Failure to Accept Good Advice

Last registered on June 18, 2022

Pre-Trial

Trial Information

General Information

Title
Failure to Accept Good Advice
RCT ID
AEARCTR-0009423
Initial registration date
June 17, 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
June 18, 2022, 10:24 AM EDT

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

Locations

Region

Primary Investigator

Affiliation
ESMT Berlin

Other Primary Investigator(s)

PI Affiliation
University of Warwick
PI Affiliation
University of Warwick

Additional Trial Information

Status
In development
Start date
2022-06-23
End date
2022-09-23
Secondary IDs
Prior work
This trial is based on or builds upon one or more prior RCTs.
Abstract
There are many reasons why individuals ignore advice. Many of these reasons will relate to the quality of the advice, but even after accounting for uncertainty about advice it may be that some continue to fail to accept advice even when it is clear the advice benefits them. Our study focuses on exactly that setting. In a controlled environment we ensure (a) that adopting advice improves the expected pecuniary payoff and (b) that this is clear to potential advisees. In this way we can be confident that anyone ignoring advice knowingly damages their pecuniary payoff. We also wish to understand whether and how the domain of the advice matters (we look at luck vs effort) and whether the characteristics of the advisee matters (we consider a number of behavioral factors and demographics). This experiment is part of a longer-running project that seeks to explore when individuals opt to continue with behavior that is likely to damage their utility.
External Link(s)

Registration Citation

Citation
Cibik, Ceren, David Ronayne and Daniel Sgroi. 2022. "Failure to Accept Good Advice." AEA RCT Registry. June 18. https://doi.org/10.1257/rct.9423
Experimental Details

Interventions

Intervention(s)
Intervention Start Date
2022-06-23
Intervention End Date
2022-09-23

Primary Outcomes

Primary Outcomes (end points)
The proportion of subjects who ignore good advice.
Primary Outcomes (explanation)
Subjects select either "accept their advice" or "ignore their advice" on the main page of the experiment, in all treatments. If they select "ignore their advice" AND would have received a higher expected pecuniary payoff by doing so, then we say they ignored good advice.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary: Susceptibility to the sunk cost effect; dispositional envy; stubbornness; IQ; gender.

Tertiary: other demographics.
Secondary Outcomes (explanation)
Susceptibility to the sunk cost effect will be measured using the scale developed in Ronayne, Sgroi & Tuckwell (2021).

Dispositional envy will be measured using the scale of Smith et al. (1999)

Stubbornness will be measured using the scale of Wilkins (2015).

IQ is measured using 10 Raven's Progressive Matrices.

For details, see the transcript attached to this entry.

References:
Ronayne, Sgroi & Tuckwell (2021), "Evaluating the sunk cost effect." Journal of Economic Behavior & Organization 186, 318-327.
Smith et al. (1999), "Dispositional envy." Personality and Social Psychology Bulletin 25(8), 1007–1020.
Wilkins (2015), “Signs that you’re being too stubborn.” Harvard Business Review, May 21.

Experimental Design

Experimental Design
We seek to determine whether individuals take good advice. We provide subjects with advice that carries a clear signal of its worth then ask them to accept or ignore the advice.

Subjects complete a task (either luck or effort) for which they are told they can receive a bonus payment of up to $2.50 depending on the quality of their answers. After the task, we tell them their score (out of five) and then introduce “the Adviser”, another subject who completed the same task some time ago and scored 4/5. We explain the Adviser agreed to offer their choices as advice to others and ask the current subject if they would like to accept or ignore the advice. If they accept, the Adviser’s score of 4/5 determines their payoff. If they ignore, their own score determines it. In all treatments, unless the subject also scores 4/5 (which makes the options payoff equivalent), either “accept” or “ignore” is strictly dominant in terms of the pecuniary payoff.

See the attached documents (the transcripts and the experimental detail and analysis plan document), which include more detail regarding the questions asked, treatments, subject recruitment and exclusion criteria.
Experimental Design Details
Not available
Randomization Method
Randomization is done by Qualtrics
Randomization Unit
Individual
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
n/a
Sample size: planned number of observations
Batch 1 (control and treatment 1): 750 Batch 2 (treatments 2 and 3): 282 Any small deviations from these exact figures will be due to sampling discrepancies generated by the employed software or the subject pool operator.
Sample size (or number of clusters) by treatment arms
Batch 1: 375 in each treatment
Batch 2: 141 in each treatment
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Our primary interest is to gain an estimate of the proportion of subjects ignoring good advice. That does not require a statistical test. We base our power calculation on a secondary item of interest: a two-sided difference in proportions test under the null that the proportions who ignore good advice in our treatments regarding task type are the same. The relevant assumptions, Stata command used and its result are below. Power: 0.8 Alpha: 0.05 P1: 0.21 (from a pilot) P2: 0.12 (from a pilot) Stata command: "power twoproportions 0.21 0.12" Result: N = 532 (N/2 = 266 per treatment) This is rounded up to N=750 to account for the application of exclusion criteria based on failure of attention or comprehension checks (see attached analysis plan for details) or because they do not face good advice (if they score the same or better than the adviser). [A pilot suggested that one or more of these criteria are met about 25% of the time, implying N=532/0.75=709. To be confident in securing adequate numbers, we will collect N=750 (375 per treatment).] Our second batch of data collection concerns two supplementary treatments. In lieu of estimates for in those settings, we scale the sample sizes in line with the power calculation for the main experiment to obtain 200 subjects (100 per treatment) after exclusion criteria are applied, giving N = 200*(750/532) = 282.
Supporting Documents and Materials

Documents

Document Name
Transcript 1
Document Type
survey_instrument
Document Description
Transcript for the control and treatment 1 groups
File
Transcript 1

MD5: 0bdf1c82657ada0c30dfffc66d4d3d4b

SHA1: aeadd7d94b9f8ecc9038f5889ca6afac98e6d364

Uploaded At: June 17, 2022

Document Name
Transcript 2
Document Type
survey_instrument
Document Description
Transcript for treatment groups 2 & 3
File
Transcript 2

MD5: d18ec90988a71098f3a9884f20ab5a54

SHA1: da52665c0caef7cbff54a66836fc783261c4de69

Uploaded At: June 17, 2022

IRB

Institutional Review Boards (IRBs)

IRB Name
ESMT Ethics Committee
IRB Approval Date
2022-02-03
IRB Approval Number
22-001-E
IRB Name
Humanities and Social Sciences Research Ethics Committee
IRB Approval Date
2022-01-12
IRB Approval Number
HSSREC 55/21-22
Analysis Plan

Analysis Plan Documents

Experimental Overview and Analysis Plan 170622.pdf

MD5: ad85efe0221d19c094d9c7cf05d340cf

SHA1: 1293e59057206b248ecdbefdee9b7b5395d75e7f

Uploaded At: June 17, 2022