Experimental Design Details
This study uses a framed field experiment design to measure how social observation and long-run cost information shape women's willingness to pay (WTP) for familiar and unfamiliar menstrual products in rural Uganda. The design combines a between-subject factorial randomization across two independent treatment dimensions with an incentive-compatible WTP elicitation mechanism and a comprehensive pre-randomization baseline survey. The study is conducted among women aged 25 and older who menstruated in the past three months, residing in Kayunga District, Uganda, in partnership with Smart Girls Uganda and ESONGA.
Factorial structure
The two treatment dimensions are crossed, generating a 3 × 2 factorial design with six experimental cells. The first dimension is the observer condition (three levels: private, peer-observed, husband-observed). The second dimension is the information condition (two levels: information, no information). Randomization to both dimensions is conducted at the individual level prior to Module 2 of the survey instrument. The observer condition is implemented physically — it determines who is present in the room during the WTP elicitation — while the information condition determines whether a printed cost-comparison card is delivered before elicitation begins.
WTP elicitation mechanism
Willingness to pay is elicited using the Becker-DeGroot-Marschak (BDM) mechanism for the product bundles (for details on the product bundles please check the PAP). Each bundle is set at approximately 15,000 UGX market value, with quantities varying across products to reflect their unit costs. Respondents receive a 15,000 UGX participation endowment at the start of Module 2. For each product, respondents state the maximum price they are willing to pay. After all prices are recorded, one product is drawn at random from a lottery pot and one price is drawn at random from a set of 16 price cards ranging from 0 to 15,000 UGX in 1,000 UGX increments. If the drawn price is at or below the respondent's stated maximum for the drawn product, the transaction is binding and real: the respondent pays the drawn price and receives the bundle, keeping the remainder of the endowment. If the drawn price exceeds the stated maximum, no transaction occurs and the respondent keeps the full endowment. The order of product elicitation is randomized within participants by SurveyCTO. The BDM mechanism makes truthful revelation of maximum WTP the dominant strategy, removing incentives to understate or overstate.
Observer condition: implementation
In the private arm, the respondent completes the BDM alone with the enumerator. No observer is present at any point during Module 2. This arm establishes the unconstrained preference baseline.
In the peer-observed arm, a known female community peer is present throughout Module 2. Both women are study participants recruited together. Role order — who completes the BDM first — is determined by a random draw at the start of the session. The first mover completes all product BDMs while the second mover observes. Roles then reverse: the second mover completes all BDMs while the first mover waits quietly. By design, the second mover observes the first mover's stated prices before completing her own elicitation. This within-arm variation in observational exposure is pre-specified as a mechanism analysis and is not the primary definition of the peer treatment.
In the husband-observed arm, the respondent's husband or male partner is invited into the session at the start of Module 2, after the Module 1 baseline survey has been completed privately. Within this arm, respondents are further assigned to one of two pre-specified sub-conditions: in the woman-decides sub-arm, the husband is present and observes silently while the woman states her own prices independently; in the joint-decision sub-arm, the husband and wife are asked to discuss and agree on a single price together for each product. The husband receives a brief standardized introduction to the products at the start of his participation, since he has not attended Module 1. The within-arm variation between woman-decides and joint-decision is pre-specified as a mechanism analysis targeting intra-household bargaining and audience effects, and is not the primary definition of the husband treatment.
Information condition: implementation
In the no-information condition, respondents receive a standardized product familiarization and BDM explanation with no additional information about product costs.
In the information condition, respondents receive a printed cost-comparison card before the BDM elicitation begins. The card is delivered by a designated research team member at the start of Module 2, prior to SurveyCTO data entry, and remains visible throughout the elicitation. The card displays the total cost of using each of the products exclusively for 1, 5, and 10 years, computed using local retail prices and menstrual cycle parameters estimated from pilot data collected in the study area. The enumerator reads the figures aloud for all three time horizons and explains that the costs reflect an average woman in this region using only that one product throughout. Where relevant, the enumerator connects the long-run cost figures to the specific bundle the respondent could take home today.
Pre-analysis and mechanism structure
The primary treatment comparisons are the three observer arms (private, peer, husband) and the two information conditions, estimated as main effects and interactions in the factorial design. The within-arm variation in the peer arm (first mover vs. second mover) and the husband arm (woman decides vs. joint decision) are pre-specified mechanism analyses, distinguished from the primary treatment comparisons in the pre-analysis plan. Heterogeneity analyses by baseline stigma, period poverty, partner involvement, and product awareness are pre-specified as secondary analyses.