The effect of gendered occupational titles on perceptions of peers’ and potential clients

We run a vignette study on a sample of high-skilled workers where different (female) professional figures for a specific profession are randomly presented in different scenarios, either as potential colleagues or service-providers, with the female or male gender-declination of the professional title. We measure whether the same individual, with the same observable characteristics, is ranked lower from potential colleagues or clients when presented with the female declination of the professional title.

The sample of the respondents is obtained through the local professional associations of the region.

The following professions are included in the study:
1) Tax accountant (in italian, commercialista): this is a high-skilled profession with a title that cannot be declined (gender-invariant; it could, theoretically, be declined with the article in front, il/la, which is not shown here). We show 4 profiles: 2 men, 2 women (names and last names randomly selected), random choice between age, experience, degree (bachelor or master), whether or not they are revisore contabile (sign of experience). Both men and women are presented as commercialista (gender invariant)

2) Medical doctor (in italian, dottore, dottoressa): this is a high-skilled profession with a title that has to be declined by gender. We show 4 profiles: 2 men, 2 women (names and last names randomly selected), random choice between age, experience, speciality. Men are presented as dottore, women are presented as dottoressa (gender variant)

3) Lawyer (in italian, avvocato, avvocata): this is a high-skilled profession with a title that can, but is often not, declined by gender. We show 4 profiles: 2 men, 2 women (names and last names randomly selected), random choice between age, experience, whether or not they are cassazionista (sign of experience). For the treatment group, men are presented as avvocato, women are presented as avvocata (gender variant). For the control group, men are presented as avvocato, women are presented as avvocato (gender invariant).

We present these professions in 5 different scenarios, presented in random order. Each respondent sees a minimum of 3 and a maximum of 4 scenarios depending on their own profession. We draw names and last names from a selected list of common names and last names in Italy and in the specific region of our study. The gender is always easily inferable from the name. The scenarios are as follows:

1) “The tax declaration has shown some irregularities. If you were to consult a tax lawyer, how would you rate the following profiles from 1 to 10?”. If a tax lawyer is answering, he/she does not see this scenario.

2) “You are in need of a medical check up for a skin-related issue. If you were to consult one of the following doctors, how would you rate the following profiles from 1 to 10?”. If a doctor is answering, he/she does not see this scenario.

3) “You are in need of a legal consultation for something related to family law. If you were to consult a lawyer, how would you rate the following profiles from 1 to 10?”. If a lawyer is answering, he/she sees “you are in need to collaborate on something related to family law”.

4)
a) “You are in need of a legal consultation for something related to penal law. If you were to consult a lawyer, how would you rate the following profiles from 1 to 10?”. If a lawyer is answering, he/she sees “you are in need to collaborate on something related to penal law”.

b) “You are in need of a legal consultation for something related to commercial law. If you were to consult a lawyer, how would you rate the following profiles from 1 to 10?”. If a lawyer is answering, he/she sees “you are in need to collaborate on something related to commercial law”.

2022-10-05

2022-11-18

- Differences in the rating given to scenarios 4.A and 4.B to female candidates presented with the title avvocata relative to the similar candidates presented with the title avvocato. We expect a potentially negative effect on the score when women are presented with the title avvocata. The outcome will also be compared to the differences in ratings between treatment arms in the same scenarios given to male candidates.

- Differences in the rating given to scenario 3 to female candidates presented with the title avvocata relative to the similar candidates presented with the title avvocato. We expect a zero or potentially positive effect on the score when women are presented with the title avvocata. As above, we will compare the outcome to the same outcome for male candidates.

We will explore heterogeneous effects by the gender of the respondent.

We randomly vary, directly on the survey platform, whether a female professional is presented with a female-declined job title or whether with a male-declined job title (in those professions whether the male declination for women exists and is in use).

The sample of the respondents is obtained through the local professional associations of the region.

The random variation is introduced in Scenarios 3 and 4a and b, presented above. All respondents see scenario 3, and, at random, between scenario 4a and 4b. 50% of respondents will see these scenarios with women lawyers presented with the female-declined occupational titles (avvocata), while the remaining 50% will see these scenarios with women lawyers presented with the male-declined occupational titles (avvocato).

The sample of respondents is obtained through the local professional associations. Our partner, the Equality Opportunity Committee of the Rovereto Lawyer’s association, has direct contact with their managerial offices. The survey is forwarded to the managerial offices of the different professional associations of the Province of Trento, which then forward it directly to their members.

Randomization is performed directly on the survey platform (Otree) .

The randomization unit is the individual respondent.

No

no clusters

We expect around 400 respondents. This estimate is based on the number of respondents who answered a survey on the same topic and target population in the previous year.

Respondents will be distributed equally between treatment and control, roughly 200 and 200.

the minimum detectable effect with a sample of 400 respondents is around 0.28 standard deviation for the standardized variable of the main outcome. The non-standardized variable ranges from 1 to 10, standard deviation is a priori unknown.