IBM SPSS Version 19 software was used for all presented statistical analyses (IBM Corp. Released 2010. IBM SPSS Statistics for Windows). Throughout the experimental chapters, preliminary analysis of the data was performed to assess normality, heteroscedasticity and sphericity. For testing normality, the power of the Shapiro-Wilk, Kolmogorov-Smirnov, Lilliefors and Anderson-Darling tests are too low for sample sizes of 30 and below (Razali and Wah, 2011). Although the sample size for Chapter 3 was 331, sample sizes in all other experimental chapters were between 12 and 35. Although typically of insufficient power, the Shapiro-Wilk test is the most appropriate for testing normality of these small sample sizes (Razali and Wah, 2011), therefore for consistency, the Shapiro-Wilk test was used to assess normality of the data throughout this thesis. Of additional note however, so long as ANOVA research designs are balanced, repeated measures ANOVAs are robust to violations of the normal distribution assumption. That is, for sample sizes over 10 to
20, the normality assumption is unnecessary (Norman and Streiner, 2008). Indeed, research designs of Chapters 4, 5 and 6 were each equally balanced and of sufficient sample size to ensure robustness of design.
Repeated measures ANOVAs were used in every experimental chapter. In each instance of repeated measures ANOVA, sphericity of the data was assessed using Mauchley’s test of sphericity. Where one-way ANOVAs or t-tests were used, heterogeneity of the data was assessed using Levene’s test for homogeneity of variances.
Regression analyses were used in Chapter 3 and Chapter 6. In Chapter 3, stepwise multiple regression analyses were used to examine the extent that secondary variables predicted pre- to post-exercise changes in the primary measures of interest. These analyses adhered to Tabachnick and Fidell’s (2007) guideline that a cases-to independent variable (IV) ratio of 40 to 1 is reasonable for stepwise multiple regression analyses. All models presented were examined for instances of significant multicollinearity. In Chapter 6, simple linear regressions were conducted in order to examine the relationships between social interaction time and the measure of intention for each environmental condition.
An alpha level of 0.05 was chosen for indication of statistical significance throughout all of the experimental chapters. In addition to the use of p values, where appropriate, partial eta squared effect size and confidence interval statistics were considered in order to offer greater detail of the magnitude of effects in a manner compatible with the data.
Chapter Three: A Comparison of Four
Typical Green Exercise Environments and
Prediction of Psychological Health
Outcomes
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
A version of this chapter has been published as a research article. The reference for this is:
Rogerson, M., Brown, D.K., Sandercock, G., Wooller, J.J. and Barton, J., 2015. A comparison of four typical green exercise environments and prediction of psychological health outcomes. Perspectives in public health, p.1757913915589845. DOI: 10.1177/1757913915589845
Table 3.1 Thesis map outlining study aims of the experimental chapters
Chapter Aims
3
To investigate potential differences in affective outcomes of running between different typical green exercise environments
To examine the importance of exercise-, individual- and environment-related variables in relation to these outcomes
4
To examine influences of visual exercise environments on directed attention, perceived exertion and time to exhaustion, whilst measuring and controlling the exercise component
5
To identify a method for ensuring consistency of the exercise component between indoor (treadmill) and outdoor (over ground) exercise
To compare psychological outcomes of outdoor versus indoor exercise
6
To compare psychological, social and behavioural intention outcomes of exercise in green outdoors versus indoors settings, whilst controlling the exercise component
3.1 INTRODUCTION
The literature discussed in Chapter 1 suggests that, compared to built or indoor settings, nature environments can enhance wellbeing-related outcomes of exercise. However, it remains less clear as to whether different types of nature environment may be optimal for these effects, as little research has compared how exercising in different outdoor environmental settings influences psychological wellbeing. Barton and Pretty (2010) reported that affective benefits of green exercise activity were enhanced for waterside settings. However, the exercise mode was not consistent between the compared environments, which may have also contributed to the reported findings. Chapter 3 sought to inform the remaining experimental chapters regarding the most appropriate kinds of typical green exercise setting to compare with equivalent built / urban or indoor exercise. It examined different green exercise settings whilst maintaining consistency of exercise mode.
A typical example of green exercise participation is that of running in a park. Acute bouts of exercise facilitate affective improvements (Yeung, 1996, Reed and Ones, 2006) such as mood (Lane and Lovejoy, 2001, Bartholomew et al., 2005) and self- esteem (Ekeland et al., 2005). As outlined in Chapter 1, research articles and systematic reviews report that, compared to exercising either indoors (Teas et al., 2007, Focht, 2009) or in built outdoor environments (Bodin and Hartig, 2003), exercise in nature-based environments can lead to greater psychological benefits (Bowler et al., 2010, Thompson Coon et al., 2011).
The ecological dynamics approach offers explanation for how green exercise improves psychological wellbeing. Compared to synthetic environments, nature environments provide more challenging, complex, varied, and intense affordances
(invitations or possibilities) (Brymer and Davids, 2012, Brymer and Davids, 2014), whereby individuals can experience a broad range of emotions and other psychological feelings such as mindfulness, peace, and calm (Brymer and Davids, 2014). This approach suggests that because laboratories contribute to promoting different affordances than do natural environments, where possible, green exercise research should employ designs which prioritise use of natural environments (Brymer and Davids, 2014). A number of studies have used opportunistic field sampling in order to examine green exercise participation via ecologically valid samples, reporting benefits to self-esteem and mood, across various green exercise activities such as horse riding and walking (Pretty et al., 2007, Barton et al., 2009, Barton and Pretty, 2010). Barton and Pretty’s (2010) meta-analysis of ten field sampling studies found that similarly to exercise per se (Ekkekakis and Petruzzello, 1999, Ekkekakis et al., 2011), duration and intensity of exercise, as well as age, can influence green exercise outcomes. However, other variables such as temperature and motivation for participation were not measured. Improvements in self-esteem (d= 0.46, 95% CI [0.34, 0.59]) and mood (d= 0.54, 95% CI [0.38, 0.69]) resulting from acute bouts of green exercise are significant (Barton and Pretty, 2010). However, it is important to understand how to maximise these benefits in order to better direct the adoption of green exercise activities for psychological health in the wider public domain.
Individual Exercise Environment p1 p3 p2
Figure 3.1 The Four Components (Categories of Variables) of Green Exercise: The Three Physical Components, and the Processes Component (p1 - p3)
A framework for categorising and considering variables in relation to green exercise can be derived from Bandura’s triadic model of reciprocal determinism and the ecological dynamics approach (Bandura, 1986a, Brymer and Davids, 2014). Figure 3.1 shows that green exercise has three physical components (categories of variables): individual, exercise and environment, and a fourth interactive processes component. The processes component comprises psychological and physiological processes within the individual, in relation to the environment or the exercise, or both. Some stimuli and accompanying processes are mutually environment- and exercise-related (p3 area of Figure 3.1). For example, when running in nature, the stimulus of visual optic flow, as perceived by the individual, is a product of exercise- related motion through the environment (Gibson, 1954, Assaiante et al., 1989) (Gibson, 1954, Assaiante et al., 1989). The ecological dynamics approach views
each of the individual, environment and exercise (task) as a system comprising a complex arrangement of factors (e.g. within the individual system: cognitive, affective and physiological states, physical flexibility, limb length). These interacting constraints act as boundaries, within which, behavioural invitations or possibilities (termed ‘affordances’) exist. In this way, the component-related variables referred to within the current study might also be thought of in terms of being constraints.
Exploration of phenomenological experiences of green exercise activities (Brymer and Gray, 2009) and the underpinning cognitive processes of green exercise effects (Harte and Eifert, 1995, Berman et al., 2008, Akers et al., 2012) has provided an initial insight into the processes component. Age and gender appear to influence psychological outcomes, addressing the individual component (Barton and Pretty, 2010). Regarding the exercise component, intensity and duration of green exercise can influence affective outcomes (Barton and Pretty, 2010, Thompson Coon et al., 2011). Examining the environment component, perceived colour of the visual exercise environment effects mood and perceived exertion (Akers et al., 2012). Climatic conditions influence mood and cognition (Keller et al., 2005) but such temporary variables are often not well-accounted for. Furthermore, few research studies have compared exercise outcomes between different nature environments, despite suggestions that the presence of water features within nature-based environments can enhance the effects of green exercise (Barton and Pretty, 2010, White et al., 2010). To better understand green exercise outcomes, the four green exercise components should be studied simultaneously.
3.1.1 Chapter Aims and the Current Study
The modern function of parks often includes usage by organised running groups who participate in set distance runs. This provides an ecologically valid opportunity to control or record relevant variables whilst measuring outcomes of green exercise participation. The current chapter had two aims: to investigate potential differences in affective outcomes of running between different typical green exercise environments; and to examine the importance of exercise-, individual- and environment-related variables in relation to these outcomes. The hypotheses were (i) that environments with greater presence of visible water features would facilitate greatest improvements in affective states via green exercise participation; and (ii) that a number of other measured individual factors, and those related to the environment, the exercise undertaken, and the processes component, would significantly predict psychological improvements.