Relaxed minds for healthier food choice at school: A field experiment in southern Mexico

doi:10.21203/rs.3.rs-3869301/v1

Stress leads to unhealthy food choices since the school-age stage. Yet, there is limited evidence particularly in low- and middle-income countries regarding the impact of stress-reduction strategies on school-age children’s food choices. Such aspects were crucial during the recent COVID-19 pandemic, which exacerbated psychological distress and unhealthier food choices among children. During the pandemic period, we conducted a field experiment in southern Mexico to assess the impact of stress-reduction strategies on the food choices of over 1,400 children aged 9–12. Half of the school-classes in the sample were randomly assigned to a stress reduction strategy namely meditation, which comprised six audios with basic relaxation techniques and intuitive messages to guide food choices. Additionally, all participants received information signalling that an amaranth snack was nutritious (i.e., the healthy snack), which was paired with a chocolate bar (i.e., the unhealthy snack) as part of a snack choice experiment. The meditation program effectively reduced stress among treated children. The effect was larger among students attending schools in lower-income areas. In general, we found that the share of students choosing the healthy snack did not differ between the treated and the control groups. Yet, we highlighted that meditation increased healthy snack choices by 4%-points among children that chose their snack for non-health or nutrition-related reasons. Upon collecting their snack, students had the chance to exchange their original choice for the other snack. Students that meditated were more likely to exchange their originally chosen “unhealthy snack” towards the healthy snack than students in the control group. Our study sheds some light on the relevance of addressing psychological well-being to complement other strategies to improve food choice at school.

Behavioral Economics

School experiment

food choices

children

meditation

dynamic inconsistency

The global efforts to promote healthy food choices and improve diets often rely on educating the public (Willett et al., 2019). However, it is increasingly recognized that information does not necessarily motivate individuals to change their diets (List and Samek, 2015; Samek, 2019; Van Kleef and Van Trijp, 2018). The effectiveness of information provision is bounded by biases and heuristics that make individuals more prone to unhealthy food choices, particularly in stressful situations. Stress has been shown to be associated with higher craving for and consumption of high-fat and high-sugar foods, especially when consumed as between-meals snacks (O’Connor et al., 2008). The recent evidence suggests that the impact of stress on unhealthy eating may be observed in children as young as eight-nine years old (Hill et al., 2018).

The period 2020–2023 was particularly stressful for many people, as the COVID-19 pandemic was found to be associated with stress-related mental illnesses, such as anxiety and depression (Kar et al., 2021). Not just the fear of illness but also the socioeconomic consequences of the pandemic: lack of financial resources, unemployment, and fear of a new recession could have increased stress (Landa-Blanco et al., 2021). These effects are likely to be especially high in low-income settings, where poverty-related stress already affects healthy behaviours (Bruijn, 2021; Just and Gabrielyan, 2018). In the present study context, Mexico, up to 50 percent of the population suffered from COVID19-related psychological distress (Landa-Blanco et al., 2021). More prolonged schools closure than in other regions, home isolation, limited access to a computer and/or internet for online education, as well as lack of parental support affected children as much as adults (Landa-Blanco et al., 2021). Moreover, the emerging research on the effects of COVID-19 pandemic on children's weight status suggests acceleration of weight gains especially among school-age children (Vogel et al., 2022).

During the last decades, Mexico, like other low- and middle-income countries, has experienced a steep increase in the consumption of ultra-processed foods and drinks (e.g., sweetened beverages, snacks, sweets and desserts), limited intake of whole foods (e.g., fruits, vegetables, high-fibre cereals), and significant reduction in physical activity (Rivera-Dommarco, 2012). These lifestyle changes together with increasingly obesogenic environments contributed to a rising prevalence of overweight/obesity and other diet-related noncommunicable diseases (Giuntella et al., 2020; Popkin and Ng, 2021; Rivera-Dommarco et al., 2018). At the same time, the policy focus on tackling the obesity epidemic might shadow other forms of malnutrition, such as micronutrient deficiencies, which have been found to be associated with regular consumption of unhealthy diets and may have negative consequences for the health and learning capacity of school-age children (Flores et al., 2010; Gaona-Pineda et al., 2018; Rivera-Dommarco et al., 2018).

The advantages of promoting healthier diets in school settings are widely recognized, as eating behaviours are formed in childhood and school grounds allow easy, continuous, intensive contact with children (Dudley et al., 2015). There is a large body of literature assessing the impact of school interventions on children’s dietary decisions (DeCosta et al., 2017; Madden et al., 2017). In the field of behavioural economics, most of the related studies focused on externally-deployed tools, such as information provision, nudges, changes in food environment or temporary incentives (Belot et al., 2016; DeCosta et al., 2017; List and Samek, 2015; Loewenstein et al., 2016). On the other hand, cognitive self-deployed strategies other than goal-setting (e.g., self-monitoring, psychological distancing, meditation, cognitive therapy) have not been yet utilized in the behavioural economics research to address healthier eating at schools (Duckworth et al., 2018; Reijula and Hertwig, 2022; Samek, 2019). Such strategies are more commonly used in interventions addressing childhood obesity, but the pool of research is still limited (Salam et al., 2020). In the specific case of meditation, such school interventions usually suffer from small sample sizes (de Lara Perez and Delgado-Rios, 2022). Altogether, there is scarce related research in the bodies of literature addressed above regarding interventions conducted in low- and middle-income countries (Brown et al., 2019; Keats et al., 2021; Salam et al., 2020).[1]

In the present study, we propose a self-deployed strategy namely meditation to improve food choices among school-age children in southern Mexico. This region has the highest rates of poverty and overweight/obesity among children below 5 years of age, and both the highest prevalence of diabetes and lowest dietary diversity in the group of adults (CONEVAL, 2019; Levy et al., 2016; Shamah et al., 2018). Meditation has been shown to reduce stress, ameliorate outcomes associated with increased self-control and reduced impulsive eating, and, in turn, increase consumption of healthy foods (Duckworth et al., 2018; Fredrickson et al., 2021; Sala et al., 2020). These benefits are especially relevant as COVID-19 aggravated stress levels. While schools had recently opened at the time of this study, children faced the consequences of the damage to academic performance caused by the prolonged period of social isolation and of the socioeconomic stress many of their parents still experienced. Focus group discussions with parents and comments by schools’ staff in the research area shed light on the emotional impact of the pandemic on school-age children from different income-levels, which likely affected their diets (UNICEF, 2021a, 2021b). Therefore, at the end of the meditation intervention, we assessed the impact of this strategy on children’s food choices via reductions in their stress level. We measured the behavioural outcome through a snack choice experiment, where participants selected either a healthy grains bar or an unhealthy ultra-processed snack. Furthermore, we assessed if meditation reduced dynamically inconsistent choices by allowing children to change their original choice while collecting their snack.

The paper has the following structure. In the next section, we present the main hypotheses of this study. Then, we present details about the sample, experimental materials and design, as well as the estimation strategy. In the fourth section, we report the results of the study. In the last section, we discuss the results and conclude.

Individuals’ decision-making processes can be modelled using a dual system framework (Kahneman, 2003). These models explain human behaviour as the outcome of brain interactions between two systems: 1) the affective system (System 1), that encompasses emotions and other motivational states and focusses on the here and now; and the deliberative system (System 2), that assesses options in a broad and goal-oriented perspective and is more concerned with long term outcomes (Fudenberg and Levine, 2006; Loewenstein et al., 2015). System 1 processes are fast, automatic, effortless, emotionally charged and governed by habit, so that this type of thinking is prone to biases and errors; whereas the operations of System 2 are slow, effortful, deliberate and costly, and as a result they typically yield more unbiased and accurate results (Kahneman, 2003; Schilbach et al., 2016).

Loewenstein et al. (2015) introduced a tractable version of two-system models where the ability of System 2 to influence System 1 is moderated by the cost of exerting willpower and the intensity of affective motivations. If the latter two factors decrease, it is more likely that an individual will make choices aligned with System 2’s preferences. The cost of exerting willpower is equivalent to the ability of exercising self-control and is decreasing in terms of increased willpower strength and reduced cognitive demands (Loewenstein et al., 2015; Ruhm, 2012). Furthermore, willpower strength can be assumed to be negatively affected by higher levels of stress, which is consistent with research linking financial stress with unhealthy eating behaviours (Beenackers et al., 2018; Maier et al., 2015; van Rongen et al., 2019). It has been shown that meditation practices reduce stress, thus strengthening System 2 relative to System 1 (Alem et al., 2021; Sala et al., 2020). In particular, school-based meditation interventions have been shown to be a valuable means of reducing stress among children (Campion and Rocco, 2009; Sapthiang et al., 2019). Then, the following mediation hypothesis arises:

Hypothesis 1

Meditation practices reduce stress among children.

As a result of the reduced stress, we expect that children will take more effortful and deliberate decisions, in this case, healthier food selections (Fredrickson et al., 2021; Sala et al., 2020). There is specific evidence that links children’s eating behaviours with emotions, which supports the notion that reduced stress will lead to fewer unhealthy food choices (Messerli-Bürgy et al., 2018; Powell et al., 2017). 2). Meditation practices improve certain self-control outcomes (e.g., eating fewer sweets, less impulsive eating) due to the enhancement of System 2 processes, such as attention outcomes (e.g., sustained attention, working memory) and emotion regulation skills (e.g., openness and acceptance towards life experiences); and to the attenuation of System-1 aspects, such as negative affect outcomes (e.g., depression, anxiety) and impulsive behaviours (e.g., craving) (Creswell, 2017; Duckworth et al., 2018; Jensen et al., 2012; Mason et al., 2016). Therefore, children treated with meditation will be more likely to choose the healthy snack in the food choice experiment. Considering the above scenario, our primary prediction is as follows:

Hypothesis 2

Meditation practices improve children’s food choices.

We predict an additional effect related to dynamically inconsistent choices. Such behaviours arise when decision-maker’s preferences (i.e., revealed preferences) change over time, in such a way that a preference can become inconsistent at another point in time (Sadoff et al., 2020). As shown by Sadoff and co-authors, the direction of dynamically inconsistent food choices among adults is systematically towards less healthy foods. In other words, if adults have the chance to change an advance choice made in the past between healthy an unhealthy foods, they are more likely to switch from healthy to unhealthy options than the other way around. In general, children discount the future more than adults and especially younger and lower-income children are more impatient than others, hence their propensity to make fewer healthy dietary choices (Belot et al., 2016; Just and Price, 2013; Loewenstein et al., 2015). Under Loewenstein’s model, a reduction in the cost of exerting willpower would lead to less dynamic inconsistencies. As meditation is expected to reduce such cost (see above), we posit the following hypothesis:

Hypothesis 3

Children practicing mediation make fewer dynamically inconsistent choices in the direction of unhealthy foods.

Data collection was conducted in Oaxaca, Mexico between March and June 2022 at 27 primary schools (see Fig. 1). At the start of the intervention, primary schools in the area had the option to open voluntarily but most of them were still closed and some COVID-19 regulations such as mandatory use of face masks were still in place at the schools that were already open. Mothers or tutors signed informed consent forms on behalf of their children. A pre-analysis plan was registered before data collection (see https://osf.io/f7yan). Likewise, the hypotheses mentioned above were specified before the data were collected. The research design received ethical approval from Wageningen University’s Social Sciences Ethics Committee.

3.1 Sample

In Oaxaca, 64% of the population lives in condition of poverty (CONEVAL, 2019). Yet, we selected both public (10) and private (17) schools, which enabled a mix of households with different income levels. The majority of the students (67%) in the sample attended a school located in a higher-income area (see Table 1), mainly because most of the schools that opened after prolonged closures were located around the main urban centre. We focused on urban and peri-urban areas (see Fig. 1), where the consumption of unhealthy processed foods is more generalized than in rural areas (Popkin and Reardon, 2018). In the context of the pandemic, there was still a substantial share of schools with online or other sorts of remote schooling, so we focused on schools where students came fully in person (16) or in hybrid (11) formats. Furthermore, we concentrated on the grades 4–6 with children aged 9 to 12, the stage of life where the relationship between emotions and food choices becomes more salient (Blissett et al., 2010).

The sample of participants with consent from their parents was 1,615 students, which was lower than our estimates in the power calculations (1,850) because the available schools had smaller class sizes than previously estimated. In the end, we collected data from 1,467 students because 9.2% of the original sample did not attend classes on the day of the survey or their group was excluded as the school made changes in the composition of their class after treatment assignment. We were not concerned about the attrition issue, as the attrition rate was small and some background indicators (i.e., mother/tutor’s age and education) of those who had attrited were similar to the sample collected. After cleaning the data from observations with incomplete sociodemographic information and outliers, the final analytical sample was 1,433. The share of girls and boys was even in the sample (see Table 1). The percentages of students in each grade were similar, but slightly higher for 4th grade, which explains that average age was around that of a fourth grader (10.34). Whereas the majority of the students’ mothers completed at most secondary education, we still had a substantial share of them having a tertiary level of education completed (44%). Most of the students had a female teacher (61%).

Two focus groups with mothers and teachers mostly from private schools were organized in the research area. In the context of the ongoing pandemic, the idea of the focus groups was to get a better understanding about school-age children’s baseline levels of stress and to determine potential impacts on eating behaviours. The participants reported high levels of distress among children (e.g., angriness, frustration, sadness) due to lack of socialization with their peers, the uncertainty about the duration of the lockdowns, witnessing tensions among their parents and being afraid of COVID-19 contagion once they were back in school. In some cases, children received psychological support. Some disruptions were also observed in eating behaviours, such as a stronger preference towards foods high in saturated fats and processed foods. Teachers mentioned that some students gained weight during the period because of reduced physical activity. We expected that some of these indicators may be exacerbated in lower-income areas due to more extended school closures, lack of access to a computer to attend online schooling, lack of access to a psychologist and pre-existing economic pressures at home.

Table 1 Summary statistics.
Indicator	Mean	Std. Dev.	Min	Max
Indicator	(1)	(2)	(3)	(4)
Students
Girl	0.50	0.50	0	1
Age (years)	10.34	0.97	8	13
4th Grade	0.37	0.48	0	1
5th Grade	0.30	0.46	0	1
6th Grade	0.34	0.47	0	1
Mothers/tutors
Age (years)	38.85	6.97	20	76
No formal education	0.02	0.13	0	1
Primary education	0.29	0.45	0	1
Secondary education	0.25	0.43	0	1
Tertiary education	0.44	0.50	0	1
Teacher
Share of students with a female teacher	0.61	0.49	0	1
Schools
Share of students in a school located in:
Lower-income area¹	0.33	0.47	0	1
Higher-income area²	0.67	0.47	0	1
Main urban centre³	0.36	0.48	0	1
Close proximity to the main urban centre⁴	0.17	0.38	0	1
Outskirts of main urban centre⁵	0.26	0.44	0	1
Area distant from main urban centre⁶	0.21	0.41	0	1
Share of students attending:
Public school	0.56	0.50	0	1
School fully in person	0.75	0.43	0	1
Religious schools	0.16	0.36	0	1
Schools with meditation or yoga programs	0.12	0.33	0	1
Notes: ¹Locations with high or very high levels of poverty according to the Index of Urban Marginalization (IUM) elaborated by the National Population Council (CONAPO). ²Locations with medium or low levels of poverty according to the IUM elaborated by the CONAPO. ³Located less than 3 kms away from the centrum of Oaxaca City. ⁴Located between 3 and 5 kms away from the centrum of Oaxaca City. ⁵Located between 5 and 10 kms away from the centrum of Oaxaca City. ⁶Located more than 10 kms away from the centrum of Oaxaca City.

3.2. Products

For the healthy snack, we decided to focus on an amaranth bar, because it is a traditional Mexican food that is popular in Oaxaca and has a higher micronutrient content than other highly consumed cereals such as wheat, rice and maize. After searching in the market, consulting a local amaranth processor (see https://www.puentemexico.org/) and the piloting phase, we chose a traditional bar called alegria in Spanish (see Fig. 2). Given its amaranth and nuts/seeds content, this snack was high in fiber, protein, iron and calcium (Orona-Tamayo and Paredes-López, 2017; Rastogi and Shukla, 2013). Following the recent labelling regulations in Mexico, this type of snack usually wears the labels “excess of calories” and “excess of sugar” in big retailers. To upgrade the snacks used in our experiment, we asked the amaranth processor to provide us with a version reduced in sugar.

Chocolate bars were identified as the most popular type of unhealthy snack in the area. To determine the specific product, we consulted parents and children and searched in the market for snacks roughly similar to the amaranth snack in terms of price (MXN 17 (USD 0.83)) and weight (40g). The pilot informed the final selection, which corresponded to the most popular chocolate out of three options. The selected option had three labels in its original packaging namely “excess of calories,” “excess of sugars” and “excess of saturated fat.” We assumed that the high content of the latter ingredients and its associated tastiness made this a temptation product, which makes it harder to resist under System 1 processes related to immediate gratification and emotional eating. Given their popularity, children would be more prone to automatic and habitual choices.

3.3. Experimental design

Our study has a block-clustered randomized design (RCT), where clusters are the school classes from grades 4–6 (the three blocks). To avoid COVID-19 contagion, six schools divided each of their classes in two subgroups, so for our study each of these subgroups was considered as a class. All children that attended school in person and had permission from their parents were part of the sample. Half of the school classes in each grade of each school were randomly assigned to a meditation program comprised of six audios. During the duration of the program, the control group did not receive any sort of intervention such as placebo audios.

The audios of the program were provided by a company specialized on self-development and meditation programs (see https://atahualpairigoyen.com/). The intensity was set to six sessions, because we expected the positive effects of meditation to be more salient after a number of sessions and the children to need some time to familiarize themselves with this new practice. We asked teachers to play the meditation audios on a daily basis and the specific schedule of the sessions was agreed depending on the availability of the trained teachers[2] (i.e., the teacher in charge of the class). The schedule was not consistent across schools as some of them had distinct types of hybrid formats in the context of the pandemic[3], thereby resulting in the meditation program lasting between 1 and 3 weeks. Specific instructions were provided to teachers to tell their students to keep their school materials in their backpacks to clear their space and to follow thoroughly the instructions provided in the audios (e.g., close their eyes) (full instructions in Annex A of the supplementary information). The teachers had the option to practice the meditation with their students voluntarily. Each meditation session lasted from 5 to 10 minutes, and its main objective was to guide children to reach a state of relaxation that may enable them to make better food choices. At the same time, we shaped the program towards achieving the specific goal (healthier food choices), a feature that was recently identified in the related literature (Haushofer et al., 2021). In particular, the meditation program included the following components: a) instructions about sitting position b) breathing and mental body scan exercises; c) playful stories that guide children to an imaginary place where they can cultivate positive emotions; d) guided stories that help children to identify which foods are natural and good for their bodies; e) an invitation to experience those foods that are good for their bodies in their daily lives. Aspects d) and e) differ from the standard nutrition information, as they are linked to feelings or perceptions of food, whereas standard information is comprised by facts about nutrition and health. All the mediation audios covered aspects a), b) and c), while only audios 2, 3, 4 and 6 covered aspects d) and e).

The research team was introduced to the students on the first meditation day (full script in Annex B of the supplementary information) but remained outside the classroom while the mediations were played. The application of the audios 1 to 5 was monitored by enumerators, generally the first two days on-site and afterwards remotely via instant messaging (i.e., WhatsApp), as teachers usually managed the application of the audios very easily. Yet, enumerators assisted the teachers to play the audios when needed, but always stayed outside while the audios were played. Immediately after the final mediation session, all treatment and control children participated in a food choice experiment. The experimental choice was recorded in a brief questionnaire (see Annex C of the supplementary information). The first question appeared alone in the first page of the survey, where children chose between two images of snacks. The rest of the questions were asked in the second page of the survey to reduce the likelihood of influencing the snack choice. These questions asked about food preferences and measured stress. The stress section consisted of a set of five questions from the stress in children (SIC) questionnaire (Michels et al., 2012; Osika et al., 2007). The stress questions involved: i) feeling of relaxation, ii) feeling of happiness and iii) whether the child enjoys going to school, which were measured as 1 for no, 2 for sometimes and 3 for yes; iv) frequency of stomach aches and v) frequency of eating when stressed (i.e., emotional eating), which was measured as 1 for very frequently, 2 for sometimes and 3 for never.

The research team told children that they got the chosen snack as a thank you token for filling in the questionnaire. They could choose either an amaranth snack (i.e., the healthy snack) or a chocolate (i.e., the unhealthy snack). Children could see an example specimen of each snack (as shown in Fig. 2). Both specimens were shown by the research team in transparent plastic packaging to avoid teacher, marketing or novelty effects. To signal that the amaranth product is healthy, we referred to it as a “nutritious amaranth bar” in the survey. All subjects received this signalling, so we did not have a pure control group. Similar to Samek (2019), we assumed that information alone would not have a large effect on behaviour. The effective use of information requires substantial System 2-related cognitive efforts to consider future implications and trade-offs regarding nutrition, health and costs (Just and Gabrielyan, 2018; Mancino et al., 2018). As meditation improves System 2’s processes and our meditation audios were purposely geared towards healthier food choices, we expected that nutrition information is more likely to affect the food selections of those children practicing mediation.

Immediately after filling in the surveys, students were moved to a different place to collect their snack, one by one, to a spot out of sight from the rest of the class. Depending on the characteristics of the school building, the other students waited for their turn sitting on the classroom or in a line controlled by enumerators and teachers to avoid that children shared their survey’s answers. Enumerators handed the surveys to the teachers, who called in the next child in turn based on the order of the surveys collected. While picking up their snack, enumerators gave the children the chance to exchange their original choice, so that we could assess the consistency of their food choices (exact wording in Annex D of the supplementary information). They did not know they could change their mind at the time of the original choice in the questionnaire. This approach was similar to Augenblick et al. (2015) and Sadoff et al. (2020), but we measured the above outcome over a shorter time span (minutes vs a week), as we assumed that children tend to be more impatient/impulsive than adults. It is possible that children switch from the unhealthy towards the healthy snack, but we expected that it would be more probable to see switches to the unhealthy option due to its tempting nature. This is in line with the findings in Sadoff et al. (2020).

Control groups encountered our research team for the first time during the survey and snack collection parts of the study, whereas treatment groups had seen some members of our team since the beginning of the meditation intervention. To have a more homogenous response from both groups during the survey and snack collection, the enumerators that oversaw those steps in the treatment groups were different from those monitoring the meditation intervention. The detailed procedure of the day of the food choice experiment is shown in Fig. 3 below.

3.4. Estimation strategy

We estimated the following model:

$${Y}_{ij}={\beta }_{0}+{\beta }_{1}{Meditation}_{j}+\delta {X}_{ij}+{m}_{s}+{t}_{g}+{\epsilon }_{ij}$$

where i and j index participant children and the cluster (school class), respectively.

${Y}_{ij}$ : Outcome variables. Our primary outcome equals 1 if a child chose the healthy snack, 0 otherwise.
- The secondary outcomes were the following:

1) stress, which is measured as the first component of a principal component analysis using the five stress questions mentioned above.

2) dynamically inconsistent food choice (i.e., willpower), dummy equal to 1 if the child exchanged his/her original snack choice from the healthy (unhealthy) snack to the unhealthy (healthy) snack at the moment of collection; 0 otherwise.

${e}_{ij}:$ Error term.

Our primary estimation method was an OLS or linear probability model. A probit model was used as a robustness check for the binary outcomes. Standard errors were clustered at the class level j (N = 130). Although there were some imbalances in the characteristics (e.g., female teacher, location) between the control and treated groups (Table 2); a joint test of orthogonality (i.e., binary probit with the treatment variable on the left-hand side and controls on the right-hand side) suggested that the meditation group was comparable to the control group (Prob > χ² = 0.822). Some controls declared in the pre-analysis plan were not used in the model (e.g., marginalization index, number of meditations practiced by teachers, student-teacher ratio, enumerator and day of the week) because of multicollinearity issues. To assess heterogenous effects without specific hypotheses, we included an exploratory analysis to distinguish gender (pre-registered), income (not pre-registered) and reasons for the snack choice (not pre-registered). For the latter analysis, we controlled for multiple hypothesis testing using the Romano-Wolf correction to control the familywise error rate (FWER) (Clarke et al., 2019).

Table 2 Randomization checks.
Variable	Meditation program		p-value^Y
Variable	Yes	No	p-value^Y
Girl (dummy)	0.50	0.50	0.950
Age	10.32	10.42	0.072
4th Grade (dummy)	0.38	0.35	0.401
5th Grade (dummy)	0.31	0.29	0.373
6th Grade (dummy)	0.32	0.36	0.086
Consumed an amaranth snack last week (dummy)	0.23	0.22	0.981
Consumed a chocolate snack last week (dummy)	0.28	0.25	0.296
Mother's age	39.06	38.60	0.210
Well-educated mother (dummy)¹	0.45	0.44	0.689
Female teacher (dummy)	0.64	0.58	0.014
Attends school located in (dummies):
Lower-income area²	0.34	0.32	0.388
Main urban centre³	0.33	0.39	0.013
Close proximity to the main urban centre⁴	0.16	0.18	0.340
Outskirts of main urban centre⁵	0.29	0.24	0.036
Area distant from main urban centre⁶	0.22	0.19	0.125
Attends (dummies):
Public school	0.54	0.57	0.213
School fully in person	0.73	0.77	0.086
Religious school	0.16	0.15	0.676
School with meditation or yoga programs	0.13	0.10	0.078
N	768	665
Notes: ¹Completed tertiary education. ²Locations with high or very high levels of poverty according to the Index of Urban Marginalization elaborated by the National Population Council. ³Located less than 3 kms away from the centrum of Oaxaca City. ⁴Located between 3 and 5 kms away from the centrum of Oaxaca City. ⁵Located between 5 and 10 kms away from the centrum of Oaxaca City. ⁶Located more than 10 kms away from the centrum of Oaxaca City. ^YIf dummy variable, p-value is based on Pearson χ2 test for the independence between variables, otherwise, it is based on a t-test on equality of means between groups.

4.1. Sensory evaluation of the products

A week prior to the food choice experiment, the share of students that consumed any chocolate bar was somewhat larger than the corresponding share for amaranth bars ((Prob > t = 0.014), it was 27% and 23%, respectively. After the treatment was administered, we asked the children about their visual perception of the products. The answers did not differ by treatment condition. The study’s products had exactly the same weight, but a slight majority of subjects in the sample (53%) perceived the amaranth snack as more filling than the chocolate snack. Participants craved more for the chocolate snack than for the amaranth snack (Prob > t = 0.000), as their average scores (scale 1–5) were 3.90 (SD = 1.11) and 3.44 (SD = 0.98), respectively.

4.2. Manipulation checks

Table 3 shows the effects of the meditation program on the different stress reduction aspects. We highlight that children treated with the meditation program reported a significantly higher score in terms of relaxation. Anecdotical comments by teachers confirmed that students looked very calm and focused after the meditation. On average, the scores of happiness and enjoys going to school were quite high (2.7–2.8 out of 3) and were not significantly different between treatment groups. Also, no differences were reported for reductions in emotional eating. Contrary to expectations, students in the control group reported higher scores for reductions in the frequency of stomach aches. Overall, students exposed to meditation reported a reduction in stress, as their stress reduction score was significantly higher than for those who did not meditate. This leads to the following statement underling the mechanism of the study (H1): Result 1: Practice of meditation reduces stress among school-age children.

Table 3

Effect of meditation on reported indicators of stress reduction.
Dependent variable	Relaxed	Happy	Enjoys going to school	Reduction in frequency of:		Stress reduction score^Y
	Relaxed	Happy	Enjoys going to school	Stomachache	Emotional eating	Stress reduction score^Y
	(1)	(2)	(3)	(4)	(5)	(6)
Meditation	0.236***	0.015	-0.022	-0.140***	-0.023	0.414***
	(0.033)	(0.026)	(0.021)	(0.030)	(0.033)	(0.058)
Girl	0.010	-0.035	0.059***	-0.082***	-0.081*	0.018
	(0.029)	(0.026)	(0.022)	(0.029)	(0.045)	(0.052)
Age	-0.033	-0.028	0.015	-0.010	-0.052*	-0.058
	(0.028)	(0.025)	(0.022)	(0.027)	(0.028)	(0.049)
Well-educated mother	-0.037	-0.005	-0.063*	0.019	0.015	-0.065
	(0.035)	(0.032)	(0.034)	(0.037)	(0.043)	(0.061)
Mother's age	-0.001	-0.001	-0.001	-0.004*	0.004	-0.001
	(0.002)	(0.002)	(0.001)	(0.002)	(0.003)	(0.003)
Female teacher	0.063*	0.088***	-0.006	0.046	0.055	0.112*
	(0.035)	(0.026)	(0.026)	(0.035)	(0.039)	(0.062)
Constant	2.911***	3.033***	2.858***	2.546***	2.562***	0.665
	(0.289)	(0.261)	(0.225)	(0.318)	(0.305)	(0.509)
R²	0.105	0.043	0.083	0.051	0.030	0.104
Mean (Control)	2.40	2.72	2.84	2.24	2.15	-0.23
Notes: OLS estimations and observations equal to 1,433 in all columns (1,432 in 4,6). Variables in columns 1–5 are measured in a scale from 1 to 3. Standard errors clustered at the school class level shown in parentheses. * p < 0.1, p < 0.05, * p < 0.01. ^YConstructed as the first principal component of the dependent variables in columns 1–5. All regressions include school and grade fixed effects. Intra-cluster correlation (ICC) ranged from 0.00 to 0.11.

4.3. Evaluating the impact of meditation on food choices

Table 4 reports the effect of the meditation program on the primary outcome. In general, 33% of the sample selected the heathy snack. Students that practiced meditation were slightly more likely to choose the healthy snack than those in the control group, but the effect was not statistically significant. The effect was slightly larger when dropping the clusters where the protocol was violated (e.g., students heard about the snack choice from family members in other class), but still non-significant. Among the associations with the demographic variables, we observed that girls and those with well-educated mothers were more likely to choose the healthy snack than boys and those with less-educated mothers, respectively. Altogether, our main estimates did not support our primary hypothesis: Result 2: Meditation does not improve children’s food choices.

Table 4

Effect of meditation on healthy snack choice.
Dependent variable	Healthy snack chosen
	(1)	(2)	(3)	(4)	(5)
Meditation	0.004	0.004	0.003	0.006	0.006
	(0.025)	(0.025)	(0.025)	(0.025)	(0.025)
Girl		0.046*	0.047*	0.046	0.046
		(0.028)	(0.028)	(0.028)	(0.028)
Age		0.029	0.029	0.028	0.028
		(0.024)	(0.024)	(0.024)	(0.024)
Well-educated mother			0.057**	0.058**	0.058**
			(0.027)	(0.027)	(0.027)
Mother's age			0.003	0.003	0.003
			(0.002)	(0.002)	(0.002)
Female teacher				-0.041	-0.041
				(0.029)	(0.029)
Constant	0.285*	-0.009	-0.110	-0.085	-0.085
	(0.164)	(0.278)	(0.284)	(0.280)	(0.280)
R²	0.035	0.038	0.042	0.044	0.044
Notes: OLS estimations and observations equal to 1,432 in all columns. Standard errors clustered at the school class level shown in parentheses. * p < 0.1, p < 0.05, * p < 0.01. All regressions include school and grade fixed effects. Intra-cluster correlation (ICC) was equal to 0.04.

Regarding the reasons for choosing the snacks, we observed that most of the participants said that they selected their snack because they like it more in general than the alternative (54%) or for taste reasons (19%). We did not estimate the effectiveness of the nutrition information provided, but identified that 16% of the subjects chose either of the snacks for health or nutrition reasons, in almost all cases this concerned the healthy snack. Surprisingly, students in the control group mentioned such aspects more often than those in the treatment group ((Prob > t = 0.005).

4.4. Consistency of choices

Table 5 depicts the impact of the meditation intervention on the likelihood of switching the original snack choice. We found that 11% of the sample changed their initial choice at the moment of snack collection. As expected, a slight majority of those changes (52%) were from the healthy to the unhealthy snack. However, the likelihood of a healthy-unhealthy switch did not differ by treatment status. On the other hand, overall switches were more likely among treated than for non-treated children, which was driven by changes from the unhealthy to the healthy option (+ 2.6%-points). 6% of the students who did such a “healthy switch” may have shown some biases (e.g., social desirability and/or experimenter demand effects), as they originally chose the chocolate bar because they do not appreciate certain sensory attributes (e.g., texture, taste) of amaranth snacks in general. Yet, the treatment effects were robust to the exclusion of those observations. Likewise, the results were robust to the exclusion of clusters where the protocol was violated (e.g., students saw the snack collection process on a previous day) and in a probit regression (Table E.2 in the Annex). Thus, our estimates did not support hypothesis 3: Result 3: Children practicing mediation do not make fewer dynamically inconsistent choices in the direction of unhealthy foods.

Table 5

Effect of meditation on time inconsistent food choices.
Dependent variable	Healthy switch^Y		Unhealthy switch^YY		All switches
	(1)	(2)	(3)	(4)	(5)	(6)
Meditation	0.027**	0.026**	0.007	0.005	0.034**	0.031**
	(0.011)	(0.011)	(0.013)	(0.012)	(0.016)	(0.015)
Girl		0.007		-0.023*		-0.016
		(0.013)		(0.012)		(0.016)
Age		0.014		0.005		0.020
		(0.012)		(0.012)		(0.017)
Well-educated mother		-0.007		-0.000		-0.007
		(0.016)		(0.013)		(0.021)
Mother's age		-0.001		0.001		-0.001
		(0.001)		(0.001)		(0.001)
Female teacher		0.025**		0.025		0.050**
		(0.012)		(0.017)		(0.020)
Constant	0.292***	0.188	0.102	0.019	0.394***	0.207
	(0.105)	(0.157)	(0.080)	(0.139)	(0.057)	(0.178)
R²	0.035	0.040	0.027	0.032	0.049	0.055
Mean (Control)	0.04		0.06		0.10
Notes: OLS estimations and observations equal to 1,432 in all columns. Standard errors clustered at the school class level shown in parentheses. ^YExchanges from the chocolate bar to the amaranth bar after the choice experiment. ^YYExchanges from the amaranth bar to the chocolate bar after the choice experiment * p < 0.1, p < 0.05, * p < 0.01. All regressions include school and grade fixed effects. Intra-cluster correlation (ICC) equal to 0.01 in all cases.

4.5. Heterogenous effects

First, we tested if the impact on the main secondary and primary outcomes differed by gender (see Table 6). The meditation treatment tended to have more positive effects for girls regarding indicators such as stress reduction and healthy snack choice, but the differences were always statistically insignificant. Second, we assessed if there were heterogenous effects by level of income (see Table 7). We highlighted that meditation reduced stress for students that attend both schools in lower- and higher-income areas, but the effect was larger in lower-income areas. The effects just mentioned remained significant after correcting for multiple hypothesis testing. Third, almost all students who chose their snack for health/nutrition reasons selected the healthy snack, so we explored the treatment effects for the rest of students (84% of the sample) choosing for other motives (see Table 8). Within this sub-sample, we observed that the meditation program was effective in increasing (+ 4%-points) the healthy snack choice. This effect was robust to the multiple hypothesis adjustments, exclusion of clusters where the protocol was violated and in probit regressions (Table E.2 in the Annex). We did not identify particular motives (e.g., past consumption, sensory attributes, dietary restrictions) that explained the mentioned effect (Table E.3 in the Annex).

Table 6

Heterogenous effect of meditation for girls.
Dependent variable	Stress reduction score^Y	Time inconsistent food choices		Healthy snack choice
Dependent variable	Stress reduction score^Y	Healthy switch	Unhealthy switch	Healthy snack choice
	(1)	(2)	(3)	(4)
Meditation	0.401***	0.034**	-0.009	-0.000
	(0.078)	(0.016)	(0.017)	(0.039)
p-value	0.000	0.038	0.581	0.994
FWER p-value	0.000	0.034	0.738	0.992
Girl	0.004	0.016	-0.039**	0.039
	(0.086)	(0.017)	(0.016)	(0.040)
Meditation x Girl	0.025	-0.016	0.029	0.014
	(0.103)	(0.025)	(0.024)	(0.056)
p-value	0.806	0.530	0.218	0.808
FWER p-value	0.932	0.785	0.319	0.932
Constant	0.675	0.181	0.031	-0.080
	(0.507)	(0.160)	(0.141)	(0.279)
R²	0.105	0.041	0.033	0.044
Notes: OLS estimations and observations equal to 1,432 in all columns. Standard errors clustered at the school class level shown in parentheses. * p < 0.1, p < 0.05, * p < 0.01. ^YConstructed as the first principal component of variables relaxed, happy, enjoys school, reductions in frequency of stomachache and emotional eating. All regressions include controls for age, mother completed tertiary education, mother's age, teacher's gender, school and grade fixed effects. * p < 0.1, p < 0.05, * p < 0.01, based on p-values shown above. FWER corrected p-values based on Clarke et al. (2019), with 3000 replications.

Table 7 Heterogenous effect of meditation for schools in lower-income areas.
Dependent variable	Stress reduction score^Y	Time inconsistent food choices		Healthy snack choice
Dependent variable	Stress reduction score^Y	Healthy switch	Unhealthy switch	Healthy snack choice
	(1)	(2)	(3)	(4)
Meditation	0.315***	0.014	0.008	0.011
	(0.076)	(0.014)	(0.015)	(0.030)
p-value	0.000	0.309	0.586	0.703
FWER p-value	0.001	0.629	0.802	0.802
Lower-income	0.012	0.215**	0.038	0.063
	(0.176)	(0.097)	(0.093)	(0.153)
Meditation x Lower-income	0.300***	0.035	-0.009	-0.015
	(0.106)	(0.024)	(0.025)	(0.055)
p-value	0.006	0.141	0.730	0.788
FWER p-value	0.018	0.316	0.914	0.914
Constant	0.522	-0.043	-0.015	-0.141
	(0.482)	(0.122)	(0.116)	(0.248)
R²	0.109	0.042	0.032	0.044
Notes: OLS estimations and observations equal to 1,432 in all columns. Standard errors clustered at the school class level shown in parentheses. * p<0.1, p<0.05, * p<0.01. ^YConstructed as the first principal component of variables relaxed, happy, enjoys school, reductions in frequency of stomachache and emotional eating. All regressions include controls for gender, age, mother completed tertiary education, mother's age, teacher's gender, school and grade fixed effects. * p<0.1, p<0.05, * p<0.01, based on p-values shown above. FWER corrected p-values based on Clarke et al. (2019), with 3000 replications.

Table 8 Heterogenous effect of meditation for students that chose for reasons other than health/nutrition
Dependent variable	Stress reduction score^Y	Time inconsistent food choices		Healthy snack choice
Dependent variable	Stress reduction score^Y	Healthy switch	Unhealthy switch	Healthy snack choice
	(1)	(2)	(3)	(4)
Meditation	0.379***	0.025*	0.014	0.040**
	(0.063)	(0.013)	(0.011)	(0.020)
p-value	0.000	0.058	0.208	0.048
FWER p-value	0.000	0.086	0.174	0.086
Constant	0.557	0.164	-0.043	-0.166
	(0.525)	(0.164)	(0.130)	(0.231)
R²	0.106	0.044	0.031	0.035
Notes: OLS estimations and observations equal to 1,200 in all columns. Standard errors clustered at the school class level shown in parentheses. * p<0.1, p<0.05, * p<0.01. ^Y Constructed as the first principal component of variables relaxed, happy, enjoys school, reductions in frequency of stomachache and emotional eating. All regressions include controls for gender, age, mother completed tertiary education, mother's age, teacher's gender, school and grade fixed effects. * p<0.1, p<0.05, * p<0.01, based on p-values shown above. FWER corrected p-values based on Clarke et al. (2019), with 3000 replications.

This paper analysed the impact of a meditation program shaped towards healthier food choices, on the dietary decisions of students in grades 4–6 at 27 schools in southern Mexico. In terms of overall preferences, our data showed that a considerable majority of the participants chose the chocolate over the amaranth bar in our experiment. This is aligned with the chocolate bar being more appealing to children in our sample than the amaranth bar. In general, our estimates revealed that students practicing meditation were not more likely to choose the healthy amaranth snack than those in the control group. Nonetheless, we highlighted that meditation effectively aided students who chose for reasons other than health or nutrition to increase the healthy snack choice. Thus, meditation is effective for children that are otherwise less health conscious. Based on the reasons of choice stated by the participants, we did not identify a particular attribute that explained this effect, such as past consumption, sensory qualities or dietary restrictions. In this sense, it is possible that meditation just gave them further “motivation” to take more effortful and deliberative decisions. The majority of these participants stated that they chose their snack because they “like it” more than the other, which is a vague response that may hide their true intrinsic motivations. On the other hand, the prices of the snacks used in the experiment were similar in the market, but it is unclear if students had different monetary valuations of the products and if they took them into account when doing their choice. We did not capture overall daily intakes of our participants, but a similar study found that meditation does not affect the overall intakes of unhealthy foods among adults (Alem et al., 2021). Unlike the latter study, we did find a significant effect on experimentally measured stated preference food choice.

In line with our expectations in the pre-analysis plan, our only significant treatment effect on food choice of 4%-points was lower than the effects in past school interventions. For the studies on incentives, the effect sizes varied from 11 to 28%-points increases in choices of fruits (Belot et al., 2016; Just and Price, 2013; Loewenstein et al., 2016), whereas those in the goal setting study varied from 10 to 18%-points increases in choices of milk (Samek, 2019). The effect in other school experiments in a middle-income country such as Indonesia, was also larger, around 13%-points for nudges (i.e., emoji faces) to increase fruit selections (de Vries Mecheva et al., 2021). While we did not measure the impact of information, based on the relatively low effect size, we posit that meditation may serve as an effective complement for other strategies. Further studies are warranted to disentangle the interaction effects of information and meditation, as sometimes information does not have a positive synergy with other strategies to improve dietary decisions among children as in de Vries Mecheva et al. (2021).

We predicted that meditation would also affect the dynamic consistency of food choice. Our results only indicated that students practicing meditation were more likely to exchange the unhealthy for the healthy snack (i.e., healthy switch) when given the chance to do that after the original choice. Children in general did more unhealthy than healthy switches, but the shares did not differ by treatment group. Compared to a study with a similar experimental design with adults, the share of participants in our research that exhibited dynamic inconsistencies was smaller (46%vs11%) (Sadoff et al., 2020). Had the second choice happened days apart from the experiment, unhealthy switches would be more prevalent, thereby giving more room for meditation to reduce the likelihood of such biases. In this sense, our results are similar to Alem et al. (2021), who found that meditation had weaker effects on time preferences.

Considering the mechanisms behind the effects of meditation, our analysis revealed that mediation reduced stress among treated participants. The effect was larger for those children attending schools in lower-income areas. Participants from such backgrounds may had more room to reduce stress levels, as they were likely experiencing higher stress than those from higher-income areas, due to long-term exposure to economic worries in their households (Bruijn, 2021; Kremer et al., 2019). Additionally, such students had prolonged school closures and lack of access to resources to attend online schooling, to cope with emotional impact of the pandemic (e.g., a therapist) and to a similar meditation program as in some of the private schools in the sample. This mechanism is backed by research linking stress with eating behaviours (Beenackers et al., 2018; Blissett et al., 2010), which we now generalize to the field of healthier diets in school settings. Again, our results are in line with Alem et al. (2021) in terms of the impact of meditation on perceived stress. Yet, further evidence is needed with alternative stress measurement techniques, as the mentioned authors had less conclusive evidence based on cortisol measures.

A few aspects are unclear in terms of the impact of meditation on our stress measures. First, the economic and pandemic-related stress sources were the most salient in the context of this study, but we are cognizant that there are other causes of worry among children such as schoolwork, bullying and violence in their communities. With this in mind, it is unclear which specific sources of stress were lessened by meditation. Second, there was an unexpected negative effect of the intervention on the reported frequency of stomach aches. Yet, assuming this indicator did not differ by treatment group at baseline, it is possible that the increased frequency was due to improved body awareness on treated participants, which is typically the case with meditation (Gallant, 2016). In fact, our meditation audios included practices were children mentally scanned their bodies. In any case, it is also possible that such techniques triggered pre-existing emotional issues that may need some guidance by an specialist, aspect that shall be recommended for future meditation interventions (de Lara Perez and Delgado-Rios, 2022).

Our findings are relevant for the ongoing debate in the field of food choices about the relative effectiveness of strategies influencing individual’s self-control such as ours, relative to more paternalistic approaches such as bans on unhealthy foods (de Vries Mecheva et al., 2021; Loewenstein, 2018). Here, we posit that strategies like meditation may well complement other approaches in a package of policies to attain healthier diets. In fact, the large availability of unhealthy foods in the market may actually be the consequence of the predominantly affective nature of consumers’ dietary behaviours, which is usually reinforced by stressors (Ruhm, 2012). Consequently, strategies influencing self-control are particularly relevant in low- and middle-income countries due to large shares of the population experiencing poverty-related stressors, which were complemented by the negative psychological effects of the COVID-19 pandemic. At the same time, the reality is that a comprehensive package of paternalistic policies such as large taxes and bans on unhealthy foods are hardly implemented in countries like Mexico, as they face strong push back from the food industry (Rivera-Dommarco et al., 2018). That being the case, meditation programs fit well in the existing student curricula and could be shaped towards the healthy choice to equip children to take better dietary decisions (Instituto Nacional de Salud Publica, 2020).

We highlight certain limitations of this work. The main results of this work cannot be generalized to Oaxaca or southern Mexico because the set of schools available for the study was constrained by the availability of open schools in the context of the COVID-19 pandemic. The sample of students was also biased becuase some minorities following only online schooling were not captured in our study. Parents of the latter children were typically more concerned with COVID-19 contagion and had the means, willingness to prolong such schooling scheme. Likewise, we had differential attrition at least in terms of gender, but other characteristics may have also varied. Altogether, we had a relatively large, heterogenous sample and expect that our results would hold for similar schools in low-income areas of Mexico, but this can only be assessed by replicating our design in other schools and other regions.

In terms of internal validity, we expect risks such as treatment diffusion to be limited, as the schedules by class differed within each school due to the pandemic regulations and it is unlikely that students would put other peers to meditate. On the other hand, we cannot discard experimenter demand effects particularly at the moment of snack collection[4]. To reduce the possibility of such biases, we switched the enumerators that monitored the meditation and those applying the survey and delivering the snacks, so they were less familiar with the experimenters when providing their responses (like the control group). We also emphasized to students that there were no right or wrong answers and that we just wanted their true opinions. In any case, the existing evidence has shown that such effects are modest and our results held when removing those observations deemed as the most suspicious of such biases (De Quidt et al., 2018; Mummolo and Peterson, 2019). Finally, we could not appropriately identify non-compliers in the treatment group so additional analyses such as intention to treat were not possible. Yet, it is certain that all treated participants in the analytical sample did the last meditation, thereby ensuring that they received to some extent the positive effects of the program.

In light of our findings, stress reduction strategies framed in the direction of healthier food choices improve dietary decisions among children that are otherwise not driven by health/nutrition motivations. Our study proved that with a large and heterogenous sample compared to similar studies in other strands of literature. In contrast with a similar experiment with adults in Alem et al. (2021), we also shed light on the higher chances of influencing food choices with meditation at the school-age stage. At the same time, strategies such as meditation may work as a self-built internal mechanism that leads to beneficial dynamic inconsistency in food choice. These strategies are especially of great assistance to improve the psychological well-being of children with a lower-income background, even more in pandemic times.

Funding

This research was funded by Wageningen University’s Development Economics Group and the Social Sciences Department (WASS). MEDV and MVB are grateful for the doctoral fellowship (217111) granted by the CONACYT and to the Netherlands – CGIAR Senior Experts Programme (SEP), grant number 17226; respectively. The funding agencies were not involved in the research activities.

Ethics approval statement

The authors declare that the mothers or tutors of participant children were fully informed about the details of the study and signed consent forms on behalf of their children. At the beginning of the intervention children were informed about their participation in the meditation program. Privacy rights of participants were observed during the research process. The research design received ethical approval from Wageningen University’s Social Sciences Ethics Committee (The Netherlands). The study was also approved by the State of Oaxaca’s Secretariat of Education (Mexico).

CRediT authorship contribution statement

Marcos E. Dominguez-Viera: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Writing-Original Draft, Writing-Review and Editing, Visualization, Supervision, Project administration. Margarita de Vries Mecheva: Methodology, Investigation, Writing-Original Draft, Writing-Review and Editing. Trang Nguyen: Methodology, Investigation, Writing-Original Draft, Writing-Review and Editing. Marrit van den Berg: Methodology, Writing-Review and Editing, Supervision, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.

Acknowledgements

We are grateful to Luis Quiroz, Diana Gutierrez, Rosalia Guerrero, Careli Lopez, David Soriano, Victor Sebastian, Alejandra Hernandez, Yamel de Escondrillas, Coral Mendoza, Alaide Vences, Talia Lerin and Amanda Ruiz for their excellent assistance on the coordination of the intervention, data collection and entry. At the institutional level, we are grateful to Charlynne Curiel from the Universidad Autonoma Benito Juarez de Oaxaca, Erisel Melchor and other staff members from the Instituto Estatal de Educacion Publica de Oaxaca, as well as with school supervisors Elvia, Felipa, Arcelia, Liney, Gustavo and Gerardo for their efforts on the facilitation of this research. We also appreciate the willingness to participate and engagement of school principals, teachers, parents and students. Finally, we thank participants who provided constructive inputs on the research design at seminars hosted by Wageningen University and Erasmus University Rotterdam.

Alem, Y., Behrendt, H., Belot, M., Bíró, A., 2021. Mind training, stress and behaviour — A randomised experiment. PLoS One 16. https://doi.org/10.1371/journal.pone.0258172
Augenblick, N., Niederle, M., Sprenger, C., 2015. Working Over Time: Dynamic Inconsistency in Real Effort Tasks. Q J Econ 1067–1115. https://doi.org/10.1093/qje/qjv020.Advance
Beenackers, M.A., Groeniger, J.O., van Lenthe, F.J., Kamphuis, C.B.M., 2018. The role of financial strain and self-control in explaining health behaviours: The GLOBE study. Eur J Public Health 28, 597–603. https://doi.org/10.1093/eurpub/ckx212
Belot, M., James, J., Nolen, P., 2016. Incentives and children’s dietary choices: A field experiment in primary schools. J Health Econ 213–229. https://doi.org/10.1016/j.jhealeco.2016.07.003
Blissett, J., Haycraft, E., Farrow, C., 2010. Inducing preschool children’s emotional eating: relations with parental feeding practices 1-3. Am J Clin Nutr 92, 359–65. https://doi.org/10.3945/ajcn.2010.29375
Brown, T., Moore, T.H., Hooper, L., Gao, Y., Zayegh, A., Ijaz, S., Elwenspoek, M., Foxen, S.C., Magee, L., O’Malley, C., Waters, E., Summerbell, C.D., 2019. Interventions for preventing obesity in children. Cochrane Database of Systematic Reviews 2019. https://doi.org/10.1002/14651858.CD001871.pub4
Bruijn, E. De, 2021. Poverty and economic decision making: a review of scarcity theory, Theory and Decision. Springer US. https://doi.org/10.1007/s11238-021-09802-7
Campion, J., Rocco, S., 2009. Minding the Mind: The Effects and Potential of a School-Based Meditation Programme for Mental Health Promotion. Adv Sch Ment Health Promot 2, 47–55.
Clarke, D., Romano, J.P., Wolf, M., 2019. The Romano-Wolf Multiple Hypothesis Correction in Stata.
CONEVAL, 2019. Diez años de medición de pobreza multidimensional en México: Avances y desafíos en política social. Mexico City.
Creswell, J.D., 2017. Mindfulness interventions. Annu Rev Psychol 68, 491–516. https://doi.org/doi:10.1146/annurev- psych-042716-051139
de Lara Perez, B., Delgado-Rios, M., 2022. Mindfulness-based programs for the prevention of childhood obesity: A systematic review. Appetite 168, 105725. https://doi.org/10.1016/j.appet.2021.105725
De Quidt, J., Haushofer, J., Roth, C., 2018. Measuring and bounding experimenter demand. American Economic Review 108, 3266–3302. https://doi.org/10.1257/aer.20171330
de Vries Mecheva, M., Rieger, M., Sparrow, R., Prafiantini, E., Agustina, R., 2021. Snacks, nudges and asymmetric peer influence: Evidence from food choice experiments with children in Indonesia. J Health Econ 79, 102508. https://doi.org/10.1016/j.jhealeco.2021.102508
DeCosta, P., Møller, P., Frøst, M.B., Olsen, A., 2017. Changing children’s eating behaviour - A review of experimental research. Appetite. https://doi.org/10.1016/j.appet.2017.03.004
Duckworth, A.L., Milkman, K.L., Laibson, D., 2018. Beyond Willpower: Strategies for Reducing Failures of Self-Control. Psychological Science in the Public Interest 19, 102–129. https://doi.org/10.1177/1529100618821893
Dudley, D.A., Cotton, W.G., Peralta, L.R., 2015. Teaching approaches and strategies that promote healthy eating in primary school children: a systematic review and meta-analysis. International Journal of Behavioral Nutrition and Physical Activity 12. https://doi.org/10.1186/s12966-015-0182-8
Flores, M., Macias, N., Rivera, M., Lozada, A., Barquera, S., Rivera-Dommarco, J., Tucker, K.L., 2010. Dietary patterns in Mexican adults are associated with risk of being overweight or obese. Journal of Nutrition 140, 1869–1873. https://doi.org/10.3945/jn.110.121533
Fredrickson, B.L., Arizmendi, C., Cappellen, P. Van, Fredrickson, B.L., Arizmendi, C., Same-, P.V.C., 2021. Same-day , cross-day , and upward spiral relations between positive affect and positive health behaviours. Psychol Health 36, 444–460. https://doi.org/10.1080/08870446.2020.1778696
Fudenberg, D., Levine, D.K., 2006. A Dual Self Model of Impulse Control. American Economic Review 96, 1449–1476. https://doi.org/10.2139/ssrn.888752
Gallant, S.N., 2016. Mindfulness meditation practice and executive functioning: Breaking down the benefit. Conscious Cogn. https://doi.org/10.1016/j.concog.2016.01.005
Gaona-Pineda, E.B., Martinez-Tapia, B., Arango-Angarita, A., Valenzuela-Bravo, D., Gomez-Acosta, L.M., Shamah-Levy, T., Rodriguez-Ramirez, S., 2018. Consumo de grupos de alimentos y factores sociodemográficos en población mexicana. Salud Publica Mex 60, 272–282. https://doi.org/10.21149/8803
Giuntella, O., Rieger, M., Rotunno, L., 2020. Weight gains from trade in foods: Evidence from Mexico. J Int Econ 122, 103277. https://doi.org/10.1016/j.jinteco.2019.103277
Haushofer, J., Mudida, R., Shapiro, J.P., 2021. The Comparative Impact of Cash Transfers and a Psychotherapy Program on Psychological and Economic Well-being. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3759722
Hill, D.C., Moss, R.H., Sykes-Muskett, B., Conner, M., O’Connor, D.B., 2018. Stress and eating behaviors in children and adolescents: Systematic review and meta-analysis. Appetite 123, 14–22. https://doi.org/10.1016/j.appet.2017.11.109
Instituto Nacional de Salud Publica, 2020. “Vida Saludable”, nueva materia en planes de estudio [WWW Document]. URL https://www.insp.mx/avisos/vida-saludable-nueva-materia-en-planes-de-estudio
Jensen, C.G., Vangkilde, S., Frokjaer, V., Hasselbalch, S.G., 2012. Mindfulness training affects attention-Or is it attentional effort? J Exp Psychol Gen 141, 106–123. https://doi.org/10.1037/a0024931
Just, D.R., Gabrielyan, G., 2018. Influencing the food choices of SNAP consumers: Lessons from economics, psychology and marketing. Food Policy 79, 309–317. https://doi.org/10.1016/j.foodpol.2018.03.003
Just, D.R., Price, J., 2013. Using Incentives to Encourage Healthy Eating in Children. Journal of Human Resources 48, 855–872. https://doi.org/10.1353/jhr.2013.0029
Kahneman, D., 2003. Maps of Bounded Rationality: Psychology for Behavioral Economics. American Economic Review 93, 1449–1475. https://doi.org/10.1128/AAC.03728-14
Kar, N., Kar, B., Kar, S., 2021. Stress and coping during COVID-19 pandemic: Result of an online survey. Psychiatry Res 295, 113598. https://doi.org/10.1016/j.psychres.2020.113598
Keats, E.C., Das, J.K., Salam, R.A., Lassi, Z.S., Imdad, A., Black, R.E., Bhutta, Z.A., 2021. Effective interventions to address maternal and child malnutrition: an update of the evidence. Lancet Child Adolesc Health 5, 367–384. https://doi.org/10.1016/S2352-4642(20)30274-1
Kremer, M., Rao, G., Schilbach, F., 2019. Behavioral Development Economics, in: Handbook of Behavioral Economics.
Landa-Blanco, M., Mejía, C.J., Landa-Blanco, A.L., Martínez-Martínez, C.A., Vásquez, D., Vásquez, G., Moraga-Vargas, P., Echenique, Y., Del Cid, G.M., Montoya, B.D., 2021. Coronavirus awareness, confinement stress, and mental health: Evidence from Honduras, Chile, Costa Rica, Mexico and Spain. Soc Sci Med 277. https://doi.org/10.1016/j.socscimed.2021.113933
Levy, T.S., Nasu, L.C., Dommarco, J.R., Hernández Avila, M., 2016. Encuesta Nacional de Salud y Nutrición de Medio Camino 2016: Informe final de resultados. Ministry of Health, National Institute of Public Health.
List, J.A., Samek, A.S., 2015. The behavioralist as nutritionist: Leveraging behavioral economics to improve child food choice and consumption. J Health Econ 39, 135–146. https://doi.org/10.1016/j.jhealeco.2014.11.002
Loewenstein, G., 2018. Self-Control and Its Discontents: A Commentary on Duckworth, Milkman, and Laibson. Psychological Science in the Public Interest. https://doi.org/10.1177/1529100619828401
Loewenstein, G., O’Donoghue, T., Bhatia, S., 2015. Modeling the interplay between affect and deliberation. Decision 2, 55–81. https://doi.org/10.1037/dec0000029
Loewenstein, G., Price, J., Volpp, K., 2016. Habit formation in children: Evidence from incentives for healthy eating. J Health Econ 45, 47–54. https://doi.org/10.1016/j.jhealeco.2015.11.004
Madden, G.J., Price, J., Sosa, F.A., 2017. Behavioral Economic Approaches to Influencing Children’s Dietary Decision Making at School. Policy Insights Behav Brain Sci 4, 41–48. https://doi.org/10.1177/2372732216683517
Maier, S.U., Makwana, A.B., Hare, T.A., 2015. Acute Stress Impairs Self-Control in Goal-Directed Choice by Altering Multiple Functional Connections within the Brain’s Decision Circuits. Neuron 87, 621–631. https://doi.org/10.1016/j.neuron.2015.07.005
Mancino, L., Guthrie, J., Just, D.R., 2018. Overview: Exploring ways to encourage healthier food purchases by low-income consumers—Lessons from behavioral economics and marketing. Food Policy 79, 297–299. https://doi.org/10.1016/j.foodpol.2018.03.007
Mason, A.E., Epel, E.S., Kristeller, J., Moran, P.J., Dallman, M., Lusting, R.H., Acree, M., Bacchetti, P., Laraia, B.A., Hecht, F.M., Daubenmier, J., J.J., D., 2016. Effects of a randomized controlled trial of a mindfulness-based intervention on cardiovascular responses to social evaluative threat in obese adults. J Behav Med 39, 201–213. https://doi.org/10.1007/s10865-015-9692-8.Effects
Messerli-Bürgy, N., Stülb, K., Kakebeeke, T.H., Arhab, A., Zysset, A.E., Leeger-Aschmann, C.S., Schmutz, E.A., Meyer, A.H., Ehlert, U., Garcia-Burgos, D., Kriemler, S., Jenni, O.G., Puder, J.J., Munsch, S., 2018. Emotional eating is related with temperament but not with stress biomarkers in preschool children. Appetite 120, 256–264. https://doi.org/10.1016/j.appet.2017.08.032
Michels, N., Sioen, I., Braet, C., Eiben, G., Hebestreit, A., Huybrechts, I., Vanaelst, B., Vyncke, K., De Henauw, S., 2012. Stress, emotional eating behaviour and dietary patterns in children. Appetite 59, 762–769. https://doi.org/10.1016/j.appet.2012.08.010
Mummolo, J., Peterson, E., 2019. Demand Effects in Survey Experiments: An Empirical Assessment. American Political Science Review 113, 517–529. https://doi.org/10.1017/S0003055418000837
O’Connor, D.B., Jones, F., Conner, M., Mcmillan, B., Ferguson, E., 2008. Effects of Daily Hassles and Eating Style on Eating Behavior. Health Psychology 27, 20–31. https://doi.org/10.1037/0278-6133.27.1(Suppl.).S20
Orona-Tamayo, D., Paredes-López, O., 2017. Amaranth Part 1-Sustainable Crop for the 21st Century: Food Properties and Nutraceuticals for Improving Human Health. Sustainable Protein Sources 239–256. https://doi.org/10.1016/B978-0-12-802778-3.00015-9
Osika, W., Friberg, P., Wahrborg, P., 2007. A new short self-rating questionnaire to assess stress in children. Int J Behav Med 14, 108–117. https://doi.org/10.1007/BF03004176
Popkin, B.M., Ng, S.W., 2021. The nutrition transition to a stage of high obesity and noncommunicable disease prevalence dominated by ultra-processed foods is not inevitable. Obesity Reviews 1–18. https://doi.org/10.1111/obr.13366
Popkin, B.M., Reardon, T., 2018. Obesity and the food system transformation in Latin America. Obesity Reviews 19, 1028–1064. https://doi.org/10.1111/obr.12694
Powell, E.M., Frankel, L.A., Hernandez, D.C., 2017. The mediating role of child self-regulation of eating in the relationship between parental use of food as a reward and child emotional overeating. Appetite 113, 78–83. https://doi.org/10.1016/j.appet.2017.02.017
Rastogi, A., Shukla, S., 2013. Amaranth: A New Millennium Crop of Nutraceutical Values. Crit Rev Food Sci Nutr 53, 109–125. https://doi.org/10.1080/10408398.2010.517876
Reijula, S., Hertwig, R., 2022. Self-nudging and the citizen choice architect. Behavioural Public Policy 6, 119–149. https://doi.org/10.1017/bpp.2020.5
Rivera-Dommarco, J.A., 2012. Deficiencias de micronutrimentos en México: Un problema invisible de salud pública. Salud Publica Mex 54, 101–102.
Rivera-Dommarco, J.A., Colchero, M.A., Fuentes, M.L., Gonzáles, T., Aguilar-Salinas, C.A., Hernández-Licona, G., Barquera, S. (Eds.), 2018. La obesidad en México: Estado de la política pública y recomendaciones para su prevención y control, First. ed. Instituto Nacional de Salud Pública, Cuernavaca, Mor.
Ruhm, C.J., 2012. Understanding overeating and obesity. J Health Econ 31, 781–796. https://doi.org/10.1016/j.jhealeco.2012.07.004
Sadoff, S., Samek, A., Sprenger, C., 2020. Dynamic Inconsistency in Food Choice: Experimental Evidence from Two Food Deserts. Review of Economic Studies 87, 1954–1988. https://doi.org/10.1093/restud/rdz030
Sala, M., Rochefort, C., Lui, P.P., Baldwin, A.S., 2020. Trait mindfulness and health behaviours: a meta-analysis. Health Psychol Rev 14, 345–393. https://doi.org/10.1080/17437199.2019.1650290
Salam, R.A., Padhani, Z.A., Das, J.K., Shaikh, A.Y., Hoodbhoy, Z., Jeelani, S.M., S., L.Z., Bhutta, Z.A., 2020. Effects of Lifestyle Modification Interventions to Prevent and Manage Child and Adolescent Obesity : Nutrients 12, 2208.
Samek, A., 2019. Gifts and goals: Behavioral nudges to improve child food choice at school. J Econ Behav Organ 164, 1–12. https://doi.org/10.1016/j.jebo.2019.05.008
Sapthiang, S., Van Gordon, W., Shonin, E., 2019. Health School-based Mindfulness Interventions for Improving Mental Health: A Systematic Review and Thematic Synthesis of Qualitative Studies. J Child Fam Stud 28, 2650–2658. https://doi.org/10.1007/s10826-019-01482-w
Schilbach, F., Schofield, H., Mullainathan, S., 2016. The Psychological Lives of the Poor. American Economic Review 106, 435–440. https://doi.org/10.1257/aer.p20161101
Shamah, T., Cuevas, L., Gaona, E.B., Gómez, L.M., Morales, M.C., Hernández, M., Rivera, J.Á., 2018. Sobrepeso y obesidad en niños y adolescentes en México, actualización de la Encuesta Nacional de Salud y Nutrición de Medio Camino 2016. Salud Publica Mex 60, 244–253.
UNICEF, 2021a. Los efectos nocivos de la COVID-19 sobre la salud mental de los niños, niñas y jóvenes son solo la punta del iceberg [WWW Document]. URL https://www.unicef.org/es/comunicados-prensa/efectos-nocivos-covid19-salud-mental-ninos-ninas-jovenes-punta-iceberg#:~:text=La alteración de las rutinas,y preocupación por su futuro.
UNICEF, 2021b. 3 de cada 10 niños, niñas y adolescentes en América Latina y el Caribe viven con sobrepeso [WWW Document]. URL https://www.unicef.org/lac/comunicados-prensa/3-de-cada-10-ninos-ninas-y-adolescentes-en-america-latina-y-el-caribe-viven-con-sobrepeso
Van Kleef, E., Van Trijp, H.C.M., 2018. Methodological challenges of research in nudging, in: Ares, G., Verla, P. (Eds.), Methods in Consumer Research, Vol. 1: New Approaches to Classic Methods. Woodhead, Cambridge, pp. 329–349. https://doi.org/10.1016/B978-0-08-102089-0.00013-3
van Rongen, S., Verkooijen, K., de Vet, E., 2019. Dealing with too little: The direct experience of scarcity does not affect snack intake. Appl Psychol Health Well Being. https://doi.org/10.1111/aphw.12163
Vogel, M., Geserick, M., Gausche, R., Beger, C., Poulain, T., Meigen, C., Körner, A., Keller, E., Kiess, W., Pfäffle, R., 2022. Age- and weight group-specific weight gain patterns in children and adolescents during the 15 years before and during the COVID-19 pandemic. Int J Obes 46, 144–152. https://doi.org/10.1038/S41366-021-00968-2
Willett, W., Rockström, J., Loken, B., Springmann, M., Lang, T., Vermeulen, S., Garnett, T., Tilman, D., DeClerck, F., Wood, A., Jonell, M., Clark, M., Gordon, L.J., Fanzo, J., Hawkes, C., Zurayk, R., Rivera, J.A., De Vries, W., Majele Sibanda, L., Afshin, A., Chaudhary, A., Herrero, M., Agustina, R., Branca, F., Lartey, A., Fan, S., Crona, B., Fox, E., Bignet, V., Troell, M., Lindahl, T., Singh, S., Cornell, S.E., Srinath Reddy, K., Narain, S., Nishtar, S., Murray, C.J.L., 2019. Food in the Anthropocene: The EAT-Lancet Commission on healthy diets from sustainable food systems. The Lancet 393, 447–492. https://doi.org/10.1016/S0140-6736(18)31788-4

For example, most lifestyle modification studies in the review by Salam et al. (81%, n = 531) were conducted in high-income countries. Within the same review, we identified only three studies focusing on cognitive self-deployed strategies that were conducted at schools in low- and middle-income countries.
Some teachers are in the classroom before recess, whereas some of them teach after recess.
For example, some students attended school two days per week on different days by class. It was also common that some students were present in the first half of the morning and the rest in the second half of the morning.
Such aspects may coincide with social desirability bias, as healthier food choices are an obvious normatively positive behaviour. Likewise, Hawthorne effects may be interconnected with the same biases at the moment of snack collection.

The authors declare no competing interests.

2SuppInformationMeditationFoodChoiceMexico.docx

Relaxed minds for healthier food choice at school: A field experiment in southern Mexico

Status:

Version 1

Abstract

Figures

1. Introduction

2. Hypotheses

3. Materials and methods

3.1 Sample

3.2. Products

3.3. Experimental design

3.4. Estimation strategy

4. Results

4.1. Sensory evaluation of the products

4.2. Manipulation checks

4.3. Evaluating the impact of meditation on food choices

4.4. Consistency of choices

4.5. Heterogenous effects

5. Discussion and conclusion

Declarations

References

Footnotes

Additional Declarations

Supplementary Files

Status:

Version 1