Annals of Behavioral Medicine

, Volume 47, Issue 2, pp 242–248

Why Do People High in Self-Control Eat More Healthily? Social Cognitions as Mediators

Authors

    • Social Psychology Unit, Department of Social ResearchUniversity of Helsinki
    • National Institute for Health and Welfare
  • Marja Kinnunen
    • National Institute for Health and Welfare
  • Pilvikki Absetz
    • National Institute for Health and Welfare
  • Piia Jallinoja
    • National Institute for Health and Welfare
    • National Consumer Research Centre
Brief Report

DOI: 10.1007/s12160-013-9535-1

Cite this article as:
Hankonen, N., Kinnunen, M., Absetz, P. et al. ann. behav. med. (2014) 47: 242. doi:10.1007/s12160-013-9535-1

Abstract

Background

Trait self-control and social cognitions both predict dietary behaviors, but whether self-control is associated with more beneficial diet-related cognitions, and the effect of self-control on diet mediated by them, has rarely been examined.

Purpose

We hypothesized that the effect of self-control on healthy diet is explained by more proximal diet-related social cognitive factors.

Methods

Altogether, 854 military conscripts (age M = 20) completed questionnaires on trait self-control and social cognitive factors (self-efficacy, outcome expectations, risk perceptions, intentions and planning) upon entering the service and a food frequency questionnaire after 8 weeks.

Results

Trait self-control was associated with more positive cognitions regarding healthy diet. The mediation hypothesis received support for fruit and vegetable but only partially for fast food consumption.

Conclusion

Individuals high in trait self-control eat more healthily because they have higher self-efficacy, more positive taste expectations, stronger intentions and more plans, compared to those low in self-control.

Keywords

Self-controlPersonalitySocial cognitionsThe Health Action Process ApproachMediationDietary behaviors

Introduction

A healthy diet, including plenty of fruit and vegetables and little saturated fat, prevents chronic diseases and mortality [1, 2]. However, young adults [1, 3] eat far fewer fruits and vegetables than is recommended [4]. In behavioral medicine, unhealthy behaviors have been explained by two major lines of research: one investigating stable individual differences in behavioral response tendencies and the other examining social cognitive variables.

An individual difference variable that has attracted increasing attention in health behavior research is self-control. Defined as “the ability to override or change one’s inner responses, as well as to interrupt undesired behavioral tendencies and refrain from acting on them” [5], self-control has been found to be a relatively stable trait [6]. In addition to being correlated with a multitude of positive outcomes in various life domains, such as academic performance and interpersonal skills [5], high self-control has also been found to be positively associated with health behaviors, such as physical activity [7, 8] and fitness [7]. Higher self-control is associated with healthier eating [9], higher fruit and vegetable intake [8, 10, 11], and lower fatty food intake [11].

Social cognition models, on the other hand, suggest that behavior is predicted by factors more amenable to change. One widely used model, the Health Action Process Approach (HAPA) [12] postulates that high self-efficacy (i.e., confidence in one’s ability to perform the healthy behavior), positive outcome expectancies (i.e., belief that the healthy behavior has more advantages than disadvantages), and high perceived health risk predict behavioral intention. HAPA further assumes that the intention is translated into action via specific action planning (i.e., how, when, and where to perform according to intentions) and coping planning (i.e., planned ways to overcome risky situations to avoid relapse). HAPA has been shown to predict a variety of health behaviors [13].

Few studies have examined the influence of trait self-control on social cognitive factors and their interrelationships in predicting health behaviors. Social cognition models generally expect the effects of more distal factors, such as personality or socioeconomic position, to be mediated by social cognitive factors [14]. Indeed, studies have found such evidence for some personality traits [15, 16].

Such mediation might also occur in the case of self-control. Impulsive behavior tendencies might lead to particular types of beliefs across many domains, including eating. As a consequence, individuals high in dispositional self-control might harbor stronger self-efficacy, expect healthy behaviors to have more positive and less negative outcomes, and more readily be aware of and perceive potential health risks. Furthermore, as self-control entails a facet of planfulness [10], those high in trait self-control might more readily form stronger intentions and regulate their eating behavior using careful planning. However, to our knowledge, no previous studies have tested whether the associations between trait self-control and eating behavior are explained by more proximal, diet-related social cognitions. Such investigation would improve our understanding of the working mechanisms between this distal, domain-general personality trait and eating behavior.

This study examines the interrelationships and relative effects of social cognitions and trait self-control on young men’s eating behavior. We hypothesize the following: (1) trait self-control is related to a healthier diet (higher fruit and vegetable consumption and lower fast food consumption), (2) higher self-control is related to more positive social cognitions related to diet, as measured by the HAPA model (i.e., less negative outcome expectancies of taste and inconvenience, more positive outcome expectancies of social outcomes and physical well-being, higher self-efficacy in social situations and fewer emotional barriers, greater perception of health risks, higher intentions for a healthy diet, and more action and coping planning), and (3) the association between trait self-control and dietary behaviors is mediated by those social cognitions. Finally, we will be able to identify which cognitions are the most important mediators of this effect and whether they differ for the two dietary behaviors.

Methods

Study Setting and Participants

The present study employs a single-sample prospective design as a part of the DefenceNutri research project conducted in two military garrisons in Finland. Military service is compulsory for every Finnish male, and almost 80 % complete it [17]. The first measurement took place in the first week of the service (time 1, T1) and the second after 8 weeks (time 2, T2). Participants were fully informed of the procedures and possible risks and gave their informed consent. Participation was voluntary. The study protocol was approved by the ethics committee of the Hospital District of Helsinki and Uusimaa. Altogether, 1,087 male conscripts entered the service in the two research units and participated in the DefenceNutri measurements during the year 2008. Of these, 854 conscripts filled in the T1 questionnaire, and the follow-up measurement of dietary behaviors at T2 was obtained from 679 (retention 79 %).

Measures

Self-control was measured with a 20-item scale adapted [7] from the original 36-item Self-Control Scale [5] (e.g., “I am good at resisting temptation,” “People would describe me as impulsive” (reversed), “I’m able to work effectively toward long-term goals”). The scale showed good internal consistency (Cronbach’s α = .85).

Multiple specific types of each HAPA component were of interest; two or more items were used to measure each specific construct [18]. Measurements of outcome expectations of eating healthy food tapped four different domains of expected consequences: physical well-being outcomes (four items, α = .76), inconvenience outcomes (two items, r = .38), social punishment outcomes (two items, r = .50), and bad taste outcomes (two items, r = .47). The two types of self-efficacy measured were emotional barriers self-efficacy (five items, α = .88) (e.g., perceived certainty of being able to carry out healthy eating intentions even when having problems and worries/feeling tense) and social self-efficacy (two items, r = .69) (e.g., “have to behave in a different way than my friends”). Subtypes of risk perception were perceived risk of weight gain (two items, r = .43) and perceived risk of health problems (e.g., cholesterol level/blood pressure, three items, α = .77). For intention, items “I intend to eat a lot of fruits and vegetables” and “I intend to avoid fatty foods” were used in the fruit and vegetable consumption model and the fast food model, respectively. Items measuring planning were introduced with the stem: “The following questions relate to how you plan your eating. I usually plan beforehand…”, followed by five items for action planning (e.g., “what I eat”, “how often I eat”) (α = .86) and four items for coping planning (e.g., “what to do when something interferes with my plans”) (α = .85).

Food consumption was measured by a 36-item food frequency questionnaire, with the stem “On how many days during the past week did you consume the following food items?” and eight response possibilities ranging from “on 0 days” to “on 7 days.” The questionnaire was based on earlier studies [e.g., 19] and had been modified to fit the target group with the help of focus groups and a food diary study conducted earlier [20]. Two indexes, formed and applied in previous research, were used as outcome variables: the Fruit and Vegetable Index [21] was the mean of two items—the frequency of fruit/berry and of fresh vegetable consumption. The Fast Food Index [3, 22] consisted of the frequency of consumption of typical sources of saturated fats among Finnish young men (French fries, chips, pizza/kebab, hamburgers/hot dogs, meat pies).

Statistical Analyses

Using the Mplus program (Version 6.11), we specified structural equation models to test for the mediation hypotheses. Correlations were investigated as well as a structural path model. From trait self-control, the exogenous variable (i.e., a variable that is not predicted by any prior variable in the model), a regression path was specified on outcome expectations (inconvenience, physical well-being, social punishment, taste), self-efficacy (social, emotional barriers), and perceived risk (health, weight gain), on which, in turn, intention was regressed (as outlined by the HAPA model). The model included paths from intention to action and coping planning, and then from these three on food consumption. Model fit was evaluated using the comparative fit index (CFI), Tucker–Lewis index (TLI), and the root mean square error of approximation (RMSEA) [23]. Missing data were handled by Full Information Maximum Likelihood estimation, considered the least biased method to date [24].

Results

On average, the conscripts reported consuming fruit and vegetables on 3 days and fast food on 1 day weekly at T1. As can be seen in Table 1, higher self-control was prospectively associated with higher fruit and vegetable consumption and lower fast food consumption at T2 (r = .21, p < .001; r = −.19, p < .001). Also, as expected, self-control had significant associations with most HAPA components (with the only exception being perceived risk of weight gain). The structural equation models of self-control predicting fruit and vegetable consumption (Fig. 1, Model 1) and fast food consumption at T2 (Fig. 1, model 2) through HAPA fit the data well (fruit and vegetable consumption: χ2 = 13.82, df = 8, p = .09, CFI = 1.00, TLI = .97, RMSEA = .029; fast food: χ2 = 8.48, df = 8, p = .39, CFI = 1.00, TLI = .98, RMSEA = .008). Self-control was associated with most social cognitions, and the strongest relationships were found with self-efficacy and action planning.
Table 1

Correlations, means, and standard deviations of the study variables

 

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

1. Self-control

1

              

2. OE physical well-being

.09*

1

             

3. OE social punishment

−.17***

−.06

1

            

4. OE bad taste

−.19***

−.25***

.31***

1

           

5. OE inconvenience

−.15***

.15***

.32***

.42***

1

          

6. SE emotional barriers

.41***

.11**

−.16***

−.24***

−.23***

1

         

7. SE social

.29***

.19***

−.23***

−.17***

−.11**

.41***

1

        

8. RP weight gain

−.02

.15***

−.02

−.01

.08*

.04

.02

1

       

9. RP health problem

.25***

.15***

−.11**

−.13***

−.08*

.18***

.24***

.02

1

      

10. FV intention

.26***

.29***

−.13***

−.34***

−.11**

.24***

.25***

.00

.19***

1

     

11. FF intention

.20***

.37***

−.06

−.34***

−.09*

.21***

.15***

.24***

.11**

.41***

1

    

12. Action planning

.34***

.15***

−.02

−.24***

−.15***

.25***

.14***

.08*

.11**

.27***

.30***

1

   

13. Coping planning

.20***

.18***

.00

−.21***

−.10**

.21***

.09**

.13*

.02

.27***

.35***

.53***

1

  

14. FF consumption T2

−.19***

−.07

.08*

.11**

.01

−.08*

−.09*

–.06

.00

−.10**

−.14***

−.13**

−.08*

1

 

15. FV consumption T2

.21***

.13**

−.09*

−.23***

−.13***

.19***

.17***

−.07

.14***

.34***

.15***

.28***

.19***

−.04

1

M (SD)

3.4 (.51)

2.9 (.64)

1.4 (.57)

2.4 (.71)

2.5 (.78)

2.5 (.61)

3.1 (.63)

3.3 (.97)

4.0 (.73)

4.3 (1.48)

3.7 (1.60)

2.3 (.67)

2.0 (.64)

.55 (.57)

2.8 (1.58)

Range min–max

1–5

1–4

1–4

1–4

1–4

1–4

1–4

1–5

1–5

1–7

1–7

1–4

1–4

0–7

0–7

OE outcome expectation, SE self-efficacy, RP risk perception, FV fruit and vegetables, FF fast food, T2 time 2

*p < .05; **p < .01; ***p < .001

https://static-content.springer.com/image/art%3A10.1007%2Fs12160-013-9535-1/MediaObjects/12160_2013_9535_Fig1_HTML.gif
Fig. 1

Predicting fruit and vegetable consumption (model 1) and fast food consumption (model 2) after 8 weeks. SE = self-efficacy, OE = outcome expectations, RP = risk perception, F&V = fruit and vegetable consumption. Correlations between the HAPA variables were estimated but are not shown for legibility

The third hypothesis was only partially supported. The total effects on eating were significant for both behaviors (fruit and vegetable: β = .21, p < .001; fast food: β = −.19, p < .001), and the total indirect effects (i.e., a sum of 29 indirect effects) were significant (fruit and vegetable: β = .14, p < .001; fast food: β = −.03, p = .02). However, the effect of self-control on eating was fully mediated only in the case of fruit and vegetable consumption (direct effect: β = .07, p = .07). For fast food consumption, the direct effect of self-control remained significant (β = −.15, p < .001). Next, we investigated the specific indirect effects of self-control on eating. For fruit and vegetable consumption, significant paths (with significance level p < .01) were the following: self-control → intention → fruit and vegetable consumption (β = .043, p < .001), self-control → bad taste expectations → intention → fruit and vegetable consumption (β = .015, p < .001), self-control → social self-efficacy → intention → fruit and vegetable consumption (β = .010, p < .01), and self-control → action planning → fruit and vegetable consumption (β = .05, p = .001). Weaker, specific indirect paths (with significance level p < .05) for fruit and vegetable consumption were the following: self-control → perceived health risk → intention → fruit and vegetable consumption (β = .006, p = .045), self-control → physical well-being expectation → intention → fruit and vegetable consumption (β = .005, p = .025), self-control → intention → action planning → fruit and vegetable consumption (β = .003, p = .025), and self-control → bad taste expectation → intention → action planning → fruit and vegetable consumption (β = .001, p = .020). For fast food consumption, all indirect effects from self-control were nonsignificant except the path self-control → bad taste expectations → intentions → fast food consumption (β = −.004, p = .040).

Discussion

The purpose of this study was to examine whether the effect of trait self-control on prospective fruit and vegetable as well as fast food consumption could be explained by social cognitive variables and, if so, via which specific mechanisms. Most of the hypotheses were supported: Higher self-control was related to later higher fruit and vegetable consumption and lower fast food consumption, as well as more positive diet-related social cognitions (especially self-efficacy and planning). The proportion of variance in food consumption explained by trait self-control alone, .04, was comparable with earlier studies [25]. The higher the dispositional self-control, the higher the self-efficacy for healthy eating was, even in the face of negative emotions or fatigue and social pressure. Low trait self-control was related to expectations that eating healthy foods causes inconvenience in daily life and trouble in social situations and that healthy foods taste bad. High trait self-control was associated with the expectation that healthy eating results in various aspects of physical well-being. Furthermore, higher trait self-control also was associated with higher perceptions of possible health risks but not higher perceptions of weight gain risk. High self-control was associated with stronger intentions to eat fruit and vegetables and to avoid fat, as well as more plans on how to eat healthy and to cope when the plans are not executed. Finally, the third hypothesis also received partial support: the effect of self-control on fruit and vegetable consumption was mediated by social cognitions, but the effect of self-control on fast food was only partially mediated.

Health behaviors have been found to be influenced by stable personality traits, in particular conscientiousness, of which self-control can be considered an important, especially health behavior-related subfacet [26, 27]. To our knowledge, this is the first study to report full and partial mediation of the effects of trait self-control via social cognitions on consumption of two types of foods. As trait self-control is a general, relatively stable aspect of personality [6], influencing it directly in health interventions is less feasible. However, our results imply that the association of trait self-control and fruit and vegetable consumption can be explained by domain-specific thoughts and skills that can more easily be promoted in interventions.

Why was the effect mediated for fruit and vegetable consumption but only partially for fast food consumption? HAPA model did not gain support in the fast food consumption model because planning was unrelated to behavior, and intention had a very small association with the behavior. This might be due to theoretical, methodological, and empirical reasons. Consuming fast foods and snacks might likely be more impulsive, compared to consuming fruit and vegetables. The HAPA model does not address environmental cues and related impulsive behavior explicitly but instead focuses on the intentional, conscious path to behavior. Alternatively, the small or nonexistent associations of the planning variables with behavior might be due to operationalization of the planning, referring to planning one’s eating in general but not specific types of food. A further possible reason why the effect of self-control on fast food consumption could not be explained by the model may be the relatively low variance in distribution of the outcome: the variance of fast food consumption was three times lower than that of fruit and vegetable consumption.

In addition to total mediation effects, we were able to investigate which social cognitions played the largest role in mediation. On fruit and vegetable consumption, the effect of self-control was largely attributable to stronger self-efficacy to resist peer pressure and taste outcome expectation, which lead to stronger intention and then, in turn, to behavior. The mediation also occurred via intention directly and through action planning. In contrast, only one indirect effect of self-control on fast food consumption achieved statistical significance: Those with lower self-control were found to expect healthier foods to taste bad, leading to lower intention to avoid fatty foods and snacks, which in turn predicts higher fast food consumption. Due to simultaneous measurement of both personality and social cognitions, the direction of causation might be questioned. For example, perhaps people who, due to their taste preferences, tend to perceive unhealthy foods as less delicious and less tempting also then report being better able to resist temptations (i.e., rate themselves higher in self-control). However, self-control is measured as a domain-general variable, and there is evidence of stability of personality and especially self-control across the lifespan [6] .

One limitation of this study is the self-report measurement of the dietary behaviors. However, the measures used were carefully designed and validated, and self-report is the only feasible and cost-effective method known for collecting such a large, representative data set. It should also be noted that the majority of the sample ate fast food 1 day or less per week. It is possible that, at higher consumption levels, the results might be different. The strengths of the study include linking a multitude of types of social cognitions with personality trait self-control in a novel way, the use of a prospective design with regard to the behavioral outcome, and employing state-of-the-art statistical methodology. Also, the sample is fairly representative of young, healthy men.

This is, to our knowledge, the first study to investigate the more proximal cognitive mechanisms between trait self-control and diet. Individuals high in dispositional self-control eat more healthily because they are more self-efficacious, expect healthy foods to taste better, and because they have formed stronger intentions and more action plans to eat healthily, compared to those low in dispositional self-control. These mediation effects were established for fruit and vegetable consumption. Further studies should investigate, by evaluating both reflective and automatic processes in the same model, whether the differing pattern of mediation for fast food consumption can be accounted for by a more impulsive process. Future studies should also investigate moderation effects, as the associations between social cognitive variables and behavior might differ depending on the level of self-control.

Conflict of Interest

The authors have no conflict of interest to disclose.

Copyright information

© The Society of Behavioral Medicine 2013