Oops I did it Again: Emotional Consequences of Food Indulgence
Masterthesis: R. L. (Renée) van Vliet June, 2015
10876286
University of Amsterdam
Faculty of Social and Behavioral Sciences First assessor: N.M.M. (Nina) Bogaerts, MSc
Second assessor: dr. A.C. (Aafje) Brandt
&
Utrecht University Self-Regulation Lab
Supervisors: prof. dr. D. T. D. (Denise) de Ridder and S. (Sosja) Prinsen, MSc
Abstract
Self-regulation failure, the inability to regulate behavior according to one’s goals, in eating behavior contributes to health problems such as obesity. Self-regulation failure is, among others, influenced by self-licensing and emotions. Self-licensing occurs when people seek and construct reasons to justify their non-healthy eating behavior. Emotions influence self-regulation in a direct and indirect (via justifications) way. The current study aimed for a better understanding of the processes around self-regulation failure. An experience sampling study in which N = 138 female participants completed, during one week, eight questionnaires a day was conducted to test the hypotheses. The current study proposed two mechanisms underlying justification-based self-regulation failure in eating behavior. The first mechanism hypothesized that justification use would decrease experienced conflict which, in turn, would increase self-regulation failure or indulgence (i.e. unhealthy eating behavior). However, the results did not demonstrate this model. The second mechanism concerned emotional consequences of indulgence. Emotional experiences after indulgence can influence eating behavior later on and are therefore relevant to study. It was hypothesized that indulgence with a justification would be associated with experiencing positive emotions, while indulgence without a justification would be associated with experiencing negative emotions. Results were not consequent with the expectations, but do provide new insights for the development of health promoting interventions. Because of several limitations, such as the absence of the factor time in the analyses, conclusions of the current study should be interpreted carefully. Nevertheless, a good first step in this area of research was made. Further research is recommended to further clarify the relationship between justifications, conflict, indulgence and emotions.
Keywords: eating behavior; self-regulation failure; self-licensing theory; justifications; conflict; indulgence; emotional consequences; experience sampling method.
Oops I did it Again: Emotional Consequences of Food Indulgence
Everyone can recall a moment in which he or she ate something unhealthy (e.g. chocolate, cookies or crisps) while actually having the goal to eat more healthily. Acting contrary to one’s health goals can be a result of being unable to successfully regulate behavior. This so called ‘self-regulation failure’ contributes to a lot of problems in the domain of eating behavior. Self-regulation failure confers indirectly, via overeating or an increase in the intake of energy-dense, high fat foods, to health problems such as obesity and being overweight. (Hagger, Wood, Stiff, & Chatzisaranris, 2009; World Health Organization, 2015). In 2014, 39% of the adult world population were overweight and 13% were obese. Being overweight or obese is a major risk factor for cardiovascular diseases (leading cause of death in 2012), diabetes, musculoskeletal disorders and some cancers (World Health Organization, 2015). Effective health improving interventions are necessary in order to be able to prevent and treat overweight and obesity. An understanding of the process of self-regulation failure can contribute to the development of such effective health improving interventions.
The term self-regulation, as introduced above, is used to describe all behavior and processes, both conscious and automatic, people use to achieve goals (Mann, de Ridder, & Fujita, 2013). Goals can be defined as ‘mental representations of a desired outcome’ as stated by Mann et al. (2013). Self-regulation in the domain of eating behavior is directed at weight goals (Johnson, Pratt, & Wardle, 2012) or at other health goals such as following a healthier diet. After one has set goals, the process of goal striving starts. Goal striving involves all behavior that contributes to successful achievement of goals (Mann et al., 2013). For example planning and shielding goals from competing goals, temptations or distractions (Mann et al., 2013). An example of goal shielding of a healthy diet goal can be eating an apple to prevent oneself to give in to the temptation of eating a cookie.
During the process of goal striving one can experience conflicts (De Witt Huberts, Evers, & De Ridder, 2014). Conflict is defined by Hofmann, Baumeister, Förster, & Vohs (2012) as “a signal that there is a discrepancy or incompatibility between the person’s present desire and other higher order goals that needs to be resolved” (p. 1319). An example of a conflict in self-regulation of eating behavior is experiencing the urge to indulge into eating a cookie (short term goal) while actually being on a diet (higher order long term goal). These conflicts take place before actually eating something. The experience of conflicts can be explained by a dual-process model approach of self-regulation. Dual-process models state that two different systems compete for control over behavior: the impulsive versus the reflective system (Strack, & Deutch, 2004; Hofmann, Friese, & Wiers, 2008; De Witt Huberts et al., 2014). The impulsive system is associated with automatic and affective processes, short term goals and behavior towards rewarding stimuli (such as the joy of eating a cookie) (Becker, Jostmann, Wiers, & Holland, 2015). The reflective system is associated with rational processes and long term goals (such as dieting to lose weight). Consequently, it can be interpreted that self-regulation failure is frequently caused by the impulsive system (De Witt Huberts et al., 2014). However, lately this interpretation is being challenged since there is growing evidence that both systems (impulsive and reflective) can cause both self-regulation success and failure (De Witt Huberts, Evers, & De Ridder, 2012a). This novel view on self-regulation failure provides new insights on the different ways that cause people to not behave according to their intentions.
A reflective route towards self-regulation failure is explained by the self-licensing theory (De Witt Huberts et al., 2012a). Self-licensing refers to seeking and construing excuses for one’s discrepant behavior before actual enactment (De Witt Huberts et al., 2014). Self-licensing is originally studied in the context of moral behavior. Merrit, Effron and Monin (2010) showed that past moral behavior can serve as a justification to act in a morally equivocal way without experiencing feelings of dissonance. A comparable principle is applicable to
self-licensing in the context of eating behavior. However, it is noteworthy that the justifications people use do not have to be within the same domain as the behavior that is being justified for. Recent studies showed that past (healthy) behavior can serve as a justification to act unhealthy, for example, allowing yourself to eat an unhealthy snack because you had a really tough day (Taylor, Webb, & Sheeran, 2014). Future (healthy) behavior or intentions of future (healthy) behavior can serve as a justification as well, like eating a cookie because you have planned to go to the gym later that day. Having a justification or reason appears to be more important than the content or quality of the reason (Shafir, Simonson, & Tverskey, 1993). So, any reason seems to be able to serve as a justification to indulge, regardless of its content (Shafir, Simonson, & Tverskey, 1993). De Witt Huberts et al. (2014), Taylor et al. (2014) and Verhoeven, Adriaanse, De Vet, Fennis, & De Ridder (2014) studied justifications use specifically in eating behavior. They discovered that some justifications are used regularly and determined different categories of justifications, for example negative emotional events, the need to gain energy or social pressure.
De Witt Huberts et al. (2014) suggests that a justification resolves the conflict that occurs while experiencing a temptation. Because people can imagine the consequences of their behavior in advance, they can experience feelings of dissonance (i.e. experiencing conflict) when imagining giving in to a temptation. Justifications are then used to resolve these feelings of dissonance by allowing violation of the long term goal (De Witt Huberts et al., 2014). Accordingly, the following mechanism can be implied: justification use results in less conflict which is then likely to result in an increase of indulgence. To date, literature assumes the existence of this underlying mechanism of justification-based self-regulation failure (De Witt Huberts et al., 2014) however, it has not been empirically demonstrated. The current study therefore tried to demonstrate this mechanism by testing a mediation model (Figure 1) including justification-use, experienced conflict and indulgence. An experience sampling method study
that followed participants for one week was designed to test this model. It was primarily hypothesized that conflict would mediate the relation between justifications and indulgence. The relation between justifications, conflict and indulgence was hypothesized to be as following: justifications were negatively related to conflict and positively related to indulgence and conflict was positively related to indulgence.
Emotions also play an important role in justification-based self-regulation failure, either via a rational and indirect route or via an impulsive and direct route. The rational and indirect route refers to a route in which negative emotions or emotional events serve as a justification which leads to an increase in indulgence (De Witt Huberts et al., 2014; Verhoeven et al., 2014). For example, eating as a way to deal with sad feelings or to comfort yourself (Verhoeven et al., 2014). De Witt Huberts, Evers, & de Ridder (2012b) studied the role of emotions in self-licensing theory. Participants only consumed more forbidden food (unhealthy snacks) when they were highly aware of their negative emotional state, not when they were unaware of their negative emotional state. This indicates that negative emotions can, indeed, serve as a justification to indulge and that this is a rational process (De Witt Huberts et al., 2012b). Furthermore, there is mixed evidence for the existence of a direct influence from emotions on food intake, not via justifications. This direct route is described by the term ‘emotional eating’. Emotional eating is a well-known phenomenon in which people tend to increase their unhealthy food intake when feeling emotional (Evers, de Ridder, & Adriaanse, 2009). This seemed to be
a more impulsive process, rather than a rational process. While the up-to-date literature shows relevant information about emotional experiences and influences before indulging, little is known about the emotional consequences of self-regulation failure. Emotional consequences of self-regulation failure, i.e. the emotions that are experienced after indulgence, are relevant to study since they can influence eating behavior later on. For example, if someone is very frustrated after indulging to a temptation, that person can indulge a second time because of his or her negative feelings. Either because he or she is an emotional eater, or because he or she uses his or her negative feelings as a justification to eat.
Ramanathan and Williams (2007) examined the emotional consequences of indulgence. Results showed that participants experience a mixture of both positive emotions, such as pleasure and pride, and negative emotions, such as frustration and guilt, after indulgence. However, Ramanathan and Williams (2007) did not study the emotional consequences of indulgence in combination with justifications use. De Witt Huberts (2014) implies that justification use resolves the inner conflict by allowing oneself to eat. It is therefore likely that people will experience little or none negative emotions (such as regret) after indulgence with having a justification because they allowed themselves to eat. Thus, emotional consequences of self-regulation failure were expected to be different when one did or did not use a justification to eat. Assessing these differences will contribute to the understanding of the process of self-regulation failure in eating behavior. Additionally, a better understanding of the emotional consequences could contribute to the development of effective health promoting interventions, for example by knowing if interventions should focus on (specific) anticipated emotions. Consequently, the current paper tried to assess the emotional consequences of self-regulation failure in eating behavior, with and without justification use. A moderation model (Figure 2) including indulgence, justifications use and emotions was tested. It was primarily hypothesized that justification-use would moderate the relationship between indulgence and emotions.
Second, it was hypothesized that one would experience high levels of positive emotions (comfort, pride, satisfaction and confidence) when indulging with a justification, while low levels of positive emotions were expected to be experienced when indulgence without a justification happened. High levels of negative emotions (frustration, regret, stress and guilt) were hypothesized to be experienced when indulging took place without a justification, while low levels of negative emotions were expected to be experienced when indulgence with a justification occurred.
Method Participants
The design of the current study was derived from the design of an experience sampling study conducted by Hofmann et al. (2012). Hofmann’s sample size was N = 208, therefore and in order to allow for dropouts and non-response the current study aimed to recruit at least 250 participants. Eventually 275 Dutch female citizens were selected to participate, of which 197 actually started the experience sampling test week. In total, 138 participants completed their participation in this study. Selection took place based on gender, age and diet or weight concerns. Women aged between 18 and 65 years who had any concerns about their diet or weight were included to participate. Women with an eating disorder were excluded from participation.
Recruitment of the participants took place using ads published on several websites of large communities (church, sport clubs and proefbunny.nl), and on Facebook. Next to internet-based recruitment techniques, ads were spread in supermarkets (Albert Heijn, Jumbo) in Houten, Nieuwegein, Utrecht and Amsterdam and in faculties of Utrecht University and the University of Amsterdam. Subjects had the possibility to sign up via internet or email. All participants received an e-mail following their application. This e-mail consisted of information about the procedure of the research and included a link to the first questionnaire.
Out of the 197 participants who started the experience sampling test week, 138 completed enough questionnaires in order to be able to use their data. Participants completed enough questionnaires when they responded to at least six questionnaires a day. A drop-out analysis was performed to assess possible differences between the two groups (the drop-out group and the non-dropout group). No difference between the two groups were found on the variables age, hours of working, degree of eating or weight goal, eating or weight goal importance, Body Mass Index (BMI: before participation), desired weight, importance of reaching the desired weight, the tendency to use justifications, education levels, whether the participant was a student or not and household composition. All p values were bigger than p > .18. An independent samples t test was performed to assess whether trait self-control differed for participants (n = 138) who completed enough questionnaires compared to participants (n = 57) who dropped out. Preliminary data screening indicated that scores on self-control were normal distributed. The assumption of homogeneity of variance was assessed by the Levene test, F = .733, p = .393; this indicated no significant violation of the equal variance assumption; therefore, the pooled variances version of the t test was used. The mean level of self-control differed significantly, t(193) = -3.35, p = .001, two-tailed. Mean self-control for the non-drop-out group (M = 3.17, SD = .58) was .32 higher than mean self-control in the drop-non-drop-out group (M = 2.85, SD = .65). The effect size, as indexed by η2, was .05; this is a medium effect. The 95% CI for the difference between sample means, M1 – M2, had a lower bound of -.504 and an upper
bound of -.131.
Descriptive information of the sample of 138 women is summarized in table 1. Eating goal refers to the degree to which participants have the goal to eat healthier, weight goal refers to the degree to which participants have the goal to lose weight. Table 2 shows the distribution of the participants into the four weight categories (underweight, healthy weight, overweight and obesity). BMI, which was measured before and after participation to the test week, was used to
determine the percentages of participants in each category. The participants lived in 43 different Dutch cities or villages. 56.5% of the participants were students, studying in 39 different fields of study (28 participants studied psychology). The other 43.5% of the participants were full- or part-time employed (41.4%) or unemployed (2.1%).
Table 1
Descriptive statistics of the sample (N = 138)
M SD
Age 27.7 10.27
Eating goal 74.07 20.91
Eating goal importance 72.35 20.12
Weight goal 51.12 20.91
Weight goal importance 52.51 29.68
Note. Age could vary between 18 – 65 years, eating and weight goal and eating and weight goal importance could vary between 0 – 100.
Table 2
Percentage of the participants with underweight, healthy weight, overweight and obesity Before participation After participation
Underweight 4.3 2.9
Healthy weight 71.1 77.5
Overweight 21.8 17.5
Obese 2.8 2.1
Note. Underweight (BMI < 18.5), healthy weight (BMI 18.5 – 24.99), overweight (BMI ≥ 25 – 29.99), obesity (BMI ≥ 30) (World Health Organization, 2006).
Materials
A complete overview of the content of all questionnaires is described in Appendix A. Only the materials relevant for the current study are described below. The variables used in the current study are ‘justifications’, ‘conflict’, ‘total indulgence’ (independent variables), ‘indulgence’ and ‘emotions’ (dependent variables). Descriptive information of these variables is summarized in Appendix B (table 3).
Justifications. Justifications use was assessed by asking participants to choose from a
35 items list of possible justifications that were applicable to them in the past two hours. This list consisted of 27 items with justifications such as ‘I was tired’ or ‘I was in a special occasion’, six ‘filler’ items and one item stating ‘none of the above’. The list was created based on research about categories of justifications in eating behavior by Verhoeven et al. (2014), Taylor et al. (2014) and De Witt Huberts et al. (2014). The current study assumes that when participants chose an item of the list, that he or she indeed used this item as a justification to indulge. Participants could indicate for every item separately if that item was applicable to them in the past two hours or not, so participants could have had multiple justifications. Participants score on justification use was calculated by taking a sum score of the first 27 items of the list. Thus, scores could vary between 0 (the participant did not use a justification) to 27 (the participant used 27 justifications).
Conflict. Conflict was measured with one item that questioned the degree to which
indulging to the previously reported temptation would be in conflict with the goal to eat healthier. This item had to be answered using a VAS, scores could vary between 0 (indicating a low degree of conflict) and 100 (indicating a high degree of conflict).
Total Indulgence. The variable ‘total indulgence’ was defined as the total amount of
food temptations participants gave in to in the two hours previously to the questionnaire. The variable was computed from two items: scores on one item that measured the degree to which
a participant gave in to the previously reported temptation were added with scores on one item that questioned how many other temptations were given in to. The first item had to be answered using a VAS with scores varying from 0 (low degree of indulgence) to 100 (high degree of indulgence). If the score on this item was equal or higher than one, the computation would add ‘one’ to the score of the latter item. This latter item had to be answered by filling in the number of temptations participants gave in to.
Indulgence. The variable ‘indulgence’ was defined as the degree to which the
participants gave in to the previously reported temptation. One item was used that had to be answered using a VAS with scores varying from 0 (low degree of indulgence) to 100 (high degree of indulgence).
Emotions. The variable ‘emotions’ refers to the level of positive and negative emotions
people experience after indulgence. Positive emotional consequences were measured using four questions asking to which degree participants currently felt satisfied, comfortable, proud, and confident. Negative emotional consequences were measured using four questions asking to which degree participants currently felt frustrated, stressed, regretful and guilty. These emotions were derived from the study about mixed emotions after indulgence by Ramanathan & Williams (2007). Participants answered the questions using a VAS, scores of each emotion could vary between 0 (low degree of experienced emotion) to 100 (high degree of experienced emotion). Procedure
Subjects who signed up to participate received an intake questionnaire to determine if they could be included to participate. The intake questionnaire also included a demographic data sheet with questions about gender, age, hometown, level of education, occupation, study, name of the school/university, profession, if non-student participants were part-time/full-time workers and household composition. Selected participants received instructions about the procedure of the study through e-mail. This e-mail contained an instruction manual, an
information video about the content, goal and procedure of the study and a link to fill in the informed consent. Information about the confidentiality of the study and the compensation of their participation was presented in the informed consent. In order to prevent participants from being biased by information about the content and the goal of the study, a cover story was used which implied that the current study looked into food temptations in daily life. This way participants thought that this study was focused on receiving information about the amount of food temptations one experiences in which surroundings.
The experience sampling period consisted of seven consecutive days in which participants received eight questionnaires a day. Participants received text-messages on their smartphone that included a link to the online survey. They received the eight questionnaires daily over a time window of 12 to 14 hours: one questionnaire per two hours with two hours between every questionnaire. Participants could choose if they wanted to receive the questionnaires between 8 a.m. and 10 p.m., 9 a.m. and 10 p.m. or 10 a.m. and 10 p.m.. The questionnaires closed automatically after 20 minutes after receiving the text-message, so participants had to answer the questionnaire within these 20 minutes. The last questionnaire of each day was send at 10 p.m. or 10.15 p.m. and participants were instructed to fill in this last questionnaire just before they went to sleep. Therefore, this questionnaire did not close, it was accessible during the whole night. Participants received the questionnaires for an extra (eight) day if they responded to five or less questionnaires on one day during the experience sampling test week. The data obtained from this extra day was added to the existing data, it did not replace the incomplete data of the day that the response of the participant was not enough.
Participants received a final questionnaire two days after participation. This questionnaire aimed to assess how participants experienced their participation.
Participants from who the data was of use (i.e. who responded to enough questionnaires) received a compensations of €20.
Statistical analysis
A mediation and eight different moderation models were hypothesized. The dependent variable in the mediation model was ‘indulgence’, the independent variable was ‘justifications’ and the proposed mediating variable was ‘conflict’. A series of three regression analyses was performed to test whether this model was significant. The variable ‘indulgence’ was used in the mediation model instead of the variable ‘total indulgence’ because the independent variable in the mediation model (‘conflict’) was only questioned with regard to the previously reported temptation, and not with regard to all experienced temptations.
The dependent variable in each of the moderation models was the degree to which one had experienced one of the eight measured emotions, so the variable ‘emotions’. The independent variable was ‘total indulgence’ and the moderating variable was ‘justifications’. Eight regression analysis including an interaction term, were performed to assess the significance of the eight different moderation models. The variable ‘total indulgence’ was used instead of the variable ‘indulgence’ because total indulgence concerned all indulgence of the two hours previously to the questionnaire and the other two variables (‘justifications’ and ‘emotions’) were also questioned with regard to the two hours previously to the questionnaire. Therefore, ‘total indulgence’ fits best in this moderation model.
Results
To test all hypotheses two different analyses were conducted, a mediation analysis and several moderation analyses.
Mediation analysis
A mediation analysis was performed using the Baron and Kenny (1986) causal steps approach. The initial causal variable was justification use, a continuous measure; the outcome variable was indulgence, continuously measured; and the proposed mediating variable was conflict, also continuously measured. Refer to Figure 1 for the path diagram that corresponds
to this mediation hypothesis. Preliminary data screening suggested that there were no serious violations of the assumption of normality. However, the assumption of linearity appeared to be violated. All coefficients reported here are unstandardized, unless otherwise noted; α = .05 two-tailed is the criterion for statistical significance.
The total effect of justifications on indulgence was significant, c = 2.105, t(1618) = -5.589, p <.001; each increase of 1 justification predicted approximately a 2-point decrease in indulgence. Justification was significantly predictive of the hypothesized mediating variable, conflict; a = .886, t(1618) = 3.334, p < .01. When controlling for justifications, conflict was not significantly predictive of indulgence, b = -.002, t(1618) = -.055, p = .956. The estimated direct effect of justifications on indulgence, controlling for conflict, was c’ = -2.104, t(1618) = -5.564, p <.001.
Indulgence was predicted poorly by justifications and conflict, with adjusted R2 = .018 and F(2, 1618) = 15.610, p < .001.
The indirect effect, ab, was -.0017. This was judged to be not statistically significant using the Sobel (1982) test, z = -.057, p = .954.
Several criteria can be used to judge the significance of the indirect path. In this case, only the a coefficient and the direct path from justifications to indulgence (c’) were statistically significant, the b coefficient and the Sobel test for the ab product was not significant. By these criteria, the indirect effect of justifications on indulgence through conflict was not statistically significant. Therefore, it can be concluded that the effects of justifications on indulgence were not mediated by conflict.
The diagrams in Figure 2 shows the unstandardized and standardized path coefficients for this mediation analysis.
Figure 3. Mediation model with unstandardized (left) and standardized (right) path coefficients. *p < .05, ** p < .01, *** p < .001, all two-tailed.
Moderation analyses
Regression analyses were performed to assess whether indulgence interacts with justification use to predict the level of different experienced emotions. In total eight regression analyses were conducted. A moderation model was tested for each of the eight emotions (satisfaction, comfort, pride, confidence, frustration, stress, regret and guilt). Total indulgence ranged from 0 to 6 and justification use ranged from 0 to 18. Preliminary data screening did not suggest problems with assumptions of normality expect for scores on the emotions, confidence, regret, guilt, stress and frustration. The assumption of linearity appeared to be violated. Prior to forming a product term to represent an interaction between justifications and total indulgence, scores on these variables were centered by subtracting the sample mean. Out of eight regression analyses, only three resulted to be (marginal) significant. The interaction term Total Indulgence × Justifications appeared to be insignificant (p > .07) in the models concerning the emotions satisfaction, comfort, frustration, regret and guilt. Therefore, these results will not be further outlined. Results concerning significant main effects of indulgence and results of the models that appeared to be significant and are outlined below.
Satisfaction and comfort. There were significant effects for indulgence on satisfaction,
b = -4.2, t (1616) = -4.618, p < .001, sr2 = .0129, and for indulgence on comfort, b = -3.062, t (1616) = -3.408, p < .001, sr2 = .0071.
Pride. The first significant regression included indulgence, total justifications and a
total-indulgence-by-justifications interaction term as predictors of levels of pride. The overall regression was statistically significant, R = .150, R2 = .022, adjusted R2 = .021, F(3, 1616) = 12.396, p < .001. There was a significant Total Indulgence × Justifications interaction, b = -.954, t (1616) = -2.143, p < .05, sr2 = .0028. There were also significant effects for justifications, b = .603, t (1616) = 2.232, p < .05, sr2 = .0030, and indulgence, b = -3.920, t (1616) = -3.520,
p < .001, sr2 = .0076. Because the interaction term was statistically significant, the interaction was retained in the model.
To visualize the nature of the total-indulgence-by-justifications interaction, examine the graph of the regression prediction lines for 0, 9 and 18 justifications1 in Figure 4. Increase in justification use was associated with experiencing higher levels of pride, but only if total indulgence was low. Higher levels of total indulgence were associated with lower levels of experienced pride; the statistically significant interaction between indulgence and justifications indicated that this association between indulgence and pride was stronger when participant had more justifications. However, a causal inference cannot be made from nonexperimental data.
1 Justifications ranged from 0 – 18. The two extreme values (0 and 18) and the median value (9) were chosen to easily
Figure 4. Predicted levels of experienced pride, total indulgence and justification use.
Confidence. The second significant regression included indulgence, total justifications
and a total-indulgence-by-justifications interaction term as predictors of levels of confidence. The overall regression was statistically significant, R = .125, R2 = .016, adjusted R2 = .014, F(3, 1616) = 8.546, p < .001. There was a significant Total Indulgence × Justifications interaction, b = -.780, t (1616) = -2.128, p < .05, sr2 = .0028. There was also significant effect for indulgence, b = -3.075, t (1616) = -3.354, p < .05, sr2 = .0068. However, there appeared to be no significant effect for justifications, b = -.107, t (1616) = -.479, p = .632, sr2 = .0069. Because the interaction term was statistically significant, the interaction was retained in the model.
To visualize the nature of the total-indulgence-by-justifications interaction, examine the graph of the regression prediction lines for 0, 9 and 18 justifications in Figure 5. Increase in justification use was associated with experiencing lower levels of confidence, however this relation was conversely in low levels of total indulgence. Overall, higher levels of total indulgence were associated with lower levels of experienced confidence; the statistically significant interaction between indulgence and justifications indicated that this association
between indulgence and confidence was stronger when participant had more justifications. However, a causal inference cannot be made from nonexperimental data.
Figure 5. Predicted levels of experienced confidence, total indulgence and justification use.
Negative Emotions
Frustration, regret and guilt. There were significant effects for indulgence on
frustration, b = 2.905, t (1616) = 2.915, p < .01, sr2 = .0049, for indulgence on regret, b = -9.126, t (1616) = -8.913, p < .001, sr2 = .0462, and for indulgence on guilt, b = 7.604, t (1616)
= 7.566, p < .001, sr2 = .0335.
Stress. The third significant regression included total indulgence, justifications and a
total-indulgence-by-justifications interaction term as predictors of levels of stress. The overall regression was statistically significant, R = .302, R2 = .091, adjusted R2 = .089, F(3, 1616) = 53.865, p < .001. There was a significant Total Indulgence × Justifications interaction, b = -.821, t (1616) = -1.985, p < .05, sr2 = .0022. There was also a significant effect for justifications, b = 2.989, t (1616) = 11.919, p < .001, sr2 = .0801, but not a significant effect for indulgence, b = 1.806, t (1616) = 1.746, p = .081, sr2 = .0016. Because the interaction term was statistically significant, the interaction was retained in the model.
To visualize the nature of the total-indulgence-by-justifications interaction, examine the graph of the regression prediction lines for 0, 9 and 18 justifications in Figure 6. Increase in justification use was associated with experiencing higher levels of stress, but not in high levels of total indulgence. High levels of total indulgence were associated with lower levels of experienced stress as justification use increased. The statistically significant interaction between indulgence and justifications indicated that this association between indulgence and stress was stronger when participants had more justifications. However, a causal inference cannot be made from nonexperimental data.
Figure 6. Predicted levels of experienced stress, total indulgence and justification use.
Discussion
The current paper tried to demonstrate two mechanisms underlying justification-based self-regulation failure in eating behavior. The first hypothesized mechanism implied that, when experiencing a food temptation, having a justification to eat would decrease inner conflict experiences which would lead to an increase in indulgence. However, the results of the current study do not indicate this relationship. The second hypothesized mechanism concerned emotional consequences of justification-based self-regulation failure (i.e. indulgence in a food
tempation). It was expected that having a justification would influence the relationship between indulgence and the emotional experiences after indulgence. These emotional consequences were expected to be positive if one had a justification to indulge, and negative if one indulged without having a justification. However, the results do not provide support for these expectations.
The results of the current study do not support the first hypothesized mechanism. Justification use appears to be associated with a decrease in indulgence, so having a justification to eat is associated with eating (indulging) less. Furthermore, results indicated that the experience of conflict does not influence the relationship between justification use and indulgence. Contrary to what was expected, results show that justification use is associated with an increase in conflict experiences. These findings can possibly be explained by the counteractive control theory (Kroese, Evers, & De Ridder, 2009). The counteractive control theory states that temptations trigger goal-directed behavior. So the experience of a food temptation activates the long term diet (health) goal which causes a decrease in indulgence (Kroese et al., 2009). This might have happened in the current study as well: having a food temptation made participants aware of their long term health goal which, even when participants had a justification to eat, resulted in less indulgence. The experience of having a justification, so having a reason to eat, while also being highly aware of your health goal is likely to cause feelings of conflict: this might explain the finding that justifications were associated with an increase in conflict experiences.
Results regarding the second hypothesized mechanism, about positive emotional consequences of food indulgence, show that indulgence (without the influence of justifications) is associated with experiencing low levels of positive emotions (satisfaction, comfort, pride and confidence). This is in line with the expectations. However, contrary to what was expected, having a justification does not influence the relationship between indulgence and the experience
of the positive emotions satisfaction and comfort. Furthermore, results show that justifications do influence the relationship between indulgence and the positive emotions pride and confidence. It appeared that high levels of indulgence were associated with low levels of pride and confidence while having a justification. Thus, after indulgence one experiences low levels of positive emotions (pride and confidence), even while being allowed to eat. It can be concluded that, regardless of having a justification or not, one will experience low levels of positive emotions after indulging in a food temptation. However, this conclusion should be interpreted carefully because of violation of the assumption of linearity and because causal inference cannot be made from the current nonexperimental data.
Regarding negative emotional consequences of indulgence, results report that higher levels of indulgence are associated with experiencing higher levels of the negative emotions frustration and guilt but, contrary to what was expected, low levels of regret. Justification use does not influence this relationship. In line with the expectations, results indicate that if one indulges while having a justifications, low levels of stress are experienced. Interpreting these results indicate that one will, regardless of having a justification or not, experience high levels of the negative emotions frustration and guilt. Furthermore, regret is experienced in low levels after indulgence and is therefore not an emotional consequences of indulgence. Moreover, stress appears to be an emotional consequence of indulgence, but not when one indulged while having a justification. Again, conclusions should be interpreted carefully because the assumption of linearity was violated and because causal inferences cannot be made from the current nonexperimental data.
Altogether, these results do not provide sufficient support for the second hypothesized mechanism concerning emotional consequences of food indulgence. As expected, one experiences low levels of positive emotions and high levels of negative emotions after indulgence. However, this relationship is, for the emotions satisfaction, comfort, regret,
frustration and guilt, not influenced by having a justification or not. Concerning the positive emotions pride and confidence, justification use did influence the relationship between indulgence and the experience of these emotions. However, this influence was contrary to what was expected because having a justification was not associated with experiencing higher levels of these positive emotions. It remains unclear why justification use has little to no influence on emotional experiences after indulgence. Further research is necessary to further clarify (possible) influences of justification use. It is proposed to, in future research, differentiate between immediate and delayed emotional consequences of indulgence. Ramanathan et al. (2007) found differences in emotional consequences of indulgence immediately after the indulgence and a little while later. It seems possible that justification use does, indeed, influences emotional experiences after indulgence, but that these influence only last for a short period. It could be that, for example, indulgence with a justification would result in positive feelings directly after the indulgence, because he/she was allowed to eat. But, after a while, one realizes that the indulgence was contrary to his or her health goals which results in negative emotional experiences. Future research could examine this suggested pathway of the influence of justification use on emotional experiences.
The current study had several limitations. Firstly, participant’s data could have been influenced by the fact that they were monitored for a longer period. This could have made participants more salient of their own behavior which might have influenced their answers to the questions. It is likely that participants became, as a result of participating in the study, more aware of their eating behavior and health goal which might have led to healthier behavior. Second, participating in the current study was very intense, which might have influenced the data as well. It is possible that participants adjusted their answers so it would be less effortful to respond to the questionnaires, for example by filling in that one did not experience a food temptation while he/she actually did experience one. By doing so, participants had to answer to
fewer questions, which causes their answers to be of less informational value. Nonetheless, these problems are difficult to tackle and are likely to happen in every experience sampling study. Third, the sample was very young and more than half of the participants were students, although they came from very different disciplines, a lot of psychology student were among them. Psychology students often have sufficient knowledge about methodologies and could therefore have (unconsciously) influenced the data. Fourth, another important limitation is that the analyses of the current study did not take the factor ‘time’ into account. It is likely that different effects will appear when controlling for time instead of using mean scores of the whole test week. As described before, different effects could appear when controlling for time as a result of differences in short term and long term effects of, for example, justifications on emotional consequences.
Nevertheless, this study does bring forward some promising results. Several strengths of this study have contributed to these findings. The current study had a strong and structured design because of its frequent and consistent measurement moments. Participants were questioned every two hours a day for one week. Because participants were questioned that often, it is likely that their answers are consistent with real life. Furthermore, the study had clear and good instructions for the participants, which resulted in a high response rate of the participants. Moreover, this study is innovative and the first in its kind: it is the first nonexperimental experience sampling study on the topic of self-licensing in eating behavior.
Results show that indulgence with justification use results in experiencing low levels of feelings of stress. Moreover, results of the current study indicate which emotions are experienced as a result of indulgence without justifications, namely high levels of negative emotions (frustration and guilt) and low levels of positive emotions (satisfaction, pride, confidence and comfort). These results can be used in the development of health promoting interventions that focus on healthy eating. Making people aware of future feelings is known to
be an effective determinant in health promoting interventions. Winterich, & Haws (2011) showed that future-focused positive emotions, such as anticipated pride, have a positive effect on self-control which results in less unhealthy food preferences and consumptions. The current research indicated a low prevalence of feelings of pride after food indulgence. It might therefore be effective to develop interventions using anticipated feelings of pride to improve eating behavior. For example by making people aware that they will not experience positive feelings of pride when they indulge. This might increase people’s motivation to act according to one’s health goals.
Moreover, the results of the current study indicate that high levels of the negative emotions frustration and guilt are experienced after indulgence. Evers et al. (2009) and De Witt Huberts et al. (2012b) showed that negative emotions or negative emotional experiences could lead to indulgence. Thus, indulgence could lead to negative emotional experiences which could, subsequently, lead to indulgence again. Health promoting intervention could make people aware of these consequences of indulgence which might motivate people to stay self-controlled when experiencing a food temptation.
It is suggested to use the data of the current study to conduct analysis including the factor time. This way, it can be clarified if the current results that were contrary to the expectations really exist or are a result of some other factor. Including the factor time will make it possible to predict levels of experienced conflict and emotions on one measurement from levels of indulgence and justification use on a previous measurement. Using such multilevel analysis will make the relationship between these variables more visible.
Altogether this study is a good first step in the domain of nonexperimental studies on the topic of eating behavior and justification use. Although results were not completely in line with the expectations, they still provide perspective for intervention development and future research.
References
Becker, D., Jostmann, N. B., Wiers, R. W., & Holland, R. W. (2015). Approach avoidance training in the eating domain: Testing the effectiveness across three single session studies. Appetite, 85, 58-65.
De Witt Huberts, J. C., Evers, C., & De Ridder, D. T. (2012a). License to sin: Self‐licensing as a mechanism underlying hedonic consumption. European Journal of Social Psychology, 42(4), 490-496.
De Witt Huberts, J. C., Evers, C., & De Ridder, D. T. D. (2012b). Thinking before sinning: Temptations elicit justification processes. Manuscript submitted for publication. De Witt Huberts, J. C., Evers, C., & De Ridder, D. T. D. (2012c). Emotional license: Negative
emotions as justification for selfregulation failure. Manuscript submitted for publication.
De Witt Huberts, J. C. D. W., Evers, C., & De Ridder, D. T. (2014). “Because I Am Worth It” A Theoretical Framework and Empirical Review of a Justification-Based Account of Self-Regulation Failure. Personality and Social Psychology Review, 18(2), 119-138. Evers, C., de Ridder, D. T., & Adriaanse, M. A. (2009). Assessing yourself as an emotional
eater: Mission impossible? Health Psychology, 28(6), 717.
Hagger, M. S., Wood, C., Stiff, C., & Chatzisarantis, N. L. (2009). The strength model of self-regulation failure and health-related behaviour. Health Psychology Review, 3(2), 208-238.
Herman, C. P., & Polivy, J. (1980). Restrained eating. Obesity, 208-225.
Hofmann, W., Baumeister, R. F., Förster, G., & Vohs, K. D. (2012). Everyday temptations: an experience sampling study of desire, conflict, and self-control. Journal of Personality and Social Psychology, 102(6), 1318.
Hofmann, W., Friese, M., & Wiers, R. W. (2008). Impulsive versus reflective influences on health behavior: A theoretical framework and empirical review. Health Psychology Review, 2(2), 111-137.
Hohlstein, L. A., Smith, G. T., & Atlas, J. G. (1998). An application of expectancy theory to eating disorders: Development and validation of measures of eating and dieting expectancies. Psychological Assessment, 10(1), 49.
Johnson, F., Pratt, M., & Wardle, J. (2012). Dietary restraint and self-regulation in eating behavior. International Journal of Obesity, 36(5), 665-674.
Kroese, F. M., Evers, C., & De Ridder, D. T. (2009). How chocolate keeps you slim. The effect of food temptations on weight watching goal importance, intentions, and eating behavior. Appetite, 53(3), 430-433.
Mann, T., de Ridder, D., & Fujita, K. (2013). Self-regulation of health behavior: social psychological approaches to goal setting and goal striving. Health Psychology, 32(5), 487.
Merritt, A. C., Effron, D. A., & Monin, B. (2010). Moral self‐licensing: When being good frees us to be bad. Social and Personality Psychology Compass, 4(5), 344-357.
Ramanathan, S., & Williams, P. (2007). Immediate and delayed emotional consequences of indulgence: The moderating influence of personality type on mixed emotions. Journal of Consumer Research, 34(2), 212-223.
Shafir, E., Simonson, I., & Tversky, A. (1993). Reason-based choice. Cognition, 49(1), 11-36. Tangney, J. P., Baumeister, R. F., & Boone, A. L. (2004). High self‐control predicts good
adjustment, less pathology, better grades, and interpersonal success. Journal of Personality, 72(2), 271-324.
Taylor, C., Webb, T. L., & Sheeran, P. (2014). ‘I deserve a treat!’: Justifications for indulgence undermine the translation of intentions into action. British Journal of Social Psychology, 53(3), 501-520.
Verhoeven, A. A., Adriaanse, M. A., de Vet, E., Fennis, B. M., & de Ridder, D. T. (2014). It's my party and I eat if I want to. Reasons for unhealthy snacking. Appetite, 84, 20 27.
Wagner, D. D., & Heatherton, T. F. (2013). Emotion and Self-Regulation Failure. Handbook of Emotion Regulation, 613-628.
Winterich, K. P., & Haws, K. L. (2011). Helpful hopefulness: The effect of future positive emotions on consumption. Journal of Consumer Research,38(3), 505-524.
World Health Organization. (2015). Obesity and overweight, Fact sheet N°311. Retrieved from http://www.who.int/mediacentre/factsheets/fs311/en/
World Health Organization (2006). Global Database on Body Mass Index, BMI Classification. Retrieved from http://apps.who.int/bmi/index.jsp?introPage=intro_3.html
Appendix A
An Outline of the Content of all Questionnaires
Intake questionnaire. First, demographics were obtained (gender, age, hometown, level of
education, occupation, study, name of the school/university, profession, if non-student participants were part-time/full-time workers and their household composition).
Second, the extent to which participants had goals about following a healthier diet or getting a healthier weight and the importance of these goals were measured with 4 items that had to be answered using a Visual Analogue Scale (VAS) with scores varying from 0 (not applicable to me) to 100 (applicable to me).
Third, participants were asked to fill in their target weight (kg) and the importance of reaching or keeping that weight (VAS, with scores varying from 0, not important, to 100, very important).
Fourth, measures of length (cm) and current weight (kg) were used to calculate BMI (Body Mass Index) scores of the participants.
Fifth, the trait self-control was measured using the brief self-control scale developed by Tangney, Baumeister and Boone (2004). This self-control scale consisted of 13 items that had to be answered using a 5-point Likert scale varying from 1 (strongly disagree) to 5 (strongly agree). Self-control scores could vary between 13 and 65 with higher scores indicating higher levels of self-control. According to Tangney et al. (2004), the self-control scale had a good reliability and validity. In the current study the reliability was good, the Cronbach alpha coefficient was .82.
Sixth, peoples tendencies to use justifications was assessed using a questionnaire developed for this study by Prinsen, the ‘Excuus Truus’ questionnaire. This questionnaire consisted of 16 items about using justifications to indulgence, again a VAS was used to answer the questions. Scores could vary from 0 (indicating a low tendency to use justifications) to 1600
(indicating a high tendency to use justifications). The reliability of the ‘Excuus Truus’ questionnaire was good, with a Cronbach alpha coefficient of .87.
Experience sampling measures. Every questionnaire involved questions about food
temptations participants experienced in the two hours previously to receiving the questionnaire. First, participants had to answer to a question (yes/no) that asked participants if they experienced any food related temptations. When they did, an open question about the content of de temptation was asked. Next, four items that had to be answered with a VAS scale (0, indicating a low score and 100, indicating a high score) questioned, subsequently, the desire strength (1), the degree of conflict (between participants health goals and indulgence to the desire) (2), degree to which participants offered resistance (3) and the degree of indulgence (the degree to which they gave in to the temptations) (4). After that, an open question asked participants how many other food temptations were experienced in the past two hours. If other food temptations were experienced, an open question followed asking how many of these temptations were given in to by the participants.
Second, justification use was assessed by asking participants to choose from a 35 items list of possible justifications. Participants could indicate for every item separately if that item was applicable to them in the past two hours or not, so participants could have had multiple justifications. Participants score on justification use was calculated by taking a sum score of the first 27 items of the list. Thus, scores could vary between 0 (the participant did not use a justification) to 27 (the participant used 27 justifications).
Third, current emotional state was examined with a graphic slider question. The question had to be answered by dragging a slider and by doing so a smiley appeared that changed its expression depending on the participants answer. Scores could vary between 1 (unhappy smiley) and 5 (happy smiley).
Fourth, emotional consequences were measured with eight items asking to which degree participants currently felt satisfied, comfortable, proud, confident, frustrated, stressed, regretful and guilty. Participants answered the questions using a VAS, scores of each emotion could vary between 0 (low degree of experienced emotion) to 100 (high degree of experienced emotion).
Fifth, self-regulation ability was assessed with four items that had to be answered using a VAS. The items concerned participant’s feelings of control about their eating behavior, their trust in being able to resist food temptations, participants motivation to act according to their diet goal and the degree of importance of the diet goal of participants. Scores on each item could vary from 0 (low self-regulation ability) to 100 (high self-regulation ability).
Final questionnaire. The final questionnaire aimed to assess how participants experienced
their participation.
First, the restrained eating scale (RS; Herman & Polivy, 1980) was assessed, the scale consisted of 10 items examining the degree to which participants were restrained eaters. All items had to be answered with a scale varying from 1 (low restrained) to 5 (high restrained). The reliability of the RS questionnaire was good, the Cronbach alpha coefficient was .75.
Second, BMI was assessed by asking participants length (cm) and weight (kg).
Third, the Eating Expectancy Inventory (EEI; Hohlstein, Smith, & Atlas, 1998) was assessed to measure eating and dietary expectancies. This scale consisted of 34 statements about eating and emotions, scores per item could vary between 1 (not true) to 7 (true). The reliability of the EEI was good, with a Cronbach alpha coefficient of .95.
Fourth, two items were questioned to examine if participants were emotional eaters. Scores on each item could vary between 1 (not emotional eater) to 7 (emotional eater).
Fifth, four questions about the representativeness of the experience sampling week were asked: one question concerning the degree to which the test week was representative of a normal week (VAS, 0 is not representative, 100 is very representative), one question (yes/no) asking if
participants were in holiday during the test week, one question (yes/no) asking if participants were in their period during the test week and a final question (yes/no) asking if participants were pregnant during the test week.
Sixth, three items about participant’s honesty and the difficulty of participating were questioned. Participants were questioned about the degree of honesty of their answers during the experience sampling week (VAS, scores varying from 0, not honest, to 100, very honest), about the level of difficulty of participating (VAS, scores varying from 0, not difficult, to 100, very difficult) and the degree to which they did not report the experience of a food temptation while actually they did experience one (scores varying from 1, never, to 5, always).
Seventh, three open debriefing questions were asked to determine if participants knew the actual goal of the study, if participants had any feedback and if participants were okay with being asked to participate in future studies.
Appendix B
Descriptive Statistics of the Independent and Dependent Variables Table 3
Descriptive statistics of the independent and dependent variables
M SD Justifications 1.96 2.25 Conflict 66.04 25.62 Total Indulgence .96 .62 Indulgence 62.44 36.50 Satisfaction 66.47 20.48 Comfort 67.15 20.40 Pride 50.37 25.55 Confidence 60.48 21.26 Frustration 17.88 22.17 Stress 20.75 23.34 Regret 13.69 19.18 Guilt 13.80 19.13
Note. Justification could vary between 0 – 27, Conflict between 0 – 100, indulgence between 0 – 100, and scores on each emotion could vary between 0 – 100. Total indulgence did not have a maximum possible score.
