Clients’ Choice of Professionals and the Use of Artificial Intelligence: A Conjoint Experiment

Last registered on September 25, 2026
View Trial History

Pre-Trial

Trial Information

General Information

Title
Clients’ Choice of Professionals and the Use of Artificial Intelligence: A Conjoint Experiment
RCT ID
AEARCTR-0019772
Initial registration date
September 18, 2026

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
September 25, 2026, 9:57 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

Name
Tommaso Nannicini
Affiliation
Contact Primary Investigator

Other Primary Investigator(s)

Additional Trial Information

Status
In development
Start date
2026-09-23
End date
2027-06-30
Keywords
Labor, Other
Additional Keywords
AI
JEL code(s)
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
This study examines how professionals’ use of artificial intelligence affects clients’ choices relative to gender, experience, and practice size. An online conjoint experiment administered by SWG will involve 2,600 adults across legal, economic, technical, and medical services. Each respondent will complete four choices between hypothetical professional profiles with randomized characteristics, before answering other AI-related survey questions. After each choice, respondents will report the minimum discount required to switch to the rejected professional. We will estimate Average Marginal Component Effects separately by professional category and pooled across categories, comparing the magnitude of the AI effect with those of the other attributes. Discount responses will provide a complementary measure of preference intensity conditional on the initial choice.
External Link(s)

Registration Citation

Citation
Nannicini, Tommaso. 2026. "Clients’ Choice of Professionals and the Use of Artificial Intelligence: A Conjoint Experiment." AEA RCT Registry. September 25. https://doi.org/10.1257/rct.19772-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
A survey-based conjoint experiment examining how clients choose legal, economic, technical, and medical professionals. Respondents choose between hypothetical professional profiles with randomly varied characteristics, including the use of artificial intelligence.
Intervention Start Date
2026-09-23
Intervention End Date
2026-10-17

Primary Outcomes

Primary Outcomes (end points)
Selection of a professional profile in each forced-choice conjoint task (question E.1).
Primary Outcomes (explanation)
Each displayed profile is coded 1 if chosen and 0 otherwise, yielding eight profile-level observations per respondent. Average Marginal Component Effects measure how the randomized attributes affect selection probability. The central comparison concerns the magnitude of the AI effect relative to gender, experience, and practice size, estimated separately by professional category and pooled across categories.

Secondary Outcomes

Secondary Outcomes (end points)
Minimum discount required to switch to the rejected professional (E.2), including the share unwilling to switch at any discount.
Secondary Outcomes (explanation)
Responses comprise six ordered categories: no discount; up to 10%; about 25%; 50%; more than 50%; and unwilling to switch regardless of discount. Cumulative switching indicators are constructed across the five discount categories, with unwillingness to switch coded zero throughout. These measure preference intensity conditional on the initial choice.

Experimental Design

Experimental Design
An online conjoint experiment with 2,600 adults, comprising 650 respondents in each of four professional categories. Each respondent completes four choices between two hypothetical professionals. Profile characteristics are randomized, and the experiment precedes the survey’s AI-related questions.
Experimental Design Details
Not available
Randomization Method
Computerized randomization within the SWG survey. Attribute levels are initially drawn independently with 50/50 probabilities. The entire pair is redrawn if profiles match on all four dimensions. Duplicate first names or surnames within a pair are redrawn separately. Left/right position is randomized for each task; attribute-row order is randomized once per respondent and held fixed across tasks.
Randomization Unit
Professional profiles within each respondent’s choice tasks. Profile position is randomized by task, while attribute-row order is randomized by respondent.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
2,600 respondents, with 650 per professional category. Respondents are the clusters for statistical inference; there is no group-level treatment assignment.
Sample size: planned number of observations
2,600 respondents completing 10,400 choice tasks, yielding 20,800 profile-task observations. Each category comprises 650 respondents, 2,600 tasks, and 5,200 profile-task observations.
Sample size (or number of clusters) by treatment arms
There are no fixed respondent-level treatment arms. Each binary profile attribute is randomized with equal probability, yielding approximately 10,400 profile observations per attribute level overall, or 2,600 per level within each professional category.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
At 80% power and a two-sided 5% significance level, approximate minimum detectable effects are 5–6 percentage points per professional category and 2.5–3 percentage points pooled, allowing for within-respondent dependence. With a binary-outcome standard deviation of approximately 0.5, these correspond to 0.10–0.12 and 0.05–0.06 standard deviations, respectively.
Supporting Documents and Materials

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

Request Information
IRB

Institutional Review Boards (IRBs)

IRB Name
Internal Review Board of the Osservatorio delle Libere Professioni
IRB Approval Date
2026-09-15
IRB Approval Number
1/2026
Analysis Plan