If someone asked you to prove that the person sitting next to you is your spouse, what would you show? A marriage certificate? A shared last name? 

In many parts of the world, you cannot rely on legal certificates because many couples are considered married without formal certification, and last names may differ. Now imagine that, for your research, you must be certain that the two people seated across from you are, in fact, spouses. The integrity of recruiting couples rests on a deceptively simple question: how do you verify a marriage using only conversation?

This was one of the more complex challenges we faced in our intra-household financial privacy study, conducted in Nairobi, Kenya- we had to design a system that ensured we recruited real couples.

Financial privacy occurs when one partner conceals their finances from their spouse. Financial privacy may help empower women by helping them retain control over resources, but it could also backfire by causing household conflict. To study this, Dr. Prachi Jain, Anisha Singh, Dr. Jessica Hoel, Brian Baraza, and Gitanksh Sethi ran laboratory-based experiments to understand how women and men in heterosexual relationships value financial privacy, whether this varies by gender, and the expected long-term effects on household dynamics. 

In our study, couples arrive at a laboratory, where they play a series of incentivized games with monetary consequences under carefully designed conditions. We also use an extensive survey and present respondents with vignettes, using realistic scenarios about couples like them, to understand beliefs about financial privacy and its effects.

The experiments allowed us to observe how the same individuals made financial decisions under varying levels of privacy. The lab created a confidential space where partners could make decisions without their spouses’ influence. Unlike many household studies, our study required both partners to make linked, yet potentially private, financial decisions. In doing so, we study individual and couple decisions in scenarios in which earnings were kept private or revealed to their spouse, which would have been impossible to observe in real-world settings.

Over 500 couples from across Nairobi came to Busara’s lab to participate, which brought us to our challenge of verifying coupledom. 

The first step began at recruitment. Unlike studies that recruit convenience samples, we needed both partners to participate in our study. To reduce the number of incomplete pairs, we first recruited men. In our context, men are more likely to face work-related and other external constraints that make participation more difficult. They may also play a larger role in deciding whether a couple joins a study. Reaching men first made joint attendance more likely. We gave couples the option to attend sessions together or separately, and offered a bonus once both spouses completed sessions, with a slightly higher bonus if they attended together. These adjustments reduced attrition from 40% to 10%.

The next layer of complexity was verifying that couples were indeed married, a sensitive and essential component. We initially co-created creative verification questions with our field team, which we asked each partner separately and compared their answers: Which staple does your spouse prefer- rice, chapati, or ugali? Which side of the bed do you sleep on? 

We quickly learned that we had wrongly assumed that shared life automatically produces shared recall of small, everyday details and what intimacy may actually look like in our context. Love may be blind, but it can also be a little forgetful. 

Through iteration, we realized that simple, factual questions worked best: How many years have you been together? Where did you meet? What is your home county? When responses differed substantially, enumerators followed up on children’s names and ages or parents’ names. This process finally identified eight ineligible pairs!

Ultimately, we find that women are significantly more accepting of financial privacy than men (44% vs. 39%, p=0.03). Yet our experiments uncovered striking heterogeneity: 36% of women and 26% of men preferred hiding income; 20% of women and 31% of men preferred transparency. 

As we continue to analyze our data, we are reminded that research on interpersonal relationships begins long before running regressions, requiring sensitivity to context, humility about assumptions, and constant adaptation in the field.

We would like to thank the Digital Credit Observatory (DCO) at CEGA and the WEE-DiFine initiative at BRAC for their funding support, which made both the initial inquiry phase and the subsequent extension phase of this study possible.