Choice architecture, gender, competition and evaluation.

Last registered on March 31, 2019

Pre-Trial

Trial Information

General Information

Title
Choice architecture, gender, competition and evaluation.
RCT ID
AEARCTR-0004030
Initial registration date
March 28, 2019

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
March 31, 2019, 11:26 PM EDT

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

Locations

Region
Region

Primary Investigator

Affiliation
University of Toronto

Other Primary Investigator(s)

PI Affiliation
University of Toronto
PI Affiliation
University of Toronto

Additional Trial Information

Status
In development
Start date
2019-03-25
End date
2019-04-07
Secondary IDs
Abstract
In most organizations, promotion into leadership typically requires self-nomination and entry into competition via an application. However, research on gender differences in self-promotion and preferences for competition suggests that this “opt-in” process might result in fewer women choosing to compete. For example, compared to men, women tend to be less (over)confident, less likely to self-promote and exaggerate accomplishments, and less likely to seek out risks and competition. These findings suggest that women might be less inclined to apply for promotions and competitive selection processes because of the typical requirement to self-nominate, promote, and compete. Drawing from findings in behavioral science, we hypothesized that changing promotion schemes from a default where applicants must opt in (i.e., self-nominate) to a default where applicants must opt out (i.e., those who pass a qualification threshold are automatically considered for promotion, but can choose not to be considered) will attenuate gender differences.

We performed a laboratory experiment using the paradigm of Niederle and Vesterlund (2007), [https://academic.oup.com/qje/article/122/3/1067/1879500] where participants solved two-digit number additions first under a noncompetitive piece rate and then under a competitive tournament incentive scheme; in a third stage, they can chose whether to enter a tournament again. We introduced an experimental variation to this paradigm: in the third stage, participants were either assigned to an opt-in system, where as in NV they could elect to compete, or to an opt out system, where the default choice was that everyone would compete, ad those who preferred not had to state their preference to be rewarded in a non-competitive (piece rate) scheme. In the opt-in framework we quite closely replicated NV’s main result, i.e. a significantly higher share of man selected the tournament than of women. However, the gender difference was eliminated when the choice to compete was described using opt-out framing. These preliminary results suggest that organizations could make use of “opt-out” promotion schemes as a behavioral intervention to reduce the gender gap in promotion rates and ascension to leadership positions.

For the current experiment, we set out to examine if an “opt-out” promotion scheme has downstream effects on any evaluator biases in selecting candidates. Our design is similar to the one in Bohnet et al. (2015) [https://pubsonline.informs.org/doi/10.1287/mnsc.2015.2186], were participants act as evaluators, and the profiles of the “candidates” are taken from our previous experiment. We are interested in testing whether the type of promotion framing, all else constant, affects the preference for male or female candidates.
External Link(s)

Registration Citation

Citation
He, Joyce, Sonia Kang and Nicola Lacetera. 2019. "Choice architecture, gender, competition and evaluation.." AEA RCT Registry. March 31. https://doi.org/10.1257/rct.4030-1.0
Former Citation
He, Joyce, Sonia Kang and Nicola Lacetera. 2019. "Choice architecture, gender, competition and evaluation.." AEA RCT Registry. March 31. https://www.socialscienceregistry.org/trials/4030/history/44452
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)
Participants selected via mTurk will be asked to evaluate hypothetical candidates. The candidates are represented by the participants to a previous experiment that we ran, where those participants solved two-digit number additions first under a noncompetitive piece rate and then a competitive tournament incentive scheme; and in a third stage, they could chose whether to enter a tournament again. Among all subjects in this early experiment, we will consider only those who, in the third stage, were rewarded with a tournament scheme. Each mTurk respondent to the current experiment will see the profile of four of those subjects; information will include gender, age, field and year of study (subjects in the original experiment were college students), as well as their performance in the second stage of the task, i.e. the number of correct responses that they gave. In addition, the respondents will be told whether the choice framing for the third stage compensation scheme was an opt-in or opt-out one.
Each participant will consider three separate pools of four applicants. For each participant, one pool with have one female and one male candidate with very similar scores in the second round of the original experiment (equal or with a difference of one correct answer); a second pool will be "male-dominated", i.e. there will be one male candidate with a much higher score than the second highest candidate, who will be a woman; the third pool will be "female dominated", with the highest scorer being a woman and her score being significantly higher than the man in second place.
Intervention (Hidden)
Intervention Start Date
2019-03-25
Intervention End Date
2019-04-07

Primary Outcomes

Primary Outcomes (end points)
Our main outcome of interest will be the choice to "promote" a male or female candidate, as a function of their recorded performance (or rank within the pool of reference), of whether the subjects in a pool were, in the original experiment, in the opt in or opt out choice framing condition, and of the type of pool (close tie, female dominated, or male dominated).

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
We will also consider secondary measures, such as the time that each respondent took to examine the characteristics of the applicants in each pool.
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
1. Participants will be recruited via mTurk. we will restrict to participants with at least 90% approval rates on the platform.

2. Participants will receive information on the previous experiment that we performed, where subjects solved addition tasks (details o the information that we gave are in the attached documents). Participants will also have to respond to two comprehension questions and those who have wrong answers will be screened out when analyzing the data.

3. Participants will be asked to consider those who participated in a tournament-based compensation scheme in the third round of the original experiment (called round 2 in the current experiment), as competing for a promotion within groups of four contestants.
They will then be presented with information on the four candidates of a given pool or group. Participants will see each applicant within a pool of four separately then together, and asked to make a choice among the four applicants. They will see the applicants and the choice for three different applicant pools.

4. Within each pool, a participant will observe the following information about each candidate (subject from the previous experiment): age, program of study, gender, year in college; number of correct responses given in the previous round in the original experiment; moreover, the information for each applicants will include whether that applicant chose to compete in a tournament where competing in the tournament was or was not the default choice for that round. Participants/evaluators will also be shown the distribution of correct responses in round 2 of the original experiment (called round 1 in the current experiment) for all those subjects who were in the condition (opt in or opt out) a given participant/evaluator is assigned to.

5. The candidates reported in each of the three pools will be randomly selected, with the following constraints:
- We exclude for the eligible sample the subjects in the original experiment whose number of correct responses in the second round (where all subjects competed in tournaments) was in the bottom 25th percentile.
- Of the three pools that each individual will consider, one pool with have one female and one male candidate with very similar scores in the second round o the original experiment (equal o with a difference of one correct answer), and significantly higher than the scores o the other two subjects in the pool; a second pool will be "male-dominated", i.e. there will be one male candidate with a much higher score than the second highest candidate, who will be a woman; the third pool will b female dominated, with the highest scorer being a woman.
- Each pool fo four applicants will have two men and two women.
The attached documents include examples.

6. For each of the three pools, a participant will be asked to select the applicants that they would hypothetically hire. The choice is incentivized. If the individual that a participant chooses is the one, among the four in the pool, who actually had the highest number of correct answers in the third stage of the original experiment, then the participant will receive a bonus o $0.75 in addition to the participation fee of $1.5. More precisely, for each participant we will randomly select one of the three pools; if, for that pool, the participant picked the applicants with the highest score in the third round of the original experiment, then the participant will receive the bonus.
Experimental Design Details
Randomization Method
Computer based randomization (oTree programmed)
Randomization Unit
Individual participant (evaluator) + pool of hypothetical applicants from previous experiment.
The order in which the three pools will be presented to each participant-evaluator will be randomized, too.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
1000 participants
Sample size: planned number of observations
1000 participants x 3 observations each
Sample size (or number of clusters) by treatment arms
Expected 500 participants will receive pools of subjects from the previous study who were in the opt-in condition.
Expected 500 participants will receive pools of subjects from the previous study who were in the opt-out condition.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
500 observation per group will allow to estimate significant differences of 10 percentage points in the likelihood of selecting a woman (all else equal) with alpha o 0.05 and power of 80%, when the proportions are close to .5.
Supporting Documents and Materials

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

Request Information
IRB

Institutional Review Boards (IRBs)

IRB Name
University of Toronto Research Ethics Board
IRB Approval Date
2018-09-06
IRB Approval Number
36556

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