The relation between consideration of immediate
3.4. General discussion
53 Table 3.6. Fit indices for confirmatory factor analysis models and structural equation models (Study 2)
χ2ab χ2/dfb RMSEAb SRMRb GFIc NNFIc AICb
Model 1a 186.179 3.448 .122 .089 .841 .863 234.179
Model 1b 164.667 3.050 .112 .090 .857 .870 212.677
Model 2a 140.949 2.659 .101 .079 .875 .896 190.949
Model 2b 121.075 2.284 .088 .081 .890 .910 171.075
Model 3 881.930 3.500 .123 .100 .691 .849 977.930
Model 4 820.130 3.267 .118 .100 .706 .864 918.130
Model 5 718.013 2.861 .107 .093 .733 .876 816.013
Model 6 628.622 2.555 .097 .091 .758 .892 736.622
Model 7 855.732 2.189 .085 .090 .742 .891 1003.732
Model 8 888.521 2.249 .087 .090 .735 .889 1028.521
Note. χ2 = Normal Theory Weighted Least Squares Chi-Square; RMSEA = Root Mean Square Error of Approximation; SRMR = Standardized Root Mean Square Residual; GFI = Goodness-of-Fit Index; NNFI
= Non-Normed Fit Index; AIC = Akaike Information Criterion.
aAll χ2s were significant at p < .001. bLower values indicate better fit. cHigher values indicate better fit.
3.4. General discussion
Rather than establishing absolute differences in time orientation, we investigated, analogous to the double dissociation methodology, whether time orientation for food and exercise differentially predict eating and exercising behavior. Both studies showed that time orientation in the food domain predicted eating behavior, but did not predict exercising behavior. Alternatively, time orientation in the exercise domain predicted exercising behavior, but predicted eating behavior less well.
Whereas several studies found that time orientation is domain-specific (Hardisty &
Weber, 2009; Weatherly et al., 2010), we show that time orientation for food and for exercise are different constructs, although they both belong to the health domain.
Regardless of the positive correlation between the two constructs, these findings indicate that time orientation is not uniform within a domain.
These results provide an explanation for the inconsistent domain-differences in time orientation (Chapman, 2003; Weatherly et al., 2010). Perhaps more important than investigating differences between domains and defining what exactly constitutes a domain (Foxall et al., 2011) is accounting for differences within domains. Because domains (e.g., health, money) encompass many different types of behavior, the mixed findings probably stem from choices researchers make regarding the specific behaviors under scrutiny. We show that type of behavior is a very important indicator, possibly more important than the broader domain to which a behavior belongs.
Chapter 3 Eat now, exercise later
Figure 3.1. Structural equation model of the relations between consideration of immediate and future consequences and self-reported eating and exercising behavior (Study 2). Non-significant coefficients are underscored.
Consistent with earlier studies (Adams, 2012), focusing on future consequences was related to more healthy behavior, whereas focusing on immediate consequences was related to less healthy behavior. Additionally, we showed that eating behavior was predicted by consideration of immediate consequences, whereas exercising behavior was predicted by consideration of future consequences. These findings confirm the distinction between CFC-future and CFC-immediate (Antonides &
Nyhus, in preparation; Joireman et al., 2008). Recently, Joireman, Shaffer, Balliet, and Strathman (2012) investigated relations between consideration of immediate and future consequences and healthy eating and exercising attitudes. They found that both eating and exercising attitudes were predicted by consideration of future consequences, but not by consideration of immediate consequences, which contrasts our findings. The difference may be caused by three dissimilarities between their and our studies. First, they used the original CFC scale instead of domain-specific scales. Second, they added two items to CFC-future, thereby changing the scale’s psychometric characteristics. Finally, they measured attitudes, whereas we measured self-reported behavior. Our findings indicate, however, that using specific scales for food and exercise allows for further specification of the predictive value of consideration of immediate and future consequences across domains.
Eat now, exercise later
3
55 A possible explanation for our findings could be that eating and exercising behavior involve different self-regulation processes. Given the relation between CFC-immediate and self-control (Joireman et al., 2008) and the fact that eating behavior is only predicted by CFC-immediate, we suggest that eating behavior mainly involves short-term self-control. For exercising behavior, however, other processes such as long-term planning are probably more important. Exercising involves choices between exercising or not, whereas eating involves choices between, for example, eating healthy or unhealthy food. Consequently, self-control is relatively more important for eating behavior (e.g., resisting the temptation of eating chocolate), whereas planning is relatively more important for exercising behavior (e.g., planning to go jogging twice a week).
We measured time orientation at a behavior-specific level, an approach that to our knowledge has not been used before but appears to be successful. This, however, does not imply that specification of time orientation should take place to infinity. It is inefficient to measure time orientation at the level of single behaviors, because this requires multiple time orientation measures (which is burdensome for participants) and, in this case, it is easier to directly measure behavior. In order to predict specific behaviors, the level of behavioral domains is probably as specific as necessary, yet not too specific.
One limitation of the present studies is that we are not able to explain why, in Study 2, consideration of the future consequences of exercising behavior does not only predict exercising behavior but also eating behavior. A possible explanation might be that eating and exercising behavior are related, but for lack of identification, it was impossible to capture this relation in models 7 and 8. Another limitation is that it is unclear what participants have in mind when they think about, for example, the future consequences of their eating behavior. Do they think about positive or negative consequences? Do they think about healthy or unhealthy behavior? Future studies might investigate these different possibilities more closely. Furthermore, we used the 12-item CFC, whereas an improved 14-item scale is now available (Joireman et al., 2012). Finally, only self-reported behavior was measured. Even though we believe that this adequately reflects actual behavior, future studies might investigate whether our results generalize to actual behavior. It would also be useful to include relations and trade-offs between eating and exercising behavior in future studies.
Our findings also open up avenues for research into the roles of self-control and planning as two distinct determinants of different types of health behavior. Future studies could more specifically relate self-control and planning to eating and exercising behavior, so as to increase our understanding of the role of different self-regulatory processes in healthy behavior. Investigating how goals for action and inaction (Hepler, Albarracin, McCulloch, & Noguchi, 2012) or differences between temptations and goals (Fishbach & Zhang, 2008) are linked to time orientation and healthy behavior would further advance our understanding of these relations.
Chapter 3 Eat now, exercise later
3
56
3.5. Conclusion
Overall, time orientation for food and for exercise are different constructs that differentially predict eating and exercising behavior. In order to predict behavior, therefore, time orientation is measured best at a behavior-specific level. Additionally, both immediate and future consequences should be taken into account, because these two dimensions of time orientation predict different types of behavior. These insights might successfully be used in order to stimulate healthy eating and exercising behavior.
Acknowledgements
The authors would like to thank Vera Wiedemann for her assistance with the data collection of Study 2.
Eat now, exercise later
3
57