Socially-Expected Leadership

Last registered on April 29, 2022

Pre-Trial

Trial Information

General Information

Title
Socially-Expected Leadership
RCT ID
AEARCTR-0009191
Initial registration date
April 28, 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
April 29, 2022, 10:28 AM EDT

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

Locations

Region

Primary Investigator

Affiliation
CIDE

Other Primary Investigator(s)

PI Affiliation
Harvard University

Additional Trial Information

Status
In development
Start date
2022-04-29
End date
2022-06-17
Secondary IDs
Prior work
This trial is based on or builds upon one or more prior RCTs.
Abstract
We provide experimental evidence to assess whether pro-social leading is motivated by altruism or by a less lofty desire to conform with expectations. The evidence from our initial experiment indicates that many individuals' pro-social leading may be driven by conformity to what they imagine is expected of them by others. This is somewhat counterintuitive as the notion of leadership seems to imply influencing others by surpassing expectations, not simply conforming with them. But what we observe looks less like inspired boldness and more like stock role-playing.

Registration Citation

Citation
Fernandez Duque, Mauricio and Michael Hiscox. 2022. "Socially-Expected Leadership." AEA RCT Registry. April 29. https://doi.org/10.1257/rct.9191-1.0
Sponsors & Partners

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

Request Information
Experimental Details

Interventions

Intervention(s)
Subjects face sequential choices about how of their 100 peso endowment to divide between themselves and a charity. We assign subjects to be either a first-mover who makes a choice that is potentially observable by a second-mover. For clarity of exposition, we will refer to the first-movers in the female gender, and to the second-movers in the male gender.

The basic design is the following. First-movers make a contribution decision which will be seen by a second-mover before he makes his own contribution decision (we will call this the "influence scenario"). First-movers also make a contribution decision which will be seen by a second-mover after he makes his own contribution decision (we will call this the "no influence scenario"). Only one of the first-mover's decisions, along with the timing of what the second-mover sees, will be implemented. The first-mover's third decision affects the probability of implementation (we call this the "replication decision"). We also elicit the first-mover's beliefs over what second-movers and first-movers do, and over what others believe second-movers and first-movers do. Finally, we ask first-movers sociodemographic questions, questions about attitudes regarding leadership, and debriefing questions.

The intervention with second-movers is much more straightforward. Second-movers decide how to divide their endowment with the charity, either before or after seeing what the first-mover contributed. Which first-mover the second-mover is matched with, and whether second-movers see their first-mover's contribution before or after their own contribution, are determined randomly.
Intervention Start Date
2022-04-29
Intervention End Date
2022-06-17

Primary Outcomes

Primary Outcomes (end points)
We will test seven hypotheses.

1) A non-negligible percentage of first-movers give more in the influence scenario. We will estimate the percentage of first-movers who give more in the influence scenarios.

2) The proportion of first-movers who give more in the influence scenario is significantly larger than the proportion who give more in the no influence scenario. We will compare the proportion in both groups with a paired t-test.

3) First-movers who contribute more in the leadership scenario make the no influence scenario significantly more likely to be implemented than the influence scenario. We will estimate, among first-movers who contribute more in the leadership scenario, the proportion who make the no influence scenario more likely.

4) Consider first-movers who contribute the same amount in both contingencies. The higher their contribution, the more likely they are to make the influence scenario more likely. We will test for this by looking at the proportion of first-movers who make the influence contingency more likely, dividing the data by terciles. We will further test for this via an OLS regression, restricted to subjects who contributed the same amount in both scenarios, where the dependent variable is whether the subject made the influence contingency more likely to be implemented, and the independent variable is the first-mover's contribution.

5) Consider the following condition on the first-mover's beliefs: the first-mover believes that the second-mover will contribute more in the no influence scenario than he would in the influence scenario had the first-mover contributed the same amount in that scenario as in the no influence scenario. First-movers whose beliefs fulfill this condition are more likely to contribute more in the influence scenario and then make the no influence scenario more likely to be implemented. To test for this, we will run an OLS regression in which the independent variable is an indicator variable equal to one if the first-mover's beliefs fulfill the above condition, and the dependent variable is an indicator variable equal to one if the first-mover contributes more in the influence scenario and then makes the no influence scenario more likely to be implemented.

6) The difference in contributions between the influence and no influence scenarios increases with the social expectations in the influence scenario and decreases with social expectations in the no influence scenario. The first effect is stronger than the second. Social expectations are measured via first-movers' guesses over what other first-movers guessed first movers would do in each scenario. In order to test this hypothesis, we follow a difference in difference specification. The unit of analysis is a subject-scenario. That is, there are two observations per subject. We then are interested in a difference in difference regression where the variables that interact is the scenario the first-mover is in and the social expectation of a specific scenario. To be explicit, the main regressors are a dummy for the influence scenario, social expectations in the influence scenario, social expectations in the no influence scenario, the interaction between the dummy for the influence scenario and social expectations in the influence scenario, and the interaction of a dummy for the influence scenario and social expectations in the no influence scenario. The coefficients of interest are the interaction terms.

7a) First-movers who contribute the same amount in both scenarios will make the influence scenario more likely to be implemented under the following condition on beliefs: if the contribution they expect from second-movers in the no influence scenario is lower than the contribution they expect in the influence scenario. To test for this, we run an OLS regresson, restricted to first-movers who contributed the same in both scenarios, in which the dependent variable is an indicator variable equal to one if the first-mover makes the influence scenario more likely to be implemented, and the independent variable is an indicator variable equal to one if the condition on beliefs stated in this hypothesis hold.

7b) First movers who contribute the same amount in both scenarios are more likely to make the influence scenario more likely to be implemented if the following condition on beliefs holds: their contribution in the influence scenario is at least as high as their social expectations in that scenario. To test for this, we run an OLS regresson, restricted to first-movers who contributed the same in both scenarios, in which the dependent variable is an indicator variable equal to one if the first-mover makes the influence scenario more likely to be implemented, and the independent variable is an indicator variable equal to one if the condition on beliefs stated in this hypothesis hold. The regression includes as controls social expectations of the influence and no influence scenarios.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
1) The average contribution in the influence scenario is significantly higher than the average contribution in the no influence scenario. We will estimate the difference in average contributions across both scenarios, tested via a t-test and a simple OLS regression.

2) This hypothesis is a secondary hypothesis to main hypothesis 5. Consider the following condition on the first-mover's beliefs: the first-mover believes that the second-mover will contribute more in the no influence scenario than he would in the influence scenario had the first-mover contributed the same amount in that scenario as in the no influence scenario. The fifth hypothesis is that first-movers whose beliefs fulfill this condition are more likely to contribute more in the influence scenario and then make the no influence scenario more likely to be implemented. The secondary hypothesis is that first-movers whose beliefs fulfill this condition are more likely to contribute more in the influence scenario. To test for this, we will run an OLS regression in which the independent variable is an indicator variable equal to one if the first-mover's beliefs fulfill the above condition, and the dependent variable is an indicator variable equal to one if the first-mover contributes more in the influence scenario.

3) There is a further secondary outcome related to hypothesis 5. This is a secondary outcome because I expect to be underpowered to test it. The hypothesis is that, among first movers who contribute more in the influence scenario, those for whom the belief conditions of primary hypothesis 5 holds are more likely to make the no influence condition more likely to be implemented. This would be tested with an OLS regression, restricted to subjects who contribute more in the influence contingency, where the dependent variable is an indicator variable equal to one if they make the no influence scenario more likely, and the independent variable is an indicator variable equal to one if the belief conditions hold.

3) Consider the primary hypotheses which have as independent variables the social expectations over first-movers' actions. These hypotheses would have have insignificant results if we replace social expectations over first-movers' actions with expectations over first-mover's actions.
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Contribution decisions.

Each first-mover makes two decisions about how to divide an endowment of 100 pesos between herself and a charity. Subjects learn about the two decisions simultaneously. To eliminate the potential for differences at margins, only one of these decisions (chosen at random) will be implemented for each subject. The first two decisions are made in different scenarios. The two scenarios are identical except in one key aspect. In both scenarios, the first-mover must choose how to split the money between herself and the charity. A second-mover will be told of one of the first-mover's decisions, and will make his own decision about how to divide his endowment. The only difference between the first-mover's first and second scenario is the timing of when the second-mover learns about what the first-mover chose to do. In one scenario, the second-mover will be informed about what the first-mover has contributed before he makes his own decision (we will call this the influence scenario). In the alternative scenario, the second-mover will only get the information after he has made his own decision (this we will call the no influence scenario). In both cases, the second-mover is able to pass judgment on the choice made by the first-mover. However, only in the influence scenario can the first-mover influence what the second-mover chooses to contribute.

The difference between what the first-mover contributes in the influence versus the no influence scenarios is our basic measure of whether there is leadership in the sense of giving extra when in a position of influence.

The design took several considerations into account to isolate the motivation for leading. To minimize social image concerns and demand effects, the first-mover's identity is not revealed to the second-mover, the experimenter or the charity. To further avoid demand effects, we tried to use value-neutral words in the instructions. For example, instead of `influence' and `no influence' scenarios, we referred to scenarios where person 2 contributes before or after seeing what person 1 contributes. We presented both scenarios simultaneously. Pilot sessions indicated that a simultaneous display of the scenarios was important for comprehension. By presenting both scenarios simultaneously, we addressed the further concern that if they made one decision before observing the second scenario, the first decision would anchor the second one.

In a third decision made by the first-mover, she must choose which of the first two decisions she would like to make more likely to be implemented. That is, suppose she had originally decided to contribute X in the influence scenario, and Y in the no influence scenario. Either the influence scenario with a contribution of X by the first-mover is implemented, or the no influence scenario with a contribution of Y by the first-mover is implemented. In her third decision she decides the probability with which the first option is implemented: either 1/3 or 2/3. We will refer to this third decision as the replication decision.

The replication decision she makes reveals which contribution-contingency pair she prefers. It allows us to measure whether subjects "take back leading" in the sense of contributing more in the influence contingency and then makes the no influence contingency more likely to be implemented. It also allows us to measure whether subjects who contribute the same in both scenarios try to put themselves in a position where they can influence others, a second sense of the term leadership in which subjects seek primacy.

Elicitation of beliefs.

After these choices have been made, we ask the first-movers a series of questions to elicit beliefs about others and measure social expectations, presented as `guessing games' (with 50 peso prizes for those who came closest to the correct answers). We asked each first-mover to guess how much the average second-mover contributed to the charity in the influence scenario (a guess for each whole dollar amount between 0 and 10 dollars that the first-mover would have contributed first) and in the no leadership scenario (one amount to guess). After this they were asked to guess how much the average first-mover contributed in the influence and no influence scenario. To generate measures of social expectations, we asked each first-mover to guess the average of what others had guessed that first-movers do. This is our measure of social expectations.


Socio-demographics, personality, exit survey.

After the guessing games, first-movers are asked to answer socio-demographic questions, personality questions and an exit survey.

First-movers are not informed about any details about second-movers other than what we have explained so far.
Experimental Design Details
Randomization Method
The randomization will be done automatically by Qualtrics.
Randomization Unit
The unit of analysis is the individual.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
300 first-movers and 300 second-movers.
Sample size: planned number of observations
300 first-movers and 300 second-movers.
Sample size (or number of clusters) by treatment arms
There is only one treatment for first-movers.

There are two treatments for second-movers: those who make their contribution decision before and after seeing what their first-mover contributed. The number of subjects per treatment depends on the first-movers' replication decisions. In principle, there can be any proportion between 1/3 and 2/3 of secont-movers who see what the first-mover gave before they make their own contribution decision. In practice, I expect this proportion to be closer to 1/2.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
The main test for this experiment is whether we find ''reluctant leaders'': first movers who contribute more when they can influence their second mover, who then make the scenario where they cannot influence the second mover more likely, and who fulfill condition on their beliefs about how first-movers will react that is described in the earlier draft of this project. The first time we ran this experiment, we found that 10% of first-movers fulfilled this condition. We are aiming to find 30 subjects to fulfill this condition, which is how we determined the sample size.
IRB

Institutional Review Boards (IRBs)

IRB Name
Institutional Review Board, ITAM
IRB Approval Date
2022-04-19
IRB Approval Number
N/A

Post-Trial

Post Trial Information

Study Withdrawal

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

Request Information

Intervention

Is the intervention completed?
No
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