Diversity and Team Performance

Researcher will use publicly posted data for a few upcoming business competitions across countries spanning 4 continents. Participants are randomly assigned to teams competing on a range of tasks (marketing, accounting, logistics/operations, business intelligence, organizational behavior). Teams can be different by gender composition, as well as information/knowledge that they can access before a task.

2024-09-10

2025-06-30

The business competition involves teams of two solving a series of business tasks: accounting, business intelligence, route optimization/logistics, sales, and coordination.

The outcome variable of interest is the performance score for each task. Note that for the business intelligence task, there are a few different scoring systems, scores from each scoring systems will be evaluated as an outcome variable.

The business competition involves teams of two solving a series of business tasks: accounting, business intelligence, route optimization/logistics, sales, and coordination. The accounting task requires real effort to conduct a series of calculations, the performance is measured by number of accounting problems correctly solved. The business intelligence quizzes participants on knowledge about a range of industries (7 industries covered for each quiz), including market sizing and naming major players and trends, and performance is measured by number of correct answers. The route optimization task requires participants to find a cost minimizing path from a starting point to an ending point on a network (represented by a graph where edges are labelled with different costs), performance is measured by cost of the path. The sales task requires the participants to come up with a sale pitch (up to 80 characters) for either a hair product, a dating
App, or a TV series, and the performance is measured by the average rating by a panel of 100 consumers recruited from Prolific. The coordination task is the Schelling game (10 different prompts, where 4 are gendered, 4 are geographical, and 2 are on random
topics), and performance is measured by number of matched responses.

A list of actual questions for one round of the accounting task is attached to this pre-registration.

A list of actual questions for one round of the route optimization task is attached to this pre-registration. In each round, one route optimization would involve the contestant being randomly assigned to only see the left side or the right side while the second route optimization allows the contestant to choose left or right (the team members can choose different sides so that they can collectively see the full map).

The actual prompt for one round of the sales pitch task is attached to this pre-registration. (which product is to be sold will be randomized)

A list of actual questions for one round of the Schelling game is attached to this pre-registration.

The business intelligence quizzes participants on knowledge about a range of industries (7 industries covered for each quiz). A sample of actual questions for one round/quiz is attached to this pre-registration.

A list of actual questions for one round of the accounting task is attached to this pre-registration.

A list of actual questions for one round of the route optimization task is attached to this pre-registration. In each round, one route optimization would involve the contestant being randomly assigned to only see the left side or the right side while the second route optimization allows the contestant to choose left or right (the team members can choose different sides so that they can collectively see the full map).

The actual prompt for one round of the sales pitch task is attached to this pre-registration. (which product is to be sold will be randomized)

A list of actual questions for one round of the Schelling game is attached to this pre-registration.

The business intelligence quizzes participants on knowledge about a range of industries (7 industries covered for each quiz). A sample of actual questions for one round/quiz is attached to this pre-registration. Two of the industries include readings (half of the participants will be in teams where each team member will be randomized to read about one of these industries, while the other half will be in teams where both team members will read about the same industry). These industry readings are generated by copy-and-pasting the "At a Glance" and "Performance" sections (for "Performance, it includes up to "Performance Snapshot") of IBISWorld Industry reports for randomly selected industries (industries as defined by IBISWorld) into ChatGPT4 with the prompt to summarize the material into about 400 words. The quiz questions for these industries are generated, again, with ChatGPT4 with the prompt to generate multiple choice questions based on the excerpt on the industry. The participants will compete on the quiz on the raw total of correct answers in one round of individual contest and in one round of team contest. They participate in a team round where they are scored across a set of different scoring systems for 2 pairs of industries (one pair of 2 industries with reading materials, and one pair of gendered industries): (a) best answer among the 2 within the team; (b) both has to be correct to score; or (c) each correct answer earns point [sum]. The other dimension of the scoring system then take the per question score as graded by (a), (b), and (c) and then either (d) take the score of the lowest scoring question across the questions in the industry group; (e) take the score of the highest scoring question across; or (f) take the average of the scores of all questions. Each of the 3X3 scoring systems generate the same expected score assuming the optimal choice where a player who read about an industry scores all questions correctly while a player who did not read about an industry has 1/4 chance of scoring each question correctly. The team will then have a chance to choose which scoring system (options are analogous to the 3X3 listed above but might differ in the actual multipliers) they wanted to be graded on for bonus scores. After completing the round of quiz scored by this chosen scoring system, the teams gets to choose whether to gain yet another batch of bonus by choosing to get this additional bonus by selecting whether they want us to score them based on a randomly selected team who either (i) read about 2 different industries or (ii) both read the same. There is one round when teams are making team submissions only where we let people choose which industry they want to read from (as opposed to randomly assigned).

The local business competition hosts have agreed to post the results, scores from each subportion of each task and under various actual scoring systems (include those not chosen), publicly after the event to ensure transparency since the winners will win materials awards and recognitions.

I hold this business competition in multiple continents (5 continents, and multiple locations) to generate more variance in the performance. Baseline variance in performance will be different by location for various tasks, and we will test the prediction by a theory/model by Alex Chan to see if the returns to diversity (for tasks that exhibits diversity benefits) are related to the variance of the underlying distribution from which types/genders of people/team members are drawn. We will also pool the data across and generate pools with varying variances to test this theory.

Participants of the business competition are randomized into teams based on a random number and a random color assigned to them upon arrival at the competition (e.g. someone will get a green card with the number 8). The pairs who received the same number (but different color) will pair up to form a team.

Individual

Yes

300 teams. Because business competition participant numbers are not predictable, we will chronologically include business competition data up until and including the one session that pushed the total teams to over 300 teams (e.g. if the first 5 sessions [starting September 2024] involved a total of 280 teams, and the 6th session included 40 teams, then the data from sessions 1-6 will be included the total number of teams will be 320; and the next two fields should be adjusted accordingly) AND that same session led to all 4 continents being covered (in which case the sample can be larger than 300 teams as well). It will start at South Africa, then Ireland then Hong Kong, then Australia and USA (if by Hong Kong, we already have 300 teams, we still incorporate Australia and then with the first session in the USA).

600 contestants (2* the above)

150 (or 50%) mixed gender teams (male+female), 75 (or 25%) all female, 75 (or 25%) all male. For industry knowledge and map knowledge, exactly 50% will get diverse information and 50% will get non-diverse information