Feedback and Performance: A Lab Experiment

2021-05-26

2021-09-30

- Performance in the second draft of the poster (total score and two sub-scores: content and design )
- Contribution to public good game

Our analysis of the performance in the second draft of the poster focuses on B-players only (see experimental design).
For the public good game contribution, we will look at both A-players and B-players (separately and pooled)

Aggregation of performance in the second draft of the poster: sum of points in content categories (10) and design categories (8), averaged over RAs who graded the posters independently

To increase precision, in our regressions we will control for the score of poster 1 (based on the the RAs grading), as well as for session fixed effects.

- Number of comments received in the feedback to the first draft of the poster
- Quality of comments (exploratory measures: survey items with subjective assessment of comments quality; correlation of number of comments received with objective score in first draft of the poster)

Our analysis of the number of comments received focuses on B-players only (see experimental design).

In an online lab experiment, participants are asked to design two drafts of a recruitment poster for the lab. Before they complete the second draft, they receive a review of their first draft from one of the other participants. Participants are allocated to different "feedback groups": in one group, pairs of participants review each other (sequentially); in another group, participants are matched anonymously. Subjects' payments are based on the peer feedback for the first draft (within a tournament) and on researchers' grading for the second draft. We study how feedback groups affect performance in the second poster and contribution in a public good game subjects play at the end of the session, as well as feedback content.

EXPERIMENT STRUCTURE
We conduct an online lab experiment over several sessions. Each session comprises multiple steps. In each session players are randomly allocated to two types (A-players & B-players).

In each session:
1. All participants complete a first draft draft of a lab recruitment poster
2. Each A-player provides feedback on one of B-player's first draft
3. B-player reads feedback and writes second draft of the poster
4. B-player provides feedback on A-player’s first draft
5. A-player reads feedback and writes second draft
6. Public good game (perfect stranger match)
7. Exit survey

Feedback in step 2 and 4 consists of a numbered list of comments. Before providing feedback, the reviewer receives information on desirable content of the poster.

TREATMENTS
Half of the A-B player pairs are randomly assigned to the "Anonymity" treatment: A reviews B; later A is reviewed by 𝐵′≠𝐵.
The other half of A-B player pairs are randomly assigned to the "Reciprocity" treatment: A reviews B; later A is reviewed by same B.

PAYMENTS
In each session, the top 25% of the first drafts receive a reward, where the top 25% are those with the lowest number of comments in the peer feedback. In addition, participants receive a payment based on the score in the second draft, as graded blindly by research assistants.These incentives are paid in addition to a flat fee for study participation.

INFORMATION
The treatment and payment structure are common knowledge

Randomization done within each session by a computer

Individual-level randomization

No

Approximately, 10 sessions with 32 participants each (16 A-players and 16 B-players). A-B pairs are randomly allocated to the two feedback treatment groups.

Our analysis focuses on B-players (i.e. those who receive feedback in step 2) for the second draft performance and number of comments received. We expect 16 observations per session, or a total of 160 observations. We also analyze contributions to the PGG for all participants (an expected total of 320 observations).

N/A