LGBTQ+ individuals, social preferences, and cooperation

2020-09-22

2022-12-31

Our primary outcome variables are:
(i) Recipients’ ID-flag choices
(ii) Amount sent by dictators in the dictator game

Our secondary outcome variables are:
(i) Recipients’ ID-string choices
(ii) Dictators’ beliefs about demographics of matched recipients
(iii) Recipients’ earnings

Participants will be assigned either the role of the Dictator or the Recipient. The experiment will be conducted asynchronously via Prolific, where recipients will be asked to make their decisions first (Part 1), followed by dictators at a later stage (Part 2).

The Recipient’s LGBTQ+ identity will be revealed to the Dictator in a non-intrusive manner.

Participants will be assigned either the role of the Dictator or the Recipient. The experiment will be conducted asynchronously via Prolific, where recipients will be asked to make their decisions first (Part 1), followed by dictators at a later stage (Part 2).

The Recipient’s LGBTQ+ identity will be revealed to the Dictator in a non-intrusive manner.

Part 1: Recipients.
In treatment ID-First, recipients will be asked to construct an ID that represents them and reveals their LGBTQ+ identity (details below). Then, they will be given the instructions of the dictator game. In treatment ID-Second, recipients will first be given the instructions of the dictator game before being asked to construct their ID. Treatment assignment will be between subjects for the recipients.

At the end of the experiment, recipients will be asked to complete a survey eliciting demographic variables and asking them questions about the decisions they have made during the experiment.

Part 2: Dictators.
Dictators will first be provided with details about the procedure under which Recipients have constructed their ID.

They will then participate in two rounds of a dictator game, where in each round they will be matched with one randomly chosen recipient from Part 1 and presented with the recipient’s ID. Dictators will be informed that the ID that they are shown was chosen by their matched recipient. They will then choose how much of an endowment (100 ECU) to allocate between themselves and the recipient.

Treatment assignment is within subject for dictators. In the first round of the dictator game, dictators will be matched to a recipient who has either chosen a Pride flag or a non-Pride flag to be part of their ID. In the second round, dictators will be matched to a recipient who chose a flag in the opposite category. That is, in the Pride-First treatment, dictators are first matched with a recipient who chose the Pride flag in Round 1, and then with a recipient who chose the non-Pride flag in Round 2. This order is reversed in the Pride-Second treatment.

Dictators are not informed that there are two rounds of the same dictator game. Rather, they are informed that there are several parts to the experiment, and the instruction for the second round of the dictator game is revealed after they have completed the first round.

At the end of the experiment, dictators will be asked to complete a survey eliciting demographic variables and asking them questions about the decisions they have made during the experiment. Dictators will also be asked to complete an Implicit Association Test.

ID Creation:
The ID creation is only for recipients and is common for all treatments (ID-first and ID-second). There are two parts to the ID: (i) an 8-character alphanumeric component, and (ii) an icon.

First, participants will be presented with the same three randomly generated alphanumeric strings (pre-determined by the researchers), and they will be asked to choose one to form part of their ID.

Next, participants will be presented with three possible icon options: the Pride flag and two non-Pride flags (each consisting of six horizontal stripes of randomly chosen colors pre-determined by the researchers). All participants will be presented with the same three options, and they will be asked to choose one to form part of their ID.

Hence, LGBTQ+ identity (or allyship) will be signaled endogenously through the choice of the pride flag by the participants.

Payment procedure:
The experiment will be conducted asynchronously and there will potentially be multiple dictators matched to the same recipient. For all dictators, one of their two decisions (across two rounds of the dictator game) will be implemented, and that decision will determine the payment of both the dictator and their matched recipient. Hence, recipients can potentially receive transfers from multiple dictators.

Recipients are informed of this procedure in the experiment. Dictators are informed that one of their decisions will be randomly chosen for payment and the earnings of both themselves and their matched recipient will be determined by this decision.

We follow a two-by-two treatment design.

On the recipient’s end, they will be randomly assigned to either the ID-First or ID-Second treatments. This will be randomly determined at the individual level within the same session such that we have a balanced sample of recipients in each treatment.

On the dictator’s end, they will be randomly assigned to either the Pride-First or Pride-Second treatments. This will be randomly determined at the individual level via a virtual coin toss that is programmed in the experimental software (oTree).

The unit of randomization is at the individual level for dictators and at the session level for recipients.

However, given that this is an online experiment conducted via Prolific, there are no “physical” sessions per se since each session will run continuously until the required number of participants is fulfilled. Both dictators and recipients make their decisions independently at their own pace, and they can start and complete the experiment at any time while the session is live. Moreover, participants do not receive any feedback until the conclusion of all experiments (i.e., both recipients’ and dictators’ sessions). Hence, for our econometric analysis of participants’ behavior, we will assume that the unit of randomization is at the individual level.

No

840 (560 dictators and 280 recipients)

840(560 dictators and 280 recipients)

Recipients:
Sample size per group = 70
Number of groups = 4 (Heterosexual vs. Non-Heterosexual, ID-First vs. ID-Second)
Note: We will recruit a gender-balanced sample (i.e., equal number of males and females) of heterosexual and non-heterosexual individuals.

Dictators:
Sample size per group = 140
Number of groups = 4 (Heterosexual vs. Non-Heterosexual (3 groups))
Note: We will recruit our sample of heterosexual individuals such that they are balanced across different political affiliations (Democrat, Republican and Independent/Other).

Recipients: - Overall, power calculation is based on: (i) a drop in Pride flag choices from 75% to 55%; (ii) one-tailed test of difference between two independent proportions; (iii) Type I error rate of 0.05 and power of 0.80. - Minimum required sample size per group = 70. Our primary conjecture is that non-heterosexual individuals will be less likely to choose the Pride flag in treatment ID-second than in treatment ID-first due to anticipated discrimination. Given this, we consider a one-tailed test for our power calculation. Our initial survey indicates that about 75% of non-heterosexual individuals choose the Pride flag in the absence of strategic concerns. Dictators: - Overall, power calculation is based on: (i) an effect size of 0.25 standard deviations change in giving amount between two recipients; (ii) two-tailed test of difference between two dependent means (matched pairs); (iii) Type I error rate of 0.05 and power of 0.80. - Minimum required sample size per group = 128. Actual sample recruited is higher to allow for heterogeneous treatment effects. Our primary conjecture is that average giving will be lower to individuals who chose the Pride flag as opposed to those who chose the non-Pride flag. However, potential in-group bias by non-heterosexual dictators (e.g., gay dictators) towards those who choose the Pride flag may mean higher given to this group. Given this, we consider a two-tailed test for our power calculation.

Pre-Analysis Plan (September 17, 2020)