesis 2, the indirect effect from affective outcome
expectancy via intention ( ¼ 0.02, [95%CI: 0.004, 0.046], p ¼ 0.10) to physical
exercise, but also via health-related outcome expectancy ( ¼ 0.03, SE ¼ 0.03
[95%CI: 0.003, 0.058], p ¼ 0.08) was marginally significant. For Hypothesis 3, a
significant indirect effect was obtained from self-efficacy via affective outcome
expectancy ( ¼ 0.05, [95%CI: 0.005, 0.097], p ¼ 0.03) to physical exercise. In
contrast, the indirect effect from self-efficacy to physical exercise via health-related
outcome expectancy ( ¼ 0.02, [95%CI: 0.062, 0.019], p ¼ 0.30) was
nonsignificant. Also for Hypothesis 4, the indirect effect (four-variables chain)
from self-efficacy via affective outcome expectancy and via intention to physical
exercise was nonsignificant ( ¼ 0.01, [95%CI: 0.002, 0.017], p ¼ 0.12). The same
applies to the indirect effect from self-efficacy via health-related outcome expectancy
and via intention to physical exercise ( ¼ 0.01, [95%CI: 0.002, 0.020], p ¼ 0.09;
Hypothesis 4).
Discussion
In this study, we examined the effects of affective and health-related outcome
expectancies on physical exercise in older adults within the framework of SCT over a
12-month period of time. We specified affective and health-related outcome
expectancies as two parallel direct predictors of physical exercise, as well as indirect
predictors via intention. Furthermore, we analysed all indirect paths within the social
cognitive model from self-efficacy via affective and health-related outcome expectancies on behaviour, and via intention on behaviour. Figure 1 shows all
hypothesised paths by the social cognitive model and the parallel outcome
expectancies. In our model, the temporal order among the study variables was
established in line with the tenets of SCT, while also recognising that no conclusive
proof of causal ordering can be derived when using a non-experimental research
design. In this study, the temporal ordering was given by operationalising Time 2
outcome expectancies, 6 months after Time 1 and 6 months before the outcome
variable physical exercise.
As hypothesised, affective outcome expectancy was a stronger predictor of
physical exercise 6 months later, compared to health-related outcome expectancy. In
addition, the mediation from self-efficacy on physical exercise via affective outcome
expectancy was significant, whereas the mediation via health-related outcome
expectancy was not. All paths (four-variables chains) from self-efficacy via one of the
types of outcome expectancy and via intention to physical exercise 12 months after
self-efficacy measurement took place were nonsignificant. The direct path from
self-efficacy to exercise was also nonsignificant, but the mediation from self-efficacy
via intention to physical exercise was significant.
These results emphasise the relative importance of affective versus health-related
outcome expectancies in predicting intentions and physical exercise in older adults,
which is in line with a number of studies (e.g., Conner et al., 2011; Kraft, Rise,
Sutton, & Røsamb, 2005; Lawton et al., 2009). Kraft et al. (2005) found in a TPB
study for exercise that affective outcome expectancy played a more important role in
the intention formation process than did health-related outcome expectancy next to
perceived behavioural control. Lawton et al. (2009) found that there was no direct
effect of affective outcome expectancy on physical exercise when intention was
Downloaded by [Newcastle University] at 01:14 16 July 2012
824
P. Gellert et al.
included into the model, which was unexpected, as an effect of affective outcome
expectancy was hypothesised by the authors and was found for other behaviours in
that study. In contrast to Lawton et al. (2009), our results showed a direct effect (not
via intention) of affective outcome expectancy as it is also assumed in the social
cognitive model. Affective outcome expectancy was the strongest direct predictor of
physical exercise, and intention was a significant direct predictor too. As there is no
direct path of outcome expectancy assumed in the TPB, the SCT appeared to be
more appropriate in explaining exercise behaviour. Conner et al. (2011) showed in an
experimental design that affective messages consistently produced greater increases
in physical exercise than the health-related and the control conditions, which
underscores the fact that exercise is driven by affective expectations or at least by
affective cues. A review by Williams et al. (2005) concluded that increases in specific
emotional responses to acute exercise (such as revitalisation and positive
engagement) could enhance behaviour.
One could argue that health-related outcome expectancies should play a major
role in predicting exercise of older adults because of their higher morbidity levels. But
we did not find stronger effects of age with affective outcome expectancy than for
health-related outcome expectancy. Furthermore, one could argue that especially for
older adults there might be a preference for emotionally gratifying stimuli or a
neglect of negative information (Shamaskin et al., 2010) and a preference for the
process of being active itself (Freund et al., 2010), rather than long-term health
benefits that reflects a product.
Limitations and future directions
Some limitations need to be mentioned. Data are based on self-reports. Although
self-report measures of physical exercise are common in health behaviour research,
adding objective measures would be appropriate (Prince et al., 2008). Weinstein
(2007) describes problems that might emerge with ongoing or repeated behaviours
that have been and are being performed frequently, such as physical exercise. In such
a case, controlling for prior behaviour is likely to underestimate the effects of social
cognition on subsequent behaviour. On the other hand, the effect of social cognitions
on behaviour can be overestimated because behaviour also causes subsequent social
cognitions and vice versa, which leads to an infinite regress (Weinstein, 2007). To
better assess such reciprocal relationships in experimental designs, modelling change
over time by micro-level analyses is desirable (Williams & Dunsiger, 2007).
Another limitation stems from the fact that we used only two- and four-item
measures for outcome expectancy. However, a more elaborated measure would be of
advantage (e.g. Physical Activity Enjoyment Scale; Kendzierski & DeCarlo, 1991), to
assess also other affective responses, such as dislike of exercise, that have been shown
to predict physical exercise (Brownson, Baker, Housemann, Brennan, & Bacak,
2001; Williams et al., 2005). The generalisability of the results might be somewhat
restricted due to the recruitment of participants via newspaper announcements and
selective dropout. Another shortcoming is related to the operationalisation of the
constructs involved. The items used to measure intentions, self-efficacy, and
health-related as well as affective outcome expectancies do differ somewhat in
their focus, so that affective outcome expectancies refer to exercise on a regular basis,
Downloaded by [Newcastle University] at 01:14 16 July 2012
Psychology and Health
825
whereas self-efficacy refers to adopting a physically active lifestyle. In future studies,
items should be more consistent in terms of target, action, context and time.
Future research should include a broader conceptualisation of the outcome
expectancy construct, distinguishing between affective and cognitive (e.g. healthrelated) outcomes. Further, there are not only indirect effects from affective outcome
expectancy via intention to exercise, but also direct effects of affective outcome
expectancy on behaviour, which could be informative for intervention and model
development. Future studies should include measures of affective states (e.g. Watson,
Clark, & Tellegen, 1988) rather than only affective outcome expectancy. More
research is needed to disentangle these effects (Richard et al., 1996), as affective
outcome expectancy is conceptualised as cognitive expectations about affective states
as a consequence of a specific behaviour and not as the affective states itself.
Future studies should also include moderators such as different outcome values
(e.g. Rodgers & Brawley, 1996), subgroups with different affective or cognitive styles
(Conner et al., 2011; Trafimow et al., 2004) or different behaviours (Lawton et al.,
2009; Trafimow et al., 2004). Although our model fits the data well, alternative
models with different causal sequence of the SCT variables (Williams, 2010) and the
possibility for reverse causation (Weinstein, 2007) shall not be precluded thereby.
This should be tested in future research. Furthermore, researchers should conduct
interventions with rigorous experimental designs (Lippke & Ziegelmann, 2008;
Weinstein, 2007).
We only looked for positive outcome expectancy, as in previous studies it was
found to be more suitable for older adults (for a review, see Williams et al., 2005).
But the question about potential effects of negative outcome expectancies especially
in combination with expected negative affective outcomes remains. Also, other
subdivisions of the outcome expectancy construct may be of interest, for example,
the temporal proximity of the behavioural outcomes, particularly distant versus
immediate consequences of health behaviour (Orbell & Kyriakaki, 2008; Strathman,
Gleicher, Boninger, & Edwards, 1994). As the short-term and long-term healthrelated outcome expectancy loaded on one factor in our study, we collapsed them
into a single construct, but further research and different measures are needed to gain
more insight into this issue (Joireman, Balliet, Sprott, Spangenberg, & Schultz, 2008;
Strathman et al., 1994).
Conclusions and practical implications
The present results highlight the relative importance of affective as opposed to
health-related outcome expectancies in predicting physical exercise in older adults
within the social cognitive framework. Further, these results emphasise also direct
(not via intentions) effects rather than exclusively mediated effects of affective
outcome expectancy on physical exercise. For practical implications, the results
suggest that interventions should focus on emotionally positive information,
especially in older adults or in those with a more limited future time perspective
(Mikels et al., 2010, Ziegelmann, Lippke, & Schwarzer, 2006), rather than only on
health-related information. Enjoying the exercise itself (Freund et al., 2010) or
highlighting the short-term affective benefits after and during physical exercise
should lead to better intervention outcomes (Conner et al., 2011). Another way to
increase the affective relevance of exercise for individuals (and therefore increasing
826
P. Gellert et al.
the affective outcome expectancy as well) could be to include emotionally close
individuals or the intimate partner in interventions and thus to foster joint exercise.
It has been shown that in those participants whose partners took part in a physical
activity intervention, physical activity increased substantially over time, as compared
to individuals whose partners were not involved in the intervention (Gellert,
Ziegelmann, Warner, & Schwarzer, in press). Finally, although there is need for
focusing on affective outcome expectancies, it is of note that affective and cognitive
behavioural consequences need not always be antagonists or independent facets, but
they could also operate in concert.
Downloaded by [Newcastle University] at 01:14 16 July 2012
Acknowledgements
This study has been supported by the German Ministry of Education and Research (BMBF)
within the project ‘Fostering Lifelong Autonomy and Resources in Europe: Behaviour and
Successful Aging: FLARE-BSA’ (Project ID 01ET0801). The first author was funded by the
PhD Programme ‘Multimorbidity in Old Age’ of the Robert Bosch Foundation. The content
is solely the responsibility of the authors.
References
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision
Processes, 50, 179–211.
Bandura, A. (1989). Human agency in social cognitive theory. American Psychologist, 44,
1175–1184.
Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman.
Brownson, R.C., Baker, E.A., Housemann, R.A., Brennan, L.K., & Bacak, S.J. (2001).
Environmental and policy determinants of physical activity in the United States. American
Journal of Public Health, 91, 1995–2003.
Carstensen, L.L., & Mikels, J.A. (2005). At the intersection of emotion and
cognition: Aging and the positivity effect. Current Directions in Psychological Science,
14, 117–121.
Conner, M., Rhodes, R.E., Morris, B., McEachan, R., & Lawton, R. (2011). Changing
exercise through targeting affective or cognitive attitudes. Psychology and Health, 26,
133–149.
Enders, C.K., & Bandalos., D.L. (2001). The relative performance of full Information
maximum likelihood estimation for missing data in structural equation models. Structural
Equation Modeling, 8, 430–457.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behavior: An introduction to
theory and research. Reading, MA: Addison-Wesley.
Freund, A.M., Hennecke, M., & Riediger, M. (2010). Age-related differences in outcome and
process goal focus. European Journal of Developmental Psychology, 7, 198–222.
Gellert, P., Ziegelmann, J.P., Warner, L., & Schwarzer, R. (in press). Physical activity in older
adults – Does a partner make a difference? European Journal of Ageing, Online First
Publication.
Hayes, A.F. (2009). Beyond Baron and Kenny: Statistical mediation analysis in the new
millennium. Communication Monographs, 76, 408–420.
Huy, C., & Schneider, S. (2008). Instrument für die Erfassung der physischen Aktivität bei
Personen im mittleren und höheren Erwachsenenalter. Entwicklung, Prüfung und
Anwendung des ‘‘German-PAQ-50þ’’ [Instrument for the assessment of middle-aged
and older adults’ physical activity: Design, reliability and application of the GermanPAQ-50þ]. Zeitschrift für Gerontologie und Geriatrie, 41, 208–216.
Downloaded by [Newcastle University] at 01:14 16 July 2012
Psychology and Health
827
Joireman, J., Balliet, D., Sprott, D., Spangenberg, E., & Schultz, J. (2008). Consideration of
future consequences, ego-depletion, and self-control: Support for distinguishing between
CFC-Immediate and CFC-Future sub-scales. Personality and Individual Differences, 45,
15–21.
Kendzierski, D., & DeCarlo, K.J. (1991). Physical activity enjoyment scale: Two validation
studies. Journal of Sport and Exercise Psychology, 13, 50–64.
Kraft, P., Rise, J., Sutton, S., & Røsamb, E. (2005). Perceived difficulty in the theory of
planned behaviour: Perceived behavioural control or affective attitude? British Journal of
Social Psychology, 44, 479–496.
Lawton, R., Conner, M., & McEachan, R. (2009). Desire or reason: Predicting health
behaviors from affective and cognitive attitudes. Health Psychology, 28, 56–65.
Lawton, R., Conner, M., & Parker, D. (2007). Beyond cognition: Predicting health risk
behaviors from instrumental and affective beliefs. Health Psychology, 26, 259–267.
Lippke, S., & Ziegelmann, J.P. (2008). Theory-based health behavior change: Developing,
testing and applying theories for evidence-based interventions. Applied Psychology:
International Review, 57, 698–716.
Little, T.D., Cunningham, W.A., Shahar, G., & Widaman, K.F. (2002). To parcel or not to
parcel: Exploring the question, weighing the merits. Structural Equation Modeling, 9,
151–173.
Lowe, R., Eves, F., & Carroll, D. (2002). The influence of affective and instrumental beliefs on
exercise intentions and behavior: A longitudinal analysis. Journal of Applied Social
Psychology, 32, 1241–1252.
Mikels, J.A., Löckenhoff, C.E., Maglio, S.J., Carstensen, L.L., Goldstein, M.K., & Garber, A.
(2010). Following your heart or your head: Focusing on emotions versus information
differentially influences the decisions of younger and older adults. Journal of Experimental
Psychology, 16, 87–95.
Muthén, L.K., & Muthén, B.O. (1998–2007). Mplus user’s guide (5th ed.). Los Angeles, CA:
Muthén & Muthén.
Orbell, S., & Kyriakaki, M. (2008). Temporal framing and persuasion to adopt preventive
health behavior: Moderating effects of individual differences in consideration of future
consequences on sunscreen use. Health Psychology, 27, 770–779.
Prince, S.A., Adamo, K.B., Hamel, M.E., Hardt, J., Gorber, S.C., & Tremblay, M. (2008). A
comparison of direct versus self-report measures for assessing physical activity in adults: A
systematic review. The International Journal of Behavioral Nutrition and Physical Activity,
5, 56.
Prochaska, J.O., & DiClemente, C.C. (1992). In search of how people change. Applications to
addictive behaviors. American Psychologist, 47, 1102–1114.
Richard, R., van der Pligt, J., & de Vries, N. (1996). Anticipated affect and behavioral choice.
Basic and Applied Social Psychology, 18, 111–129.
Rodgers, W.M., & Brawley, L.R. (1996). The influence of outcome expectancy and selfefficacy on the behavioral intentions of novice exercisers. Journal of Applied Social
Psychology, 26, 618–634.
Rotter, J.B. (1954). Social learning and clinical psychology. Englewood Cliffs, NJ: Prentice
Hall.
Scheibe, S., & Carstensen, L.L. (2010). Emotional aging: Recent findings and future trends.
Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 65B,
135–144.
Scholz, U., Sniehotta, F.F., & Schwarzer, R. (2005). Predicting physical exercise in cardiac
rehabilitation: The role of phase-specific self-efficacy beliefs. Journal of Sport and Exercise
Psychology, 27, 135–151.
Schwarzer, R., Luszczynska, A., Ziegelmann, J.P., Scholz, U., & Lippke, S. (2008). Socialcognitive predictors of physical exercise adherence: Three longitudinal studies in
rehabilitation. Health Psychology, 27, S54–S63.
Downloaded by [Newcastle University] at 01:14 16 July 2012
828
P. Gellert et al.
Shamaskin, A.M., Mikels, J.A., & Reed, A.E. (2010). Getting the message across: Age
differences in the positive and negative framing of health care messages. Psychology and
Aging, 25, 746–751.
Strathman, A., Gleicher, F., Boninger, D.S., & Edwards, C.S. (1994). The consideration of
future consequences: Weighing immediate and distant outcomes of behavior. Journal of
Personality and Social Psychology, 66, 742–752.
Trafimow, D., & Sheeran, P. (1998). Some tests of the distinction between cognitive and
affective beliefs. Journal of Experimental Social Psychology, 34, 378.
Trafimow, D., Sheeran, P., Lombardo, B., Finlay, K.A., Brown, J., & Armitage, C.J. (2004).
Affective and cognitive control of persons and behaviours. British Journal of Social
Psychology, 43, 207–224.
van den Berg, H., Manstead, A.S.R., van der Pligt, J., & Wigboldus, D.l.H.J. (2005). The role
of affect in attitudes toward organ donation and donor-relevant decisions. Psychology and
Health, 20, 789–802.
van der Pligt, J., & de Vries, N.K. (1998). Expectancy-value models of health behavior: The
role of salience and anticipated affect. Psychology and Health, 13, 289–305.
Watson, D., Clark, L.A., & Tellegen, A. (1988). Development and validation of brief measures
of positive and negative affect: The PANAS scales. Journal of Personality and Social
Psychology, 54, 1063–1070.
Weinstein, N.D. (2007). Misleading tests of health behavior theories. Annals of Behavioral
Medicine, 33, 1–10.
Williams, D.M. (2010). Outcome expectancy and self-efficacy: Theoretical implications of an
unresolved contradiction. Personality and Social Psychology Review, 14, 417–425.
Williams, D.M., Anderson, E.S., & Winett, R.A. (2005). A review of the outcome expectancy
construct in physical activity research. Annals of Behavioral Medicine, 29, 70–79.
Williams, D.M., & Dunsiger, S. (2007). Suggestions for testing health behavior theories:
Implications for mediator analysis. Annals of Behavioral Medicine, 34, 223.
Ziegelmann, J.P., Lippke, S., & Schwarzer, R. (2006). Subjective residual life expectancy in
health self-regulation. Journals of Gerontology: Psychological Sciences, 61B, 195–201.