x

NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Task Importance and Herd Behavior
Last registered on July 22, 2019

Pre-Trial

Trial Information
General Information
Title
Task Importance and Herd Behavior
RCT ID
AEARCTR-0004342
Initial registration date
July 13, 2019
Last updated
July 22, 2019 1:04 PM EDT
Location(s)
Region
Primary Investigator
Affiliation
Other Primary Investigator(s)
PI Affiliation
Swarthmore College
Additional Trial Information
Status
In development
Start date
2019-07-13
End date
2019-07-31
Secondary IDs
Abstract
Social conformity and information-based herding have been studied extensively in the social sciences, but there is little experimental evidence on the relationship between task importance and herd formation. In the upcoming experiment, we aim to replicate the main findings of a previous experiment on herd behavior, using some minor design tweaks, to both ensure the robustness of our prior work and to more explicitly test (in a pre-registered manner) some hypotheses motivated by exploratory findings from our previous study.
External Link(s)
Registration Citation
Citation
Bhanot, Syon and Charles Williamson. 2019. "Task Importance and Herd Behavior." AEA RCT Registry. July 22. https://doi.org/10.1257/rct.4342-2.0.
Former Citation
Bhanot, Syon, Charles Williamson and Charles Williamson. 2019. "Task Importance and Herd Behavior." AEA RCT Registry. July 22. http://www.socialscienceregistry.org/trials/4342/history/50465.
Experimental Details
Interventions
Intervention(s)
Our experiment takes place in two stages. All individuals are treated symmetrically in the first stage. The intervention occurs in the second stage of the experiment, where individuals will be randomly assigned to one of four tracks determining the value of each trivia question. In addition, subjects will receive herd information in the second stage.
Intervention Start Date
2019-07-22
Intervention End Date
2019-07-29
Primary Outcomes
Primary Outcomes (end points)
Given the two-stage nature of this study, it is important for us to protect the integrity of the experiment. Therefore, we are withholding details from the public portion of this preregistration and reserving those details for the hidden portion. We describe our primary and secondary outcome variables in the analysis section to be released after the completion of the experiment.
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
Given the two-stage nature of this study, it is important for us to protect the integrity of the experiment. Therefore, we are withholding details from the public portion of this preregistration and reserving those details for the hidden portion.
Experimental Design Details
We will recruit participants on mTurk for our study, which consists of two stages, with the second stage occurring at a specified time and date one week after the first stage (known to them upfront). Participants will be encouraged to take stage 1 only if they can return for stage 2, and we will pay participants for each stage separately. We will strongly incentivize participants to complete both stages by providing a minimum payment of $6.50 for stage 2, while offering only $1.50 for completing stage 1 (each stage takes the same amount of time, roughly 15-20 minutes). Participants who agree to take part in stage 1 will then be asked to answer 60 multiple-choice trivia questions. Each trivia question will be accompanied with a question asking the participant to report their confidence in the answer they chose on a six-point scale from “No confidence at all” to “Certain.” Each pair of questions (one trivia question and one confidence reporting question) will be presented on its own screen, and participants will not be able to go back and change their answers. The order in which the trivia questions appear will be randomized across participants. We will also use five randomly-placed attention check questions to make sure that participants were paying attention. After participants complete the survey, they will be asked to confirm that they intend to participate in stage 2 at the specified time and date The second stage of the study will be designed to explore participants’ likelihood of engaging in herding behavior, and how that propensity varies under different financial incentive levels. Returning participants will see the same 60 trivia questions they saw in stage 1, and will be again asked to answer them. However, there will be two key differences in the process. First, when participants are presented with a trivia question in stage 2, they will also be reminded of their answer to the question in stage 1 and the most popular answer to the question from stage 1 across all participants. Second, in stage 2 participants will be financially incentivized to answer each question correctly, with questions assigned to one of four different levels of financial incentive: $0, $1, $2, or $3. Participants will be told at the beginning of the second stage that correct answers to the 60 questions will be associated with financial incentives of $0, $1, $2, or $3, and that one question from each of the four possible financial incentive levels will be randomly selected to be used for payment. We will randomly vary the order that questions appeared and the financial incentives associated with each question, across participants, while keeping the difficulty level of questions similar across incentive levels. Incentives and ordering will be randomized as follows: participants will be randomly assigned to one of four possible “versions” of the stage 2 survey, designed to allow us to evenly vary the size of the financial incentive associated with any given question. Before stage 2 is implemented, the 60 questions will be classified into one of fifteen difficulty levels (four questions per difficulty level) based on the percent of participants who answered the question correctly in stage 1. Financial incentives will then randomly be assigned to questions in the four versions such that each question will appear with a different incentive in each of the four versions, and each incentive will only appear once per difficulty level in each of the four versions. To prevent participants from looking up the answers when money is on the line, we will require participants to explicitly agree to not look up the answers prior to answering the questions with financial incentives. Additionally, participants will be told that if there are any signs of cheating (higher number of clicks or longer answer times relative to stage 1 for that participants), they will not receive any payment. This was an effective deterrent in our first iteration of this study. After participants complete the trivia question section of the survey, they will be asked to answer a handful of demographic questions, at which point they self-report their age, gender, ethnicity, education, and geographic region of residence.
Randomization Method
Individuals will be randomly assigned to one of four tracks using the Qualtrics software. In addition, the order of the questions will also be randomly assigned by the Qualtrics software.
Randomization Unit
Individual
Was the treatment clustered?
No
Experiment Characteristics
Sample size: planned number of clusters
NA
Sample size: planned number of observations
260 participants
Sample size (or number of clusters) by treatment arms
We will randomly assign individuals to one of the four treatment groups, so they should be roughly the same size.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
We conducted a power calculation for both the impact of financial incentives on herd behavior and the stage 1 confidence gap between men and women. For the effects of financial incentive, we tested for a minimum effect of 3 percentage points on the probability of engaging in herd behavior, clustering standard errors at the individual level. Note that this is in line with findings from our first iteration of this study, with a very similar design. We also did a power calculation for detecting a 0.15 unit difference in confidence between men and women where the standard deviation for stage 1 confidence is 1.56 for men and 1.43 for women. Since detecting the impact of financial incentives on herding requires more observations, we use this power calculation. The sample size required to detect a 3 percentage point effect is about 7,000 opportunities to engage in herd behavior. Since there are likely to be, on average, around 30 herding opportunities per stage 2 participant, we will need 230 participants. We will recruit 260 participants for stage 1 in anticipation of some attrition. Furthermore, we include a .do file containing our power calculation.
Supporting Documents and Materials

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

Request Information
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
Swarthore College IRB
IRB Approval Date
2017-08-02
IRB Approval Number
1415084
Analysis Plan

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

Request Information
Post-Trial
Post Trial Information
Study Withdrawal
Intervention
Is the intervention completed?
No
Is data collection complete?
Data Publication
Data Publication
Is public data available?
No
Program Files
Program Files
Reports, Papers & Other Materials
Relevant Paper(s)
REPORTS & OTHER MATERIALS