Experimental Design
Survey 1
Recruitment: We conduct door-to-door sampling in urban neighborhoods in Patna, India. To minimize spillovers, we will randomly sample 1 in every 2-4 houses on a given street.
Sample: After approaching a house, we explain we are conducting a survey and ask to speak with a married woman who is aged between 18-60 and has completed 5th grade. We also check that their husband currently lives in Patna, is also aged 18-60, and passed 5th grade.
Demographics: After informed consent, we collect basic demographic details from the wife including her work history and education. In this section, we ask the wife whether she is interested in new part-time work opportunities using a 4-point Likert scale. We stop the survey after the demographics section if participants say they are “not at all interested” in new job opportunities.
Description of workplace: We then inform participants of the women-only call center office we have recently opened and that we are looking to hire women for the call center on a part-time basis. We explain the job to participants and highlight that this is a one-time, two-week offer. We hand the participants a flyer about the workplace and explain it to them. We then tell them the job includes a peer support program, in which they will be matched with a peer from a different office who will provide them with career advice. Finally, we tell participants the salary (which we randomly vary across participants). We then ask participants whether they would like to apply for the job and peer support program.
Survey 2
Our team returns to the household at the scheduled time. We proceed with the second survey if both the husband and wife are present and consent to participate.
Job and peer support programs: We re-explain the job to the husband and wife and confirm that the wife is still interested in the job. We then explain the peer support program to the couple in more detail.
We then briefly explain the four different peer support programs we offer. The programs are designed to vary the amount of spousal jealousy husbands may feel if their wife were paired with a male peer. One of the programs, (the “video” program) minimizes the amount of jealousy husbands may feel, while also ensuring there are no differences in the content or quality of support participants would receive based on the gender of their peer.
The four programs are the:
1. “One-on-one” program: In this program, participants speak directly with their peer without anyone else being present.
2. “Husband” program: husbands are given the option of joining all conversations between their wife and the peer. If they are unable to join any of the peer support sessions they are sent recordings of the conversations.
3. “Group” program: In this program, participants take part in group peer support.
4. “Video” program: In this program, participants do not speak with their peer but instead watch pre-recorded videos of them.
Program randomization: Next we hand the participant a sealed envelope which contains the peer support program they have been assigned to, and tell them that we will open the envelope at the end of the survey. We randomize a larger fraction of participants to the “one-on-one” program (47.5%), and “husband” program (47.5%) so that we are powered to study differences in job and peer support attendance for participants in these groups. We randomize remaining participants to the “group” program (2.5%) and the “video” program (2.5%). We stratify our randomization based on whether the wife has worked in the past (or currently) and whether they have completed 10th grade.
Fee to guarantee a female peer: Next, we explain that we do not have many female staff working in our other office who are available to provide peer support, so we are charging a fee to anyone who would like to guarantee having a female peer in the peer support program they are assigned to. We tell participants if they choose to pay the fee it will be deducted from their salary. If they choose not to pay the fee, they will receive the full salary and a “standard” peer. For the standard peer, we inform them that they will be paired with an employee from our other office. And that most of the permanent staff in that office are men. We randomize the fee size across participants to provide a willingness-to-pay measure. To ensure participant safety, when we implement the job and programs, we will assign all participants to a female peer.
Participant choices: We then go through the four peer support programs one-by-one, first explaining the program to the participant in more detail and presenting them with two flyers depicting the standard (male) peer and the female peer. We then ask participants comprehension questions about the program to ensure they understand its key features. Finally, we ask participants which peer they would like to select if they are assigned to that program. We repeat this procedure for all four programs, reversing the order in which we elicit their preferences for random subset of participants.
Implementation comprehension: Next, we ask participants two additional comprehension questions. The first, confirms that they understand their choices will not affect their program allocation. And the second, confirms that that they understand how their choices will be implemented. If they answer any of these questions incorrectly, we correct them, and ask them again. If they answer incorrectly twice, we explain things to them a third time and proceed to the next question. We then give participants the opportunity to update their decisions if they would like to. Once the participant’s choices have been finalized, the surveyor records how the household made their choices (with discussion between the husband and wife, by the wife alone, or by the husband alone).
Platonic language: After making their choices, we ask participants how they would like to be addressed if they were (hypothetically) assigned a male peer assigned in the one-on-one program. We offer them three options – (i) their name (e.g. Rashmi), (ii) their name with the suffix “ji”, a marker of respect (e.g Rashmi-ji), or (iii) “didi”, meaning sister. We hypothesize that some couples may prefer the wife be addressed as “didi” to signal a platonic relationship between the wife and the peer. We will explore heterogeneity in this outcome by an index of spousal jealousy.
Internalize preference choice experiment: Next, we test to what extent preferences for a female peer are driven by women’s own internalized preferences, and to what extent they are driven by concerns about what it would signal to their husband. We ask women to choose between two professional development videos that they will be shown when they start the job. We randomize which of the videos features a male peer and which features a female peer, giving us a measure of the wife’s preference for a female peer holding the content of the video fixed. We then test the extent to which these preferences change when the wife’s choices are observed by the husband, randomizing whether we ask this question in front of the husband or in private.
Spousal jealousy: We will also measure spousal jealousy and gender norms regarding impropriety using four survey questions. We ask these questions to the wife, in private. The questions ask (i) directly about spousal jealousy, (ii) personal beliefs about how appropriate it is for married women to interact with other men (iii) second-order beliefs, and (iv) a measure of controlling behavior on the part of the husband. We will create an index using these questions. As secondary analysis, we will analyze heterogeneous treatment effects using the index. We also collect data on whether women interacted with men in their last job (if they previously worked).
Willingness-to-pay: To elicit willingness-to-pay preference over a wider range of prices, we ask participants which peer they would select for each program if the fees were (hypothetically) higher. We only ask participants what they would do if they selected a female peer in the main choice experiment (assuming if they did not pay the fee at a lower price, they would not do so at a higher price).
Beliefs and spillovers: We then measure participants’ expectations that the standard peer will be a man. To document spillovers, we again measure whether or not participants have spoken to anyone else who participated in the study, and if so, what they spoke about.
Out-of-sample validation exercise: To validate that our treatment arms are varying spousal jealousy as we hypothesize, we will conduct an additional data collection exercise with a separate sample of married women who are not offered jobs. We will explain the job and peer support program to these participants and ask them to explain why they think some couples may be comfortable with a male peer in the “group”/“video”/“husband” programs but not in the “one-on-one” program. We ask these questions to a separate sample to minimize any social desirability bias or discomfort that may come from ascribing one’s own choices to spousal jealously.
Conclusion: We schedule and record when participants would like to come to the call center for their first day of work. We then open the envelope to reveal to participants which program they were assigned to and implement their choice for that program.
Sample: Our final sample for this experiment will consist of participants who complete the second survey (and do not drop out before their treatment assignment in the envelope has been revealed to them).
Workplace attendance: We will record participants workplace attendance and participation in the peer support program. At the workplace, we will ask women additional questions including who they interacted with in the past week, and the fraction that were men and women.