'Social Learning, Behavioral Biases and Group Outcomes - Groups and Convergence'
(AsPredicted #134073)
Download PDF

---

---

1) Have any data been collected for this study already?
No, no data have been collected for this study yet.

2) What's the main question being asked or hypothesis being tested in this study?
This follows the experiment 'Social Learning, Behavioral Biases and Group Outcomes' (AsPredicted #130499), in which I investigate the impact of social learning on group outcomes, in the absence of any information about other participants.

In the Baseline study, participants observed one answer from another participant. The key finding is that social learning may both amplify or reduce behavioral biases, with the relative performance-relative confidence correlation being a strong predictor for the effect of learning. In this experimental branch, participants are matched with three other participants live, and then undergo multiple rounds of learning. The aim of this additional branch is to test two questions:
i) Are findings robust when social learning takes place in groups? In other words, when participants observe more than one answer at a time, does social learning have a negative (positive) effect on group outcomes when the relative confidence-relative performance correlation is negative (positive)?
ii) Are findings robust to iteration? That is, when participants interact in more than one round, is the effect of learning consistent with the case of one-shot interaction? Also, is there evidence of convergence?

3) Describe the key dependent variable(s) specifying how they will be measured.
(i) Distribution of optimality rate by group, in each round.
(ii) Overall optimality rate in the last round: share of participants whose answer is optimal in the last round.

4) How many and which conditions will participants be assigned to?
Two conditions, each associated with a specific task, RM and CRT. Each task is associated with either a negative (RM) or a positive (CRT) relative confidence-relative performance correlation.

5) Specify exactly which analyses you will conduct to examine the main question/hypothesis.
To answer the first question, the outcomes after the first round of learning will be relevant. For the second question, all other rounds are relevant, but in particular, the outcomes of the last round are relevant for evidence on convergence. The measures are:

(i) Using skewness, compare the distribution of optimality rates by group with a symmetric distribution with mean 0.5.
(ii) Test if the number of participants who answered optimally is significantly higher

Additionally, test that the overall optimality rate in the last round is significantly different from 0.5.

6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations.
Participants who fail a comprehension question are excluded from the study and therefore do not count towards the number of completes.

7) How many observations will be collected or what will determine sample size?
No need to justify decision, but be precise about exactly how the number will be determined.
1200 participants in total, 600 per condition (150 groups per condition).

8) Anything else you would like to pre-register?
(e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?)
Nothing else to pre-register.