Selection in strategy formulation: The impact of domain salience

Last registered on June 20, 2025

Pre-Trial

Trial Information

General Information

Title
Selection in strategy formulation: The impact of domain salience
RCT ID
AEARCTR-0016229
Initial registration date
June 16, 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 20, 2025, 11:36 AM EDT

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

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Affiliation

Other Primary Investigator(s)

PI Affiliation
Rotterdam School of Management
PI Affiliation
ESADE Business School
PI Affiliation
Bocconi University

Additional Trial Information

Status
On going
Start date
2025-05-21
End date
2026-06-20
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Entrepreneurs translate past experiences and knowledge – what we call individual domains – into “theories of value” that shape strategic search, i.e., the problems they choose to address, as well as the development and selection of strategic alternatives (Camuffo, Gambardella and Pignataro, 2024; Felin and Zenger, 2017; Wuebker et al., 2023). Past research shows that decision makers overweight information that happens to be contextually prominent, or salient, while overlooking equally relevant but latent cues (Bordalo et al., 2012, 2022; Chetty et al., 2009; Kahneman, 2003; Tiefenbeck et al., 2018; Tversky and Kahneman, 1974). Yet, despite theoretical claims that salience shapes strategic reasoning (Felin et al., 2023; Gavetti, 2012), there is no experimental evidence on whether deliberately making a domain salient alters strategy formulation or firm performance. Moreover, we do not know (i) whether latent domains can be activated through a short training intervention; (ii) how such activation changes the mix of domains and attributes strategists incorporate in their theories; or (iii) the downstream consequences for firm performance.

Our theoretical framework treats domain salience as a general cognitive mechanism. To obtain a clean test we concentrate on one domain – environmental sustainability – because its attributes are recognisable, have wide applicability to different sectors or business models, and are economically consequential. We design an intervention embedded within a standard university startup training program, where an instructor will present examples of start-ups whose theory of value hinges on sustainability attributes, i.e. the “sustainability exposure treatment”. Slides and speaking notes will illustrate how environmental sustainability attributes map onto each firm’s theory of value, making the sustainability domain salient to strategists. Matched control cases will be exposed to very similar startups in terms of products, business model etc, but omit any cues related to sustainability.

We will implement a set of randomised controlled trials across different startup training programmes (e.g. in Spain, Italy and The Netherlands), targeting to enrol a total of 450 early-stage start-ups. We choose to focus on entrepreneurs because they act as the primary strategists in early-stage ventures. Their choices are central to startup development and success, making them ideal subjects for examining how domain salience affects strategic exploration. All participants will receive training on theory-based experimentation to help them describe and present their theories of value (Camuffo, Gambardella, Messinese, et al., 2024). Next, half will be randomly assigned to the sustainability exposure treatment, the rest to the placebo.

By tracking textual changes in venture theories, pivoting behaviour and economic performance over twelve months, we will answer three primary research questions: (RQ1) Can a relatively simple intervention render a domain salient? (RQ2) Does induced salience shift entrepreneurs’ strategic framing and theories of value? (RQ3) Does that shift translate into short- and long-run performance effects?
This multi-country design delivers the first causal test of domain-salience theory in strategic management and entrepreneurship and clarifies how domains become encoded in entrepreneurs’ theories of value.
External Link(s)

Registration Citation

Citation
Berchicci, Luca et al. 2025. "Selection in strategy formulation: The impact of domain salience." AEA RCT Registry. June 20. https://doi.org/10.1257/rct.16229-1.0
Experimental Details

Interventions

Intervention(s)
We will randomly allocate participants to receive the sustainability domain exposure treatment. The main treatment will consist of of an in-class lecture delivered on the second part of the training program’s two-day, in-person bootcamp. The treatment will be delivered by two entrepreneurship lecturers who are unrelated to the research team and ‘blind’ to the RCT hypotheses.

The intervention will consist of a two-hour lecture with examples of start-ups whose theory of value hinges on the sustainability domain. Slides and speaking notes will illustrate how environmental sustainability attributes map onto each firm’s theory of value, making the sustainability domain salient to strategists. The control group will receive a very similar class in terms of duration and content. Control participants will be exposed to identical startups in terms of products, business model, performance etc, but the description will omit any cues related to sustainability.

Following the bootcamp, we will provide a treatment “reminder” in the form of an additional masterclass delivered one month later. During the reminder treatment, the lecturers will again provide examples of startups whose theory of value relies on the sustainability domain to the treatment group and identical examples without any mention of sustainability to the control group. Lastly, we will share a video to act as an additional reminder. The treatment group will receive a short video showcasing the training program in which the sustainability-related content will be clearly visible. The control group will receive an identical video in duration and style, but without any sustainability cues.

In order to avoid contamination across participants receiving the treatment, we will set up independent cohorts and run independent classes. We will have a total of 4 cohorts, two receiving the treatment and two acting as controls. The lecture halls will be in different parts of the campus to minimize interaction across participants in different experimental conditions. In order to account for the influence of lecturer teaching style and motivation, we will have both instructors teaching both experimental conditions.
Intervention Start Date
2025-06-20
Intervention End Date
2025-07-24

Primary Outcomes

Primary Outcomes (end points)
• Sustainability domain use: whether a sustainability attribute is included in an entrepreneur's theory of value
• New domain use: whether an attribute (attributes) coming from a new domain (or multiple new domains) is (are) included in an entrepreneur's theory of value
• Performance: key indicators of startup development (e.g., survival, sales, employees, and funding received)
Primary Outcomes (explanation)
To trace the evolution of entrepreneurs’ theories of value and start-up performance, we will implement a series of structured data-collection waves. We will collect information through:
1. the recruitment survey (demographics, sector, performance etc)
2. the pre-treatment survey (theory description and confidence, performance)
3. the 1st post-treatment survey (theory description)
4. the 2nd post-treatment survey (theory description, performance)
5. the 3rd post-treatment survey (performance)
6. the 4th post-treatment survey (performance)

The primary outcomes that we will analyse in this study include:
• Sustainability domain use: this measure will be generated based on the theory descriptions we will collect. In practice, we will train all participating entrepreneurs with the theory-based experimentation framework and ask them to provide a textual and schematic description of their startup’s theory of value, explaining the key attributes that are present and the causal links between them. This will allow us to create an indicator variable equal to one when the startup theory includes at least one attribute drawn from the sustainability domain (i.e. it is related to environmental sustainability). To identify which attributes are drawn from the sustainability domain, we will use keyword-based text search, where we will identify which attributes refer to sustainability concepts (e.g. “recycling”, “energy saving”, “renewable energy”, “environmental-friendly”, etc). We will also measure sustainability domain use at the intensive margin, i.e. the percentage of all attributes that are drawn from the sustainability domain.
• New domain use: this measure will be generated again based on the theory descriptions we will collect, and is defined as a relative measure in comparison with the initial theory description. So, this variable captures the use of new domains (other than the sustainability domain) following our treatment. To identify new domain use, we need to first categorize all theory attributes into domains. For that, we will rely on machine learning techniques – in particular classifiers such as LLMs – that will use the corpus of startup descriptions to categorize them into different domains (Dell, 2025). Once we do this exercise, we will define an indicator variable equal to one when a new domain is used. We will also measure new domain use at the intensive margin, i.e. the percentage of all attributes that are drawn from new domains. As an alternative approach, we will use LLM-based encoders (Dell, 2025) to compare the pre- and post-treatment theory descriptions and assign a similiarity score between them. A low similarity score likely indicates the use of new domains. We will manually verify both approaches to ensure that our measures reflect the use of new domains.
• Performance: we will measure performance using key indicators of startup development: e.g., survival, sales, employees, and funding received. These will be measured both at the extensive (e.g. an indicator variable equal to one if the startup has achieved sales in the past months) and intensive margin (e.g. the amount of sales in the past months). We also plan to use experienced startup evaluators or investors and ask them to rate the startup’s chances of success and the likelihood of investing in these.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
The treatment will be delivered as part of the program’s in-class training. For this reason, we will create 4 cohorts, two of which will receive the sustainability domain exposure treatment and two will receive the placebo.

Participants will be assigned to the 2 experimental conditions (treatment group vs. control group) by stratified randomization, with language used as stratum. This is because we will use English as a language of instruction as well as the local language to ensure the content and treatment is effective for those that are not fluent in English. So, each experimental condition will have a cohort that is instructed in English and a cohort instructed in the local language.

We will perform randomization checks to compare mean values of key observable entrepreneur and startup characteristics across the two groups.
Experimental Design Details
Not available
Randomization Method
Stratified randomization (language as stratum)
Randomization Unit
Individual
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
N.a.
Sample size: planned number of observations
Sample size according to power calculations: 210 participants. Planned sample size: 450 participants (150/ RCT location). Within each location, participants are split into 4 cohorts (2 treated, 2 control).
Sample size (or number of clusters) by treatment arms
225 participants in treatment group, 225 participants in control group.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
0.50 Power calculations were conducted for an individually randomised trial that compares the mean outcomes of treatment and control groups, focusing on entrepreneurs’ performance and theory development. With equal allocation, a two–sided α = 0.05 and 95% power (α err prob = 0.05 Power (1-β err prob) = 0.95), and assuming a moderate effect size (d=0.50), the total sample requirement is 210 participants. We use a medium effect size as a relatively conservative benchmark: the intervention is far more intensive than a simple nudge, and founders join the accelerator precisely to reshape their ventures, so at least a modest treatment impact is plausible. An additional advantage in our context is that we have repeated measures for some of our depedent variables and we will be able to perform longitudinal analyses, a fact that reduces the sample requirements. However, we have to account for the heterogeneity analysis that we need to undertake which should require – roughly speaking – splitting our sample in half. Given this, we opt for a sample size of 450 participants, which is more than double than the one estimated based on the original power calculations.
IRB

Institutional Review Boards (IRBs)

IRB Name
ESADE Research Ethics Committee Report
IRB Approval Date
2025-06-02
IRB Approval Number
018/2025
Analysis Plan

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information