Experimental Design Details
This study employs a 3×3 between-subjects factorial design that independently varies two dimensions of constraint salience in a discrete choice experiment (DCE) on solar panel financing in Yemen. Participants are randomly assigned to one of nine cells defined by the crossing of DCE timing and constraint framing conditions using SurveyCTO's random() function.
DCE Structure
Each participant sees 13 choice sets from one of three pre-assigned blocks (39 unique scenarios total). Each choice set presents four alternatives: a cash purchase from a bank, a cash purchase from a merchant, a loan from a bank, and a loan from a merchant. All alternatives vary in total cost (300,000–700,000 YER), and loan options additionally vary in duration (6 or 12 months), flexibility (none or skip one payment), and documentation requirements (simple or full). Of the 13 choice sets, 7 contain an alternative that is unambiguously identifiable as a cash purchase, enabling construction of a participant-level cash preference rate.
Experimental Dimensions
Dimension 1: DCE Timing
The first dimension varies the survey modules that precede the DCE, thereby manipulating the extent to which participants have reflected on their financial situation before making financing choices.
(1) Early. The DCE comes right after demographics and basic solar questions. Participants have not yet thought about their budget, providing a baseline measure of financing preferences under minimal constraint salience.
(2) After Expenditure. The DCE comes after participants have listed their household spending across 14 categories (food, rent, electricity, water, debt repayment, healthcare, cooking fuel, hygiene, communications, transport, petroleum, diesel, education, and other) for both the past 30 days and the expected next 30 days. The act of enumerating specific expenses makes budget constraints vivid and cognitively available when participants then encounter the DCE.
(3) After Expenditure + Financial Assessment. The DCE comes after the expenditure listing plus a financial assessment module covering: income sources, regularity, and volatility; financial access (bank accounts, prior loan experience, collateral availability, borrowing from family); attitudinal measures (debt aversion, religious norms around borrowing, generalized trust and risk tolerance); perceived consequences of loan default across 10 specific dimensions (asset seizure, house seizure, legal action, imprisonment, debt collector visits, social shame, family pressure, denial of future credit, workplace problems, and community harm); interest rate expectations; and inflation expectations. This condition makes both budget constraints and the specific risks and considerations associated with borrowing salient.
Dimension 2: Constraint Framing
The second dimension varies the instructions presented to participants immediately before the DCE choice tasks, manipulating the frame within which participants evaluate their options.
(1) Constrained. "Think about the options based on your current financial situation. Consider your income, expenses, and financial obligations when choosing." This instructs participants to apply their actual constraints.
(2) Unconstrained. "Imagine you have 1,000,000 YER in the bank" (Early timing) or "Imagine you have no financial constraints" (After Expenditure and After Expenditure + Financial Assessment timing). "Choose what you genuinely prefer without worrying about your current financial situation." This is designed to elicit preferences stripped of constraint-driven considerations, revealing what participants would choose if financial barriers were removed.
(3) Neutral. No framing instruction is provided. Participants interpret the hypothetical scenario however they naturally would. This reveals how people naturally approach hypothetical financing decisions and provides the benchmark against which the other conditions are compared.
The Nine Treatment Cells
Cell 1 (Early × Constrained). Participants complete demographics and basic solar questions, then receive the constraint instruction before the DCE. Any constraint salience arises solely from the textual instruction, without prior experiential activation. This cell tests whether a framing instruction alone, absent reflection on one's budget, is sufficient to shift financing preferences relative to the neutral baseline.
Cell 2 (Early × Unconstrained). Participants complete demographic and basic solar questions, then receive unconstrained instructions before the DCE. Having neither reflected on their budget nor been encouraged to apply constraints, and having been explicitly instructed to set financial concerns aside, this cell provides the closest approximation to preferences free of constraint considerations. It serves as the aspirational benchmark for what participants would choose if financial barriers were removed.
Cell 3 (Early × Neutral). Participants complete demographics and basic solar questions, then encounter the DCE without framing instructions. This cell serves as the primary reference condition, capturing how participants naturally approach hypothetical financing decisions with minimal prior reflection and no directional guidance.
Cell 4 (After Expenditure × Constrained). Participants enumerate household expenditures across 14 categories for two time periods, then receive the constraint instruction before the DCE. This cell combines experiential and instructional constraint salience. Comparing Cell 4 to Cell 1 isolates the incremental effect of expenditure reflection beyond the constraint instruction alone. Comparing Cell 4 to Cell 6 isolates the incremental effect of the constraint instruction beyond expenditure reflection alone.
Cell 5 (After Expenditure × Unconstrained). Participants enumerate household expenditures, then receive the unconstrained instruction before the DCE. This cell creates a deliberate tension between the experiential budget activation and the instruction to disregard constraints. Comparing Cell 5 to Cell 2 tests whether the unconstrained instruction elicits similar preferences regardless of prior budget reflection — that is, whether hypothetical preference elicitation is robust to preceding survey context. Comparing Cell 5 to Cell 6 tests whether the unconstrained instruction shifts preferences away from constraint-driven choices among those who have recently reflected on their budget.
Cell 6 (After Expenditure × Neutral). Participants enumerate household expenditures, then encounter the DCE without framing instructions. This cell isolates the pure effect of budget reflection under natural interpretation conditions. The comparison between Cell 6 and Cell 3 provides the cleanest estimate of how expenditure enumeration alone affects financing preferences, absent any framing intervention.
Cell 7 (After Expenditure + Financial Assessment × Constrained). Participants complete both the expenditure module and the financial assessment module, then receive the constraint instruction before the DCE. This represents the maximum constraint salience condition, combining budget reflection, deliberation on borrowing risks and consequences, and an explicit constraint instruction. Comparing Cell 7 to Cell 4 tests the incremental effect of the financial assessment module — including reflection on default consequences, repayment capacity, and institutional trust — beyond expenditure reflection and constraint framing alone.
Cell 8 (After Expenditure + Financial Assessment × Unconstrained). Participants complete both modules, then receive the unconstrained instruction. This cell tests whether the unconstrained instruction can elicit constraint-free preferences even after extensive reflection on budget constraints, default consequences, and borrowing risks. The comparison between Cell 8 and Cell 2 provides the most stringent test of whether hypothetical preference elicitation is robust to preceding survey context, as the two cells share the same framing instruction but differ maximally in the preceding cognitive experience.
Cell 9 (After Expenditure + Financial Assessment × Neutral). Participants complete both modules, then encounter the DCE without framing instructions. The comparison between Cell 9 and Cell 3 captures the total effect of maximum prior reflection on financing preferences under natural interpretation conditions. The comparison between Cell 9 and Cell 6 isolates the incremental effect of the financial assessment module beyond expenditure reflection, under neutral framing.