Creating the Minimal Group Paradigm

1. Participant Recruitment

1. Conduct a power analysis and recruit a sufficient number of participants.

2. Organize materials.

1. Create forty images of dot clusters with various amounts (see example of one in Figure 1).

Figure 1: Example of dot estimation task. Twenty-three dots are displayed in the image above. Participants are only given 125-500 ms to view each image before estimating the number of dots that were on the screen.

2. Generate ordered matrices, each of which should consist of 14 boxes containing two numbers.
   1. Label each row, "These are rewards and penalties for member number [insert code number] of your group", or "of the other group".
   2. Create six matrices that satisfy one of three criteria (two each; Figure 2).
      1. Type A: The maximum penalties exceed maximum rewards. Order the boxes such that the two maximum joint payoff and maximum fairness terms are in the middle.
      2. Type B: No penalties and constant joint payoff. Order the boxes such that the two maximum fairness terms are in the middle.
      3. Type C: The maximum rewards exceed maximum penalties. Order the boxes such that the maximum joint payoffs are at both extremes and the two maximum fairness terms are in the middle.
   3. Replicate these six matrices for three different types of choices corresponding to the following labels for each row:
      1. In-group choices: The top row refers to rewards and penalty outcomes for one in-group member and the bottom row refers to outcomes for another in-group member.
      2. Out-group choices: The top row refers to rewards and penalty outcomes for one out-group member and the bottom row refers to outcomes for another out-group member.
      3. Differential (intergroup) choices: One row refers to rewards and penalty outcomes for an in-group member and the other row refers to outcomes for an out-group member. Randomize the order of these rows such that three matrices reflect in-group choices on the top and the other three reflect out-group choices on the top.

Figure 2: Typical payout matrices used in this experiment. Each matrix consists of two rows, reflecting monetary tradeoffs that affect other in-group or out-group members. Matrix types (A, B, and C) are indicated on the left, with two matrices belonging to each. All six matrices are replicated three times, one for each decision type (in-group, out-group, and differential). A differential choice in matrix 1, for instance, might read, "These are rewards and penalties for Member 3 of your group" (top row) and "Member 2 of the other group" (bottom row) along with 14 tradeoff terms. Thus, term 14 (i.e., the most unfair option) would deduct 19 points from an outgroup member and give 6 points to a fellow in-group member. Terms 7 and 8, on the other hand, reflect the maximally fair options since they minimize the joint penalty (-1) for both players.

3. Randomize the order of the 18 matrices.
4. Title the front of each section, "For member of the [insert group identification]" where group identification reflects one of the four experimental conditions: (1) under-estimator groups, (2) over-estimator group, (3) better accuracy group, and (4) worse accuracy group.

3. Data Collection

1. Seat each participant at a computer.
2. Present each of the 40 dot-cluster images very briefly, anywhere from 125-500 ms.
3. Allow time between images so that participants can estimate the number of dots they see and record their responses on the computer.
4. To assess the effects of the categorization on intergroup behavior, inform participants that you are interested in studying other types of judgments and, for convenience, you will be dividing them into two groups for making these judgments.
   1. Randomly assign half the participants to either the neutral condition or the value condition.
5. Inform participants in the neutral condition that one group will consist of participants who provided the highest estimates for the number of dots and the other group will consist of those who provided the lowest estimates.
   1. Randomly assign half of these participants to "under-estimators" and the other half to "over-estimators".
6. Inform participants in the value condition that one group will consist of participants who provided more accurate estimates for the number of dots and the other group will consist of those who provided less accurate estimates.
   1. Randomly assign half of these participants to the "better accuracy" group and the other half to the "worse accuracy" group.
7. Lead participants to separate cubicles and inform them that they will soon make real monetary decisions where they can (anonymously) reward and punish other participants.
8. Have the participants do the matrices with their corresponding group identification. Instruct them to indicate their choices by selecting one box per matrix (Figure 3).

Figure 3: Example page of matrix booklet from participant's perspective. Six of the 18 matrices are displayed on each page of the matrix booklet. In the example above, there are two matrices of each type (A, B, and C) and two matrices for each choice (in-group, out-group, and differential). For instance, matrix 3 is type C with a differential choice since (1) the maximum rewards exceed maximum penalties and (2) participants must decide between outcomes than impact a fellow in-group member (top row) or an out-group member (bottom row).

9. Fully debrief participants.

4. Data Analysis

1. Score the matrices from 1 to 14, where 14 stands for the choice in the matrix which gives the member of the in-group the maximum possible points on that matrix and 1 gives the in-group the minimum possible points; a score of 7.5 represents maximal fairness.
2. Perform one-sample t-tests to determine whether the individual means scores within groups were significantly different from the point of fairness.