Correlation Neglect in Risky Choice

Last registered on February 06, 2024


Trial Information

General Information

Correlation Neglect in Risky Choice
Initial registration date
February 02, 2024

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
February 06, 2024, 5:17 PM EST

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


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

Request Information

Primary Investigator

Texas A&M University

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.
This study examines the effects of neglecting correlation in choice under risk. While alternative theories of choice under risk, such as regret aversion theory (Loomes and Sudgen, 1987) and salience theory (Bordalo, Gennaioli and Shleifer, 2012) suggest agents attend to the underlying correlation and derive utility from state contingencies, it remains an open question how people process these decisions. The repercussions of correlation neglect have paramount implications for the modern portfolio theory for risk-averse agents who seek to minimize variance for a given expected return (Markowitz, 1952). Relatedly, it may lead to missed hedging or arbitrage opportunities (Eyster and Weizsacker, 2016). How people process correlated risky choice in the evaluation of trade-offs between different options is largely endogenous though likely the decision process is influenced by the framing or presentation of lotteries. This experiment exogenously induces or suppresses subjects' focus towards the underlying correlation between lotteries through a dynamic display. The experiment will be augmented with eye-tracking measures to estimate the effects of decision heuristics on choices. While eye-tracking data are endogenous given that a subject elects how to disperse attention throughout the screen, the random treatment assignment will be used as an instrument to estimate the effects of these data given that they induce differential decision processes.
External Link(s)

Registration Citation

Vitaku, Valon. 2024. "Correlation Neglect in Risky Choice." AEA RCT Registry. February 06.
Experimental Details


The interventions I employ involve sequentially displaying the state space or lottery options when correlated. I compare these interventions to examine risk-taking behavior and decision-processes using eye-tracking techniques. This question is important as literature on insurance up-take has revealed people often and excessively over-insure. This experiment investigates one of the potential causes. Further it has significant implications in the division of consumer surplus and profits for insurance policymakers.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
risky choices
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
evaluation processes (eye-tracking)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Individual subjects are randomly assigned to one of two treatments (sequential lotteries, or sequential states). Subjects are initially faced with two practice (unincentivized) binary lottery choices presented in a random order so they can understand that lotteries are correlated. I elected to present lotteries in wheels since sectors of a circle provide a straightforward illustration of the state space. After subjects select a lottery, the two wheels spin at the same rate and stop at the same angle. Following practice rounds, subjects are faced with a one-shot risky choice problem which determines their experimental earnings. This problem is randomly selected for each subject from a list which contains two different problems.
Experimental Design Details
Not available
Randomization Method
Subjects randomly seated in the lab. Randomization into treatments done by a computer.
Randomization Unit
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
Sample size: planned number of observations
Sample size (or number of clusters) by treatment arms
150 subjects per treatment (sequential lotteries, sequential states)
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Subjects will be faced with a one-shot binary risky choice problem (randomly selected from a list that contains two problems) after undergoing 2 practice (unincentivized) problems in a random order. With 300 subjects (150 subjects/treatment arm) overall, and 150 subjects per choice problem (75 subjects/treatment arm), I am powered to detect a medium effect size (0.463) according to Cohen's d (two-sided, significance level=0.05, power=0.8) for each of the problems.

Institutional Review Boards (IRBs)

IRB Name
IRB Approval Date
IRB Approval Number
Analysis Plan

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

Request Information