Adolescents Girls, Agency and Appetite: Evidence from a Food Voucher Experiment in High Schools in Indonesia

Classes of female adolescents aged 12-16 are randomized into three types of vouchers. Participants will receive the same voucher twice a week over a three week period.

The experiment and interventions will be conducted in 5 schools in Solo, Indonesia. A target sample of 1000 adolescent girls aged 12-16 will take part in the voucher intervention. Each participant will be randomized into one of three groups with equal chances: (i) control (non-food, stationary voucher of equal value to treatment vouchers), (ii) healthy food voucher (restricted voucher; treatment 1) and (iii) choice between health and unhealthy food voucher (choice-based voucher; treatment 2). Vouchers are personalized and will be handed out over a three week period (2 times a week). The food vouchers can be redeemed at food stalls on school grounds on the day. The stationary voucher of equal value can also be redeemed at school on the day. All groups receive basic information on healthy food choices.

2024-08-12

2024-08-29

Healthy meal choice

This binary outcome variable is based on each student's choice in a controlled free meal choice between a healthy vs. a relatively unhealthy option.

-Healthy foods or drinks consumed per day
-Unhealthy foods or drinks consumed per day

Consumption frequencies are collected via students' food diaries (featuring a list of typical meal/drink items that are classified into healthy vs. unhealthy by our nutrition co-PI). The diary is collected 6 times during the intervention period. We will use the number of healthy as well as the number of unhealthy meals/drinks consumed per day as dependent variables.

Classes are randomly allocated to the three experimental voucher conditions. We will collect baseline survey and controlled meal choices, as well as two rounds of follow-up controlled meal choices. We will also collect body height and weight at baseline, as well as weights at follow ups. We will also collect food diaries during the intervention period.

Procedures and Activities:

There are four activities in the project namely baseline survey, intervention followed by two times follow-up surveys.

1) Baseline survey: to be conducted in early August 2024 before the intervention. We will first collect original baseline data on food consumption and habits and participant characteristics/traits. We will also perform a free meal choice experiment to measure preferences. In terms of anthropometrics, the subject's height and weight measurements will also be carried out in this last survey. Height measurement is done with microtoise with an accuracy of 0.1 cm, and weight measurement with OMRON digital scales with an accuracy of 0.1 cm.

2) Intervention: after randomly splitting the sample of classes into three roughly equal intervention groups, non-food voucher (control), food voucher treatment 1- restricted and treatment 2- choice-based. We will conduct interventions twice a week for three weeks in August 2024 (6 times total).

3) Follow-up survey 1 will be conducted once in September 2024. Subjects will be asked to join the controlled free meal choice experiment. This is done to examine the subjects' food preferences after being given the intervention for three weeks.

4) Follow-up survey 2 will be conducted once in November 2024. Subjects will be asked to join the meal choice experiment again. This is done to examine the subjects' food preferences over a longer time horizon.

Sample and Treatments:

Our planned sample consists of 1000 adolescents girls aged 12-16 years in 5 purposively selected high schools in Solo, Indonesia. There are three experimental groups with personalized vouchers: The first group is a control group that receives non-food voucher that can be used to buy stationery at school that day. The second group, treatment group 1, receives restricted food vouchers that can be used to buy healthy food in the school cafeteria that day. The third group, treatment group 2, receives choice based food vouchers with options that can be used to buy healthy and less / unhealthy food in the school cafeteria that day. Treatment group 2 is asked to make a choice (ticking the choice on the voucher).

All subjects are asked to fill in a food diary of food consumption on the day of the intervention. Note that vouchers will be given twice a week for three consecutive weeks. Thus, the intervention will be carried out for three weeks. All subjects also receive basic information on a healthy and unhealthy consumption choices and a food stall at the cafeteria.

The research team will work with the school cafeteria to provide healthy and unhealthy meals based on Isi Piringku's (My Plate) recommendations from the Ministry of Health. Based on the recommendations of Isi Piringku, healthy meals include: carbohydrates, vegetables as much as the amount of carbohydrates, fruit, animal/vegetable side dishes; less healthy meals include: carbohydrates, small vegetables/fruits, animal/vegetable side dishes; Unhealthy meals include: carbohydrates, animal/vegetable side dishes with no vegetables/fruits at all.

Predictions:

We will intervene in the food environment by providing restricted vouchers, making healthy food more affordable. We assume that current utility is non-separable and is determined by past and current choices. The restricted food voucher will therefore form habits into healthier food at follow-up. The choice-based food voucher facilitates adolescents to exert agency as they have control to decide whether they use the voucher to redeem for healthy or unhealthy foods. This treatment is inspired by Brownback et al., (2023, RESTAT) suggesting that agency may “encourages more goal-oriented behavior, which should further increase demand for healthy food.”

We are thus testing two hypotheses to measure the effectiveness of policy interventions for altering female adolescents' healthy eating behavior as proxied by our primary outcome variable:

1.Food subsidies in the form of food vouchers (treatment 1 or 2) increase healthy meal choices at follow-up (compared to control).
2.The choice-based food voucher is relatively more effective than the restricted voucher as it triggers agency.

Our primary outcome variable to test these hypotheses is the student's choices in our controlled free meal choice experiment (two follow-ups). We will present pooled and separate short and longer term effects. We expect treatment effects (if any) to be qualitatively weaker in follow-up round 2.

In addition, we will test for corresponding impacts on healthy and unhealthy consumption frequencies using food diaries (during the intervention period only). The effects on this secondary outcome are ambiguous and depend on budget constraint changes.

Models:

We will present simple differences in means (control is the excluded category) using linear regression models pooled and for each follow up. We will also present adjusted estimates controlling for baseline meal choice, age, BMI z-score, as well as school dummies. Standard errors will be clustered at the class-level. In a robustness test, we will explore the presence of spillovers using information on the experimental assignment of the three best friends. We will test for experimental balance on baseline meal choice, age and BMI z-score. Missing covariates are replaced with sample means and this extrapolation is accounted for with a missing value dummy.

Extensions:

In correlational and exploratory tests of mechanisms, we will further examine if treatment effects vary with the following baseline covariates:
-Views on fruits and vegetables (health, taste, snack, costs)
-Time and risk preferences (survey-based, scale 0-10)
-Psychological need satisfaction using Activity-Feeling States (AFS)
-Agency engagement
-Personality traits
-Children’s Social Desirability Short Scale

We will test for above mechanisms via simple interactions terms in our linear regression models.

Randomization is done in office using Stata 18.5 based on the baseline data we collected. We initially registered an individual randomization stratified by class. After collecting the baseline data, we observed a low ICC of 0.01046 at class level, which allowed us for a cluster randomization (MDEs of individual and clustered randomization are qualitatively similar). We thus updated our power calculations below and randomized 70 classes (instead of individuals) into control, treatment 1 and treatment 2 stratified by school. This change in randomization approach is meant to better guard against spillovers and make administration of vouchers easier. After randomizing classes into three groups, we checked balance on the meal choice (primary outcome), as well as age, BMI for age, weight, height and class size. We detected no significant differences at conventional levels.

We randomize classes stratified by school.

Yes

70 classes located in 5 schools as included in baseline data.

1011 students as included in baseline data.

The targeted 1011 baseline participants in 70 classes in 5 schools are allocated to the three groups in a roughly balanced way (stratified by schools): 328 students (23 classes) in control group, 337 students (23 classes) in treatment 1 (restricted voucher) and 346 students (24 classes) in treatment 2 (voucher with choice).

Using our baseline data we performed an updated power calculation for a clustered randomization (instead of an individual one as planned). Taking the baseline healthy choice of 0.2275, an average of 14 kids per class, 23 classes in control and treatment each, we use the following command in STATA: power twoproportions 0.2275, m1(14) m2(14) rho(0.01052) power(0.8) k1(23) k2(23). We are powered for a MDE of 0.1055. Inclusion of covariates including baseline meal choice may further increase power.