Preference for Homogeneity? Occupational Segregation and Gender Stereotypes (Field Part)

- We vary the initial gender composition of the four-person group (all female or all male) and whether a woman or man is added to the group
- We also vary the task the groups perform (building shelves, memory, network).

2021-05-31

2021-08-31

- Choices (binary productivity estimates) between a gender-homogeneous team that remains homogenous and a team that becomes heterogeneous (six choices, one for each combination of task (building shelves, memory, or network) and initial gender composition (all-female or all-male)).

- Beliefs about the choices of (eleven) other participants between a gender-homogeneous team that remains homogenous and a team that becomes heterogeneous (six choices, one for each combination of task and initial gender composition).
- Binarized version of this belief measure (six choices, one for each combination of task and initial gender composition).
- Choices (binary productivity beliefs) between an all-female team and an all-male team (three choices, one for each task).

Gender-homogeneous groups consisting of four women or four men perform one of three different tasks. We then replace one worker with a new subject (either female or male), and study how this affects performance expectations and beliefs about the performance expectations of others.

The experiment consists of three parts and a post-experimental questionnaire.

In part 1, we investigate the performance assessments of groups that initially consist of either four men or four women (stage 1) where we replace one person by either a man or a woman (stage 2), that is, we elicit the productivity estimates about the performance of the new (stage 2) group that can be either more gender-heterogeneous or as homogeneous as the old (stage 1) group. We employ stereotypically male and female tasks: assembling a bookshelf and solving a memory game, and a task for which men and women are expected to provide complementary inputs: writing down chains of movies and actors. We incentivize participants’ choices by comparing their predictions to the reference performances obtained in so-called group sessions. In the group sessions, different subjects worked in groups on the tasks.

In part 2, we elicit participants’ beliefs about the fraction of other participants betting on each group. The participants are asked how many of 11 randomly chosen other participants bet that the group with a man performed better. We elicit subjects’ beliefs about others’ performance expectations for each combination of the initial group composition and task.

In part 3, we also obtain a measure of the task stereotypes by eliciting participants' expectations of the performance of the all-female and all-male groups in stage 1.

The experiment ends with a questionnaire in which we elicit participants’ gender, age, job, and educational attainment. We also elicit participants’ experience with recruitment decisions. Finally, to obtain a measure of whether the participants paid attention to their choices or not, we ask them to rate how attentive they were during the experiment.

The randomization is done by a computer.

The randomization unit is the individual.

No

400 participants.

400 participants.

400 participants (the treatment variation is within-subjects).

See pre-analysis plan.