Literature review of studies on physical activity using the TPB
Identifying the determinants of decision-making processes is at the forefront of studies
in physical activity (Buchan et al., 2012) and the TPB is one of the most comprehensive
and widely researched theories used in the prediction and understanding of physical
activity behaviour (Ajzen, 2011; Armitage & Conner, 2001; Downs & Hausenblas,
2005b). The TPB states that the proximal determinant of behaviour is the intention to
perform or not to perform a behaviour and intention is determined by attitude, SN and
PBC towards the behaviour (Jackson, Smith, & Conner, 2003).
One of the early reviews of the literature on the predictive capability of TPB constructs
was conducted by Blue (1995). Blue (1995) reviewed studies using TPB or TRA as their
theoretical framework in analysing exercise behaviour from 1980 to 1995. The review
included 23 studies involving 5,014 participants: 16 studies used TRA and the
remaining 7 used TPB as their framework. In this review, Blue (1995) found that
intention was predictive of an individual’s behaviour. Studies included in this review
measured behaviour between 2 weeks and 2 months after intention was measured and
it was discovered that time frames did not affect the intention-behaviour correlation. The
review also found that the direct effect of attitude on behaviour was weak and the
correlation between SN and intention was positive but not significant. As for the
comparison of TRA and TPB, the review found that the inclusion of PBC significantly
increased the prediction of intention. Conversely, Dzewaltowski, Noble and Shaw
(1990) found that PBC did not add to the prediction of behaviour. However, the results
of two other studies revealed that PBC also added to the explained variance of exercise
behaviour (Kimiecik, 1992; Madden, Ellen & Azjen, 1992). Finally, this review provided
support to the idea that TPB is superior to TRA in predicting exercise behaviour.
A subsequent meta-analysis examining the utility of TRA and TPB for explaining and
predicting exercise behaviour was carried out by Hausenblas, Carron and Mack (1997)
and built on the work of Blue (1995). Manual searches were also made using the
reference list from Blue (1995) and Godin (1993). The principle inclusion criteria in this
review were that the studies focussed on exercise and incorporated at least two
constructs of TRA or TPB in their methods. The results provided strong support for the
validity of TRA and TPB for explaining and predicting exercise behaviour (Hausenblas,
1997). The combined analysis of data showed that intention has a large effect on
exercise behaviour and attitude has a large effect on intention to exercise. SN had only
moderate effects on intention to exercise. These results illustrated that insofar as
exercise behaviour is concerned, the construct of intention is more strongly related to
the construct of attitude than PBC or SN (Hausenblas, 1997). This finding also
demonstrated the superiority of TPB compared to TRA in predicting exercise-related
behaviour (Hausenblas, 1997). It also suggested that in the context of TPB, the more a
person feels the goal is under volitional control, the stronger the relationship between
intention and behaviour. In addition, there is a strong relationship between PBC and
behaviour and attitude when individuals have great commitment to exercise and they
hold favourable beliefs about the activity and believe that they can perform the activity
(Hausenblas et al., 1997).
Bozionelos and Bennett (1999) investigated the predictive utility of TPB in the context of
exercise to identify the possible mediators and moderators of the relationship. The study
involved 114 students from a United Kingdom (UK) university. Questionnaires were
distributed twice with a 3-week interval between. The questionnaires at Time 1 were
standardised and derived from pilot interviews with 20 students whilst questionnaires at
Time 2 only concerned their level of exercise. The constructs measured were attitude,
SN, PBC, behavioural intention, behavioural criterion, perceived barriers, role beliefs
and personal norms, and sex role identity. The findings that SN had little impact on the
prediction of exercise behaviour reinforced the view that significant others have
marginal influence over people’s exercise behaviour. Behaviour was predicted by
attitude, intentions and PBC but not social normative beliefs. Past behaviour was also
found to be a strong predictor of both intention and behaviour. The timing of this study
towards the end of the academic year and overlap with examinations probably
influenced the results. Both attitude and PBC failed to emerge as a significant predictor
of exercise behaviour. The timing of the study, which was conducted prior to year-end
examination, may have caused students to engage in counter-attitudinal behaviour and
they may have failed to judge their control over behaviour during an examination period.
The researchers suggested that the examinations could have exerted pressure on
students to engage in counter-attitudinal behaviour.
A prospective study conducted by Jackson et al. (2003) examined the utility of TPB by
extending the variables to see which would account for additional variance in physical
activity, these included descriptive norms, personal norms, self-identity and past
behaviour along with the three TPB constructs. Two hundred employees of a college in
the north of England completed questionnaires at two time points. Findings showed that
attitude, self-efficacy, PBC and SN accounted for 40.8% of the variance in intention with
attitude and self-efficacy as significant predictors of intention. Contrary to expectation,
descriptive norms were not found to be significant predictors. Jackson et al, (2003)
suggested this was due to descriptive norms being measured using a single-item scale,
which had weakened the correlation between descriptive norms and intention. Past
behaviour contributed an additional 3.4% of the variance in intention. Practically, this
means that promoting physical activity to individuals who had been sedentary
throughout their life would not be particularly useful. As for the prediction of behaviour,
this study found that 22% of the variance of physical activity behaviour was accounted
for by intention, self-efficacy and PBC but descriptive norms did not add to the
prediction of intention. Descriptive norms were a significant predictor of intention for
individuals reporting low to moderate past behaviour. On a practical level, intervention
targeting descriptive norms should be designed for individuals with less experience in
physical activity.
Another meta-analytic study on TRA and TPB in relation to exercise behaviour was
conducted by Downs and Hausenblas (2005a). This study analysed 111 studies
published since 1975 and sought to examine the strength and the predictive utility of
the TPB or TRA constructs and the influence of moderator variables in exercise
behaviour. The analysis discovered a large magnitude of effect for intention-behaviour,
intention-PBC and intention-attitude. The effect was found to be moderate for SN-
behaviour and PBC–behaviour. The effect between attitude-intention was also found to
be larger than PBC-intention and SN-intention. These findings are consistent with other
TPB or TRA reviews and suggest that people’s intention is the strongest determinant of
exercise behaviour and their attitude has the strongest influence on the intention
(Downs & Hausenblas, 2005b). It was also found that intention and PBC predicted
behaviour with intention emerging as the stronger predictor. Attitude was the strongest
determinant of intention but only marginally stronger than PBC in predicting intention.
Consistent with other findings, SN did not predict intention and indicated that significant
others had less impact on exercise intention or behaviour (Downs & Hausenblas,
2005a). An important aspect to note about the above review is that the association
between intention and behaviour was large when the behaviour was measured within 1
month compared to durations of more than a month. This result showed that the
predictive power of intention decreases as the time between intention and behaviour
measured increases (Downs
& Hausenblas, 2005b). The association between self-efficacy and intention was found
to be larger than PBC–intention or perceived barrier-intention.
A meta-analytic review by Hagger et al. (2002) evaluated the predictive validity of TRA
and TPB in physical activity behaviour with the aim of extending the work of
Hausenblas (1997). This study also examined self-efficacy and past behaviour as
additional variables to the TPB and investigated the moderating role of: 1) attitude-
intention relationship, 2) age and 3) time period of assessment of past behaviour.
Online and manual searches were made through journals, reference lists, conference
proceedings and abstracts from Dissertation Abstracts International and Psychological
Abstracts. The researchers examined 72 papers including 31 studies reviewed by
Hausenblas et al. (1997). Their findings corroborated those of Hausenblas (1997)
whereby medium to large effects for the relationship between intention-behaviour,
attitude-intention and PBC-intention were found. A smaller effect was shown for the SN-
intention relationship. Their findings illustrated that SN has a peripheral effect on
intention. This review supported the conclusion that TPB is superior to TRA in the
physical activity domain and PBC is as substantial as attitude as a predictor of
intention. The results also showed that PBC has both direct and indirect effects on
behaviour mediated by intention.
