Can experimental store predict Real-World Choices? An external validity study on fresh fruit and vegetables purchasing

Last registered on June 23, 2025

View Trial History

Pre-Trial

Trial Information

General Information

Title

Can experimental store predict Real-World Choices? An external validity study on fresh fruit and vegetables purchasing

RCT ID

AEARCTR-0016257

Initial registration date

June 23, 2025

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

June 23, 2025, 3:02 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country

France

Region

paris

Primary Investigator

Name

Daniela Lima Rente

Affiliation

EconomiX (UMR 7235), Paris Nanterre University and CNRS ; Interprofessional Technical Center for Fruit and Vegetables (CTIFL), France

Contact Primary Investigator

Other Primary Investigator(s)

Additional Trial Information

Status

In development

Start date

2025-06-24

End date

2025-07-11

Keywords

Behavior

Additional Keywords

Experimental economics, Experimental store, External validity

JEL code(s)

C93 ; C91 ; D91 ; D12

Secondary IDs

Prior work

This trial is based on or builds upon one or more prior RCTs.

Abstract

Laboratory studies offer rigorous control over experimental variables, ensuring high internal validity. However, their external validity is often questioned due to the artificial nature of the tested environments and the limited diversity of participants. In contrast, field studies allow for the observation of behaviors that are closer to real-world conditions, albeit with lower internal validity. To address these limitations, new methodological approaches have emerged. Early attempts relied on the use of 2D pictorial stimuli. Subsequent technological advances enabled the development of virtual reality, and more recently, immersive 3D virtual reality, providing more engaging environments and eliciting behaviors more similar to those observed in physical retail settings. Nevertheless, these technologies remain insufficiently realistic for certain products, particularly fresh fruits and vegetables, where direct sensory evaluation plays a crucial role.

In this context, the CTIFL established in 2021 the Experimental Sales Area (Zone Expérimentale de Vente, ZEV), a physical reconstruction of a fresh produce section, involving a panel of consumers representative of the French population. This study aims to assess the external validity of the ZEV by comparing the effect of a melon tasting on purchasing behaviors observed both in the laboratory (ZEV) and in a real store setting.

External Link(s)

Registration Citation

Citation

Lima Rente, Daniela. 2025. "Can experimental store predict Real-World Choices? An external validity study on fresh fruit and vegetables purchasing." AEA RCT Registry. June 23. https://doi.org/10.1257/rct.16257-1.0

Sponsors & Partners

Experimental Details

Interventions

Intervention(s)

The intervention consists of implementing a melon tasting setup across two experimental contexts:

- In the laboratory setting (CTIFL's experimental retail space): melon pieces are offered in a self-service format, presented on a tray placed directly on the display at the melon section.

- In the field setting (a fresh produce store in Paris): the same tasting setup is deployed, with strict standardization of the presentation format, tray location, and quantity of melon offered.

The objective is to assess whether providing sensory information (i.e., taste) prior to the purchase decision can reduce consumers’ risk aversion and positively influence their purchasing behavior.

Intervention (Hidden)

Intervention Start Date

2025-06-24

Intervention End Date

2025-07-11

Primary Outcomes

Primary Outcomes (end points)

Primary outcome variables:
- Melon purchase rate,
- Quantity of melon purchased,
- Amount spent on melon,
- Diversity of products purchased, including the purchase rate, quantity, and expenditure on other products,
- Post-purchase declarations, including perceived risk, reasons and barriers to purchase, and evaluation of the quality of the sampled melons.

These variables are designed to assess the impact of the intervention (tasting) on purchasing decisions, with particular attention to perceived risk aversion and consumer confidence in product quality.

Primary Outcomes (explanation)

The main outcome variables are directly observable (purchase incidence, quantity purchased, and amount spent), based on actual purchase records collected during the experimental sessions. In addition, more subjective dimensions — such as perceived risk or confidence in product quality — are constructed from post-purchase questionnaire responses. These dimensions are measured using Likert-type scales, based on statements related to trust or perceived risk associated with the product. An aggregate risk perception score will be constructed either by computing the mean of relevant items or through a dimensionality reduction technique such as Principal Component Analysis (PCA), depending on the statistical consistency of the items.Risk aversion levels are measured using an experimental lottery task.

These outcomes will allow for analysis of the intervention’s effect on behavioral drivers underlying purchase decisions. Furthermore, comparing results from the laboratory and field settings will provide insights into the external validity of findings generated in the CTIFL experimental store.

Secondary Outcomes

Secondary Outcomes (end points)

Secondary Outcomes (explanation)

Experimental Design

The experiment is based on a dual-context design: a laboratory experiment conducted in an experimental retail store (CTIFL), and a field experiment carried out in a specialized fresh produce store in the Paris region. In both settings, two experimental groups are defined: a control group (no tasting) and a treatment group (self-service tasting via a tray of melon pieces placed directly on the display). In the laboratory setting, participants receive an experimental budget to spend, while in the field setting, customers are surveyed after completing their purchases, under natural shopping conditions.

Purchasing decisions are recorded, along with stated motivations and barriers to purchase. The post-purchase questionnaire also include sociodemographic variables, individual risk aversion, perceived confidence in product quality, and participants’ preferences regarding sales techniques.

The objective is to evaluate the effect of tasting on purchasing behavior and to assess the external validity of results obtained in the controlled (lab) environment.

Experimental Design Details

Randomization Method

In the laboratory experiment, participants were recruited through a specialized panel provider, who was responsible for randomly assigning individuals to the experimental conditions (with or without tasting) and for assembling a sample representative of the French population.
In the field setting, all store customers were invited to participate after completing their purchases. Assignment to experimental conditions was based on a pre-randomized schedule of tasting and non-tasting sessions (on a daily basis), in order to minimize contextual biases.

Randomization Unit

In the laboratory, participants were randomly assigned to either the tasting or no-tasting condition. Those assigned to the same condition then participated together in group sessions.

In the field setting, randomization was implemented at the session level, defined by time slots within a given day. Each hourly session was randomly assigned to either the tasting or no-tasting condition, according to a pre-randomized schedule. This approach allows both conditions to occur on the same day while minimizing contextual biases related to time of day or customer flow.

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

Laboratory experiment: 20 sessions (10 sessions per treatment group, with 9 participants per session)
Field experiment: 36 sessions (3 sessions per day alternating between control and treatment conditions, over a period of 12 days)

Sample size: planned number of observations

Laboratory experiment: 180 participants (10 sessions × 9 participants × 2 treatments). Field: The objective is to recruit a total of 200 customers.

Sample size (or number of clusters) by treatment arms

Laboratory: 90 participants in control, 90 participants in treatment (tasting).
Field: 100 customers in control, 100 customers in treatment (tasting).

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Using the study design and effect sizes observed in the laboratory, the smallest effect we can reliably detect (Cohen’s w) for the main binary outcomes is about 0.051. A power calculation using a chi-square test with 1 degree of freedom, a significance level of 0.05, and a desired power of 0.8 indicates that a sample size of about 2,998 observations is required to detect this effect. However, due to budget constraints and logistical challenges related to recruitment in the field, we limited our sample size to approximately 180-200 participants. Given this actual planned sample size of 200 observations, the statistical power to detect an effect of this size is approximately 11%.

Supporting Documents and Materials

IRB