Group size and diversity

How to cite

Pereira, Brais Álvarez; Aman-Rana, Shan; Delfino, Alexia. 2022. "Group size and diversity." AEA RCT Registry. AEARCTR-0007988. https://doi.org/10.1257/rct.7988-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Brais Álvarez Pereira* , Shan Aman-Rana , Alexia Delfino

* Primary investigator

Registry ID
AEARCTR-0007988
DOI
https://doi.org/10.1257/rct.7988-1.0
First registered
Last updated
Status
Completed
Trial dates
2018-08-01 – 2018-08-30
Countries
Guinea-Bissau ()
JEL codes
C92, D00, D23, D73, D89, D91, O12
Registry record
http://www.socialscienceregistry.org/trials/7988

Abstract

In this paper we analyse the relationship between group size and the effect of diversity on workers’ performance. We focus on diversity in skills, but also investigate the additional impact of diversity in gender on performance. First, we explore the link between diversity and group size theoretically in a model with knowledge spillovers in production, which predicts that the effect of diversity (in skills) on individual performance increases with group size. We test this hypothesis in a lab experiment in Guinea-Bissau. Nursing and Economics students solve an individual test twice, first in isolation and then communicating with other participants, in homogeneous and diverse groups of different sizes. Our main outcome variable is the change in the percentage of correct answers between the two tests taken by a given individual. We test whether this change is positive and largest when the respondent works in larger and diverse groups compared to smaller homogeneous - or diverse - groups. We additionally ask whether the effect of skills diversity on performance is affected when accompanied by gender diversity, and whether this interaction is different in small or large groups and for male or female respondents.

Experimental design

The design simultaneously studies three factors that could affect performance: team size, skill diversity, and gender diversity. With this aim, for the second round of the test, participants were randomly distributed into teams of size 2 or 4, composed only of nurses or of nurses with one economist of either gender.
Thus, in total, we have the following 6 types of groups:
- size 2, skill and gender homogeneous
- size 4, skill and gender homogeneous
- size 2, skill heterogeneous and gender homogeneous
- size 4, skill heterogeneous and gender homogeneous
- size 2, skill heterogeneous and gender heterogeneous
- size 4, skill heterogeneous and gender heterogeneous
We have both male and female nurses in the experiment, and will look at effects overall and splitting the sample by gender.

Design details

Randomization method
Administrative lists were used to randomize invites to participants (randomisation done in the back office using computer). Those who came to the university on the day of the experiment were randomized in a public lottery, stratified by gender.
Randomization unit
Individuals randomised into different types of groups.
Clusters
86 clusters (experimental groups)
Observations
258 participants
Arms
86 teams / 6 treatments = 14.3 groups by treatment
Primary outcomes
Each participant repeats a test made of 25 multiple-choice questions twice, first in isolation and then in a group (with experimentally-varied composition). The main outcome variable is the difference in the % of correct answers between the second (in group) and first (in isolation) test rounds. The % of correct answers is computed as the number of correct answers over 25 (multiplying by 100). Each question had five options and a unique correct answer: the first 10 were general-knowledge questions; the next 5 were questions from the other discipline (economics questions for nurses and nursing questions for economists); the last ten were questions from the participant’s field of study (nursing for nurses and economics for economists). We analyse the results overall and by splitting the sample by gender of the test-takers (our stratification variable). Our participants are both nurses and economists. However, economists are mainly used as sources of variation in team diversity (see treatments below), thus the analysis focuses on the outcomes for nurses.
Secondary outcomes
- Split of main outcome variables by knowledge area of the questions

Cited by

No citation data available for this registration yet.

Keywords