Facets of Mindfulness in Stages of Behavior Change Toward Organic Food Consumption

Out of 610 largely completed surveys, 567 participants correctly passed the two quality-check items. Since structural equation modeling ideally requires complete data, we excluded seven cases that had missing values on relevant measures. Finally, 560 cases remained for the analyses. An a priori power analysis, using G*Power (Faul et al. 2007), suggested that a sample size of at least 492 is required to reveal a medium size effect regarding the mean differences between the stages (f² = 0.15, α = 0.05, 1-β = 0.80).

The mean age of the participants was 30 (SD = 10.5). The educational degree was high, and most of the participants worked in a full-time (29.5%) or half-time (16.5%) occupation. About one-third of the participants were undergraduate students (36.1%). Most (55%) lived in a city with a population of over 100,000 people. The proportion of people with vegan (6.4%) and vegetarian (16.1%) diets was high compared to the German population, but the majority reported an omnivore diet (72.4%). Few participants reported a diet with fish but without meat (3.2%). Almost half participants (43.6%) had no experience with meditation (e.g., vipassana) or meditative bodywork (e.g., yoga). The other half reported experience but no regular meditation practice (48.4%), while a minority stated that they practiced at least once a week (8%) (see demographic characteristics in Table 1).

Data collection was done through an online survey using the software Qualtrics. The survey invitation was spread in different online communities with and without relation to the topics of food to ensure enough participants in every stage of action. As compensation, the participants were offered to participate in a lottery of 25 × 20 Euro Amazon and 50 × 10 Euro drugstore vouchers.

The analysis of the data was done in R version 3.4.3. The means of all stage variables were compared across the different stage groups using ANOVA and post hoc tests with Bonferroni correction to test the plausibility of the stage model. Variance homogeneity was tested in advance using a Levene test. The fit of the basic stage model (model 1, see Fig. 2) was tested through a structural equation model (SEM) using the R package lavaan. The multivariate normal distribution was tested with the package MVN. A confirmatory factor analysis (CFA) was done to proof the factor structure of the mindfulness dimensions. The model fit for the CFA and SEM was evaluated by looking at the comparative fit index (CFI), Tucker–Lewis index, root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). CFI and TLI of ≥ .95, an RMSEA, and an SRMR of ≤ .08 indicated an acceptable fit (Hu and Bentler 1999). As the data were not multivariate normal distributed, the maximum likelihood estimation was applied with the Satorra–Bentler correction for the test statistics. The significance of the indirect effects was tested with bootstrapping (N = 5.000) and the estimation of 95% bias-adjusted percentile (asymmetric) confidence intervals.

Addressing Q1 and Q2, the relationships between the five mindfulness dimensions, intentions, organic food consumption, and stage-specific variables were explored with SEM by the following steps: First, a reduced stage model only with the three intentions, organic food consumption behavior and with the mindfulness facets as predictors of those (model 2, see figure in supplementary materials). Second, an extended stage model with the mindfulness facets as predictors for intention and behavior to test whether the relationships between mindfulness, itentions, and behavior remain stable when the complete basic stage model is included (model 3, see Fig. 3). Third, an extended stage model with the mindfulness facets as predictors of the stage-specific variables (model 4, see Fig. 4). Addressing Q3, the mindfulness mean differences of the members of the stages groups were examined using an ANOVA.

Overall, the means of the stage variables were as expected: the higher the stage, the higher the means (see Table 2). Only perceived behavior control differed slightly from this pattern. As the items on the self-efficacy scales were only presented to participants who stated at least a low intention to prefer organic foods, missing values occurred on those scales, especially in members of the preaction stage. This is consistent with the assumption that members of the preaction stage have no or a very low intention and, consequently, no or very low self-efficacy toward the behavior change. Therefore, the variables of maintenance and recovery self-efficacy were not included in the following SEM. The internal consistencies of all the measures ranged from satisfactory to good (see Table 3).

The SEM of the basic stage model (Fig. 2) showed a good fit (χ²(77) = 185.907, p < .000; CFI = .979, TLI = .971, RMSEA = .053, SRMR = .039), which indicates a plausible model. The CFA with mindfulness revealed a broadly acceptable fit (χ²(160) = 444.654, p < .000; CFI = .935, TLI = .922, RMSEA = .060, SRMR = .053). To maintain comparability, and due to the lack of substantially theoretical reasons, we refrained from modifying the model. An ANOVA including the participants who provided information about their income (N = 289) revealed no significant differences in the per capita income of the members of different stages (F(3,292) = 0.65, p = .584).

The model 2, which included intentions and behaviors as dependent variables and the five mindfulness dimensions as predictors, revealed significant relationships between observing and goal intention (β = .317, p < .000), between describing and goal intention (β = −.119, p = .024) and between nonreactivity and implementation intention (β = .102, p = .049). The model showed overall an acceptable fit (χ² (292) = 668.218, p < .000; CFI = .948, TLI = .938, RMSEA = .051, SRMR = .046). No significant relationships between the mindfulness dimensions and intentions remained in the model 3, where the stage-specific variables were included. This indicated that the relationships found in the previous model do not remain stable when the other stage variables are added (Q2.1). The overall fit of model 3 was broadly acceptable (χ² (502) = 969.028, p < .000; CFI = .952, TLI = .944, RMSEA = .043, SRMR = .051). The changes in the explained variance in the different models are summarized in Tables 4. In model 3 (see Fig. 3), compared to the basic stage model, a small positive change in explained variance through the inclusion of mindfulness occurred in the goal intention (ΔR² = .012) and the implementation intention (ΔR² = .012).

Model 4 tested the mindfulness dimensions as predictors of the stage-specific variables (see Fig. 4). It revealed positive relations between observing and social norms (β = .157, p = .046), personal norms (β = .301, p < .000), and attitude (β = .283, p < .000). Further, a weak relationship between the non-judging facet and attitude (β = .165, p = .024) was found. The model fit remained overall acceptable (χ² (517) = 963.833, p < .000; CFI = .954, TLI = .948, RMSEA = .041, SRMR = .046). On the basis of model 4, the following indirect effects were tested: (1) the indirect effect of observing on personal norms, mediated by social norms, (2) the indirect effect of observing on goal intention, mediated by personal norms, (3) the indirect effect of observing on goal intention, mediated by attitude, and (4) the indirect effect of non-judging on goal intention, meditated by attitude. All of the tested indirect effects were significant, noticeable through a confidence interval above zero (see Table 5). Compared to the basic stage model (model 1, see Fig. 2), the inclusion of the mindfulness facets caused a considerable increase in the explained variance in the personal norm (ΔR² = .108, see Table 4).

The comparison of the different stage groups in terms of the mindfulness dimensions yielded an overall significant difference in the dimension observing (F (3, 556) = 4.61, p = .003), with a small main effect for stage membership (η² = .02, 95% CI [.004; .053], f = .16). The post hoc test revealed a significant medium-sized mean difference (p = .002, d_cohen = .43) of observing between the stage of predecision (M = 3.29) and postaction (M = 3.64). Further, in the post hoc tests, non-judging was significantly different (p = .034, d_cohen = .32) in the preaction (M = 3.31) and postaction (M = 3.51) stages (see Table 6).

The test of the relationship between mindfulness, intentions, and organic food consumption in the reduced stage model (model 2) shows a direct connection between observing and goal intention. Even though this relationsship is not stable, when the complete stage model is included, it might indicate that observing is connected with the development of a general goal intention toward organic food consumption. A greater sensitivity toward one’s experiences through mindful observation may constitute a greater basis for reflection and value clarification (Shapiro et al. 2006) as well as meaning in life (Garland et al. 2015). This in turn can be seen as a basis to develop a goal intention that is consistent with one’s general life goals. Second, the items of observing include aspects of perception of the natural environment (i.e., “I pay attention to sensations, such as the wind in my hair or sun on my face.”). This could point to the relevance of the direct awareness of natural environments to engage in pro-environmental behaviors, specifically food consumption. Future studies should consider variables such as connectedness to nature that are directly related to the natural environment and, at the same time, predictors of pro-environmental behaviors (Barbaro and Pickett 2016; Mayer and Frantz 2004). Since nature connectedness is also a source of happiness (Zelenski and Nisbet 2014), research on this pathway is promising. Further, in the reduced stage model (model 2), nonreactivity showed a positive but weak connection with the implementation intention. Nonreactivity is a central dimension of mindfulness that is relevant for emotion regulation strategies (Desrosiers et al. 2014), and helps to control impulsive acting toward food consumption (Keesman et al. 2017; Park and Dhandra 2017). The relationship between nonreactivity and the implementation intention suggests that nonreactivity might go along with a planned implementation process. Further experimental studies could focus on the role of nonreactivity in breaking habits with regard to pro-environmental behavior. A weak significant negative relationship between describing and the goal intention occurred unexpectedly in model 2. At this point, no post hoc explanation seems plausible, and, thus, we refrain from drawing conclusions with respect to our hypotheses. It should also be noted that the describing facet is not usually seen as a core element of the tradistional concept of mindfulness in Buddhisn and that the assessment of the ability to describe feelings with words in the context of mindfulness might be especially difficult (Grossman and van Dam 2011). Further, the relationships between the mindfulness facets and the intentions (model 2) do not remain significant when the stage-specific variables are included (model 3). This highlights the limitations of mindfulness as a direct predictor of intentions and organic food consumption behavior when directly compared to the predictive value of personal and social norms, attitudes, and perceived behavior control. A comparison of the model 3 with the basic stage model shows that the mindfulness aspects contribute no substantial explained variances in intentions or behaviors. This is in line with the assumption that mindfulness is a distal predictor of pro-environmental behavior.

The analysis of mindfulness aspects and the stage-specific variables revealed relationships of observing with the social and personal norms as well as with attitude. At this point, indirect effects of observing on goal intention and personal norm were identified. First, personal norms mediated the connection between observing and goal intention. This may indicate that observing in this context might be related to reflective processes, which in return result in a personal obligation toward the preference of organic food or a sustainable lifestyle in general. With regard to personal norms, meaning in life and different sources of meaning (Schnell 2011; Steger et al. 2006), life goals, and values (Unanue et al. 2016) could potentially explain the mechanism of this relationship and should be addressed in future studies. Second, a positive attitude toward the preference of organic food mediated the relationship between observing and goal intention. According to the TPB, a positive attitude toward a behavior is developed through the evaluation of perceived benefits or disadvantages as well as the outcome expectations of certain behaviors (Ajzen 1991). With regard to organic food consumption, especially beliefs about health benefits, better taste and environmental concerns are strong predictors for a positive attitude toward organically grown foods (Aertsens et al. 2009; Michaelidou and Hassan 2008). Moreover, a heightened mindful awareness toward internal and external stimuli goes along with health orientation and health behavior in general (Geiger et al. 2018b). Further, due to the sensitization of one’s senses through mindful observing (Hong et al. 2014), high observing might also go along with a higher demand for good taste and enjoyment of food. Therefore, a positive attitude toward organic food seems more likely, as people tend to expect the taste of organic food to be superior compared to conventional food (Hughner et al. 2007). Overall, to investigate the role of observing for attitude and goal intention at a deeper level, health orientations as well as beliefs about the taste of organic food could be valuable. Third, an indirect effect of observing on personal norm, mediated by social norm, was identified. Since a higher ability to observe internal and external stimuli may also be connected to the salience of social norms, a positive relationship between observing and social norms is plausible. At this point, in further studies, a distinction between descriptive and injunctive social norms or global and local social norms could be valuable in understanding the mechanism (Kormos et al. 2015; Vesely and Klöckner 2018). Fourth, an indirect effect of non-judging on goal intention, mediated by attitude, occurred unexpectedly. As non-judging involves a non-evaluative stance toward thoughts and feelings, a direct connection to any intention does not seem plausible. According to the present results, people who judge their thoughts less have higher goal intention toward organic food consumption. One explanation could be that those not distracted by negative thoughts or rumination develop a goal intention more easily than those who do not. In fact, a decrease in rumination is considered one of the central mechanisms of mindfulness (Svendsen et al. 2017). A closer look at the suggested paradoxes of mindfulness as well as mechanisms such as reperceiving could be valuable to understand this relationship (Shapiro et al. 2006; Shapiro et al. 2018).

In model 4, the gain in the explained variance through mindfulness is quite low. This indicates a limited value of mindfulness to explain the factors that influence the purchase of organic foods and should be noted with regard to the interpretation of the present results. One exception is that the explained variance in personal norms grew substantial when mindfulness was included. This might demonstrate the potential of mindfulness to go along with a moral obligation toward organic food consumption. With reference to the different suggested pathways from mindfulness to sustainable consumption (Geiger et al. 2018a; Hunecke and Richter 2019), this result suggets the role of mindfulness to go along with personal values that are connected to organic food consumption.

As expected, the mindfulness aspect of observing was significantly higher in members of the postaction stage than in those of other stages. This may indicate that observing, or the degree to which people notice internal and external experiences, can be relevant in successfully passing the stages toward acquiring the new behavior of organic food consumption. This is in line with the theoretical assumptions of the stages and intervention recommendations of the SSBC. While in the stage of predecision, the enhancement of awareness and self-focus is central (Bamberg et al. 2011), and people in the stage of postaction have already processed through these steps and have higher levels of observing as a consequence. A significant mean difference was unexpectedly found in the dimension of non-judging between people in the preaction and postaction stages, indicating that those in the preaction stage were more judgmental toward their thoughts and feelings. A post hoc interpretation may be that this is due to the reflective and uncertain nature of the preaction stage, in which people want to change their behavior but are not yet sure how to achieve this goal.

The cross-sectional design of the study does not allow causal conclusions regarding the impact of mindfulness on organic food consumption intentions and behavior. The design also limits the reliability of the mediation analyses, and this can lead to biased estimates in single cross-sectional studies (Maxwell and Cole 2007). Future studies can collect data from (quasi-) experimental designs to address questions of causal mechanisms between mindfulness and pro-environmental behaviors. While data were collected conscientiously in selected communities to reach a broad spectrum of people with regard to the assumed stages of action, it is not representative but a convenience sample, which limits the generalizability of the results.

We successfully applied a modified version of the SSBC to organic food consumption and tested the plausibility of the theoretical model analyzing mean differences and using SEM. However, the high intercorrelations between the intentions indicated similarity of the different intentions even though the indicators for multicollinearity (variance inflation factor) and the fit indices were within the common limits. Although we included central stage-specific predictors and the most important variables with regard to organic food consumption, not all the predictors suggested in the original model were controlled. This could have led to different results in relation to the mindfulness aspects. Further, the assessment of pro-environmental behaviors using self-reports is an adequate and commonly used approach, but it can still involve validity problems with regard to actual behavior (Kormos and Gifford 2014). The stage model has a focus on the process of behavior change that is marked by four different stages. We used the categorization in the stages to examine the differences between the participants in terms of to mindfulness to gain additional information beyond the stage variables. However, in terms of the assumed stages, it was not possible to proof the stability of the stage or stage progression over time due to the cross-sectional study design. For this purpose, longitudinal studies are required, and the stage model approach would be useful to track the changes in specific behaviors from one stage to another.

The focus of the present study was on the preference of organic food consumption, because it is a central part of a sustainable lifestyle in terms of ecological impact. However, there are further potential motivators that we have not directly considered in our model, such as beliefs about health benefits and better taste of organically grown foods (Aertsens et al. 2009). Furthermore, other ecologically relevant fields of action such as mobility, energy conservation have to be investigated to fully estimate the potential of mindfulness to foster a sustainable lifestyle. The identification of the relevant mindfulness aspects in relation to stage-specific variables is helpful for planning future (quasi-) experimental studies. In the present study, observing showed the strongest relationships with social and personal norms and attitude. To further investigate this relationship, a look at conceptual close psychological constructs such as self-awareness (Vago and Silbersweig 2012) could be useful. Further, the aspect of observing can be distinguished in inner and outer awareness (Bergomi et al. 2015). Since mindfulness is often used as an umbrella term for various concepts and meditation practices, a consistent definition or measurement is not identifiable in the present research literature (Van Dam et al. 2018). It should be noted that we assessed self-attributed mindfulness with a questionnaire in a sample with mostly non-meditators in this study. It is very likely that self-reported mindfulness has limited similarity with the experience of mindfulness practice, and the interpretation of the items varies in meditator and non-meditator samples (Grossman and van Dam 2011). For instance, in a study by Baer et al. (2008) using the FFMQ, the observing facet varied in its connection with psychological symptoms in meditator and non-meditator student samples. Therefore, the results of the present study may be different in a meditator sample. Moreover, the study does not allow conclusions regarding the potential of mindfulness meditation to promote pro-environmental behavior. The results of the present study can only be interpreted in terms of the specific measurement used. Although the shortened version of the FFMQ used was largely proofed regarding its construct validity in a previous study (Hunecke and Richter 2019), a short version can limit the comparability. While the use of a multidimensional instrument prevents the derivation of an extremely simplified understanding of mindfulness, some aspects of mindfulness are still missing, especially paradoxes of mindfulness and its practice, such as acceptance vs. change and effort vs. non-striving (Shapiro et al. 2018), and they are not adequately assessed with the common instruments. The focus on the five different aspects of mindfulness is beneficial to understand the relevant aspects of mindfulness for organic food consumption and the stage-specific variables. The present study, therefore, offers useful recommendations for future studies. However, when it comes to the practice of mindfulness, the complete separation of these aspects is not beneficial, owing to the interactions between the different mindfulness dimensions (Desrosiers et al. 2014). To ensure the benefits for emotion regulation of mindfulness meditation, the aspect of observing should not be trained in isolation but always along with a non-judging stance (Baer et al. 2008; Iani et al. 2019).