Testing Undue Inducement

Last registered on January 28, 2022


Trial Information

General Information

Testing Undue Inducement
Initial registration date
January 27, 2022

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
January 28, 2022, 8:44 AM EST

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



Primary Investigator

University of Zurich

Other Primary Investigator(s)

Additional Trial Information

In development
Start date
End date
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Around the world, laws limit incentives for transactions such as human research participation, egg donation, or gestational surrogacy. A key reason is the notion of undue inducement---the conceptually vague and empirically largely untested idea that incentives cause harm by distorting individual decision making. Two experiments, including one based on a highly visceral transaction, show that incentives bias information search. Such behavior is consistent with both irrational and Bayes-rational behavior. I examine whether choices admit welfare weights on benefit and harm that justify permitting the transaction but capping incentives.
External Link(s)

Registration Citation

Ambuehl, Sandro. 2022. "Testing Undue Inducement." AEA RCT Registry. January 28. https://doi.org/10.1257/rct.8903-1.0
Experimental Details


The experiment consists of two parts, each of which proceeds in multiple rounds.
In each round of the first part, the subject decides whether to participate in a gamble that depends on
a state of the world s. The state is good (s = G) with prior probability μ or bad (s = B) otherwise. If
the subject participates in the gamble and the state is bad, she suffers a monetary loss L. If the state is
good, she does not suffer a loss. The subject receives the incentive payment m if she participates; otherwise
she receives nothing. Before deciding whether to participate, the subject chooses one of two information
structures I = {IG,IB} from which she observes a stochastic signal about the state. Information structure IG is statewise biased towards G relative to IB. Decisions in this part provide all information required to test UIH-behavioral (formalized in Proposition 1 and Corollary 1).
In the second part of the experiment I elicit the welfare benchmarks needed to test UIH-normative in an ex-ante welfare framework. In each round, the subject reveals her certainty equivalent for a lottery that leads to a gain g ≥ 0 with probability p or to a loss l < 0 with probability (1 − p). Unbeknownst to the subject, the parameters g,l, and p correspond to the participation decisions she faced in the first part of the experiment after having observed a signal from the information structure she had chosen. Crucially, the success probability p is equal to the posterior a Bayesian would have held at that stage. For instance, if, in the first part, the subject faced the incentive m and had observed a signal σ from information structure I, then the corresponding lottery in part two is given by g = m, l = −L + m, and p = γ_{σ,I}.

In part 1 of the experiment, the subject decides whether to accept a ‘venture’ that may either ‘succeed’ or ‘fail,’ in exchange for a ‘venture participation payment.’ If the venture succeeds, she can keep the venture participation payment, and no further consequences occur. If the venture fails, she must ‘pay damages’ and may or may not keep the venture participation payment (depending on treatment; see below). Before deciding whether to participate, the subject chooses between a ‘bold advisor’ (the participation-biased information structure) and a ‘cautious advisor’ (the abstention-biased information structure). She then observes a stochastic signal which reads either “The [type] advisor recommends: Participate in the venture!” (accompanied with a thumbs-up symbol on green background) or “The [type] advisor recommends: Don’t participate in the venture!” (accompanied with a thumbs-down symbol on red background). In order to alert the subject to the fact that a new state is drawn in each round, each round begins with a display of 20 red and green symbols signifying ventures that are successes and failures. She clicks a first time to hide the colors, and clicks three more times to shuffle the ventures (animation). A final click on a button randomly selects one of the ventures. The subject learns that the venture thus selected is hers for the round.

Part 2 of the experiment is framed neutrally. It simply describes the amount of money the subject can gain or lose from participating in each gamble, the corresponding success probability, and the certain amount of money the subject receives or loses if she refuses the gamble.

Subjects receive a completion payment of EUR110 to which gains are added and from which losses are discounted. The amount a subject can lose from participating in the gamble in the bad state is L = EUR100. Subjects face incentive amounts m ∈ {20, 30, 70, 80} euros. The set of information structures is given by IG = (1,0.5) and IB = (0.5,0).

Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
1. Choice of advisor as function of incentive m
2. Welfare as measured by the certainty equivalents from stage 2
Primary Outcomes (explanation)
Welfare equals the certainty equivalent elicited in stage 2 if the subject accepted the corresponding gamble, and equals zero of the subject rejected.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Experimental Design Details
Randomization Method
All randomization is within subject, done by the computer.

Clarification on random assignment: This is a within-subjects design. That is, some subjects first participate in treatment T1, then in treatment T2, and so on, up to T18. Other subjects proceed through these treatments in different orders. The order in which people proceed through the treatments is randomized. Every subject proceeds through all treatments.

Randomization Unit
Randomization is within subject.
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
70 subjects
Sample size: planned number of observations
18 accept / reject decisions from each subject.
Sample size (or number of clusters) by treatment arms
Not applicable; this is a within-subjects design.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials

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

Request Information

Institutional Review Boards (IRBs)

IRB Name
Human Subjects Committee of the Faculty of Economics, Business Administration, and Information Technology at the University of Zurich
IRB Approval Date
IRB Approval Number
OEC IRB # 2022-007


Post Trial Information

Study Withdrawal

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

Request Information


Is the intervention completed?
Data Collection Complete
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials