The Hidden Costs of Poverty: Mental Load and Technology Learning

Workers are individually randomly assigned to one of two treatment arms. In the cash-relief arm (Arm 1, n = 200), each worker receives a cash advance of $38, approximately five days of earnings and 2–3 weeks of typical lean-season salary delivered privately at their home two days before the work session. The advance is framed as an early installment of their expected compensation, consistent with occasional company practice. Workers in this arm receive the remaining expected earnings plus their actual piece-rate earnings at the end of the session; total compensation is identical across arms, with only the timing of the first installment varying. In the control arm (Arm 2, n = 200), workers receive no advance and no unusual contact before the session; all compensation is paid at the end of the session, so they arrive in their natural lean-season financially stressed state. On arrival, all workers are offered a binary technology choice: (A) manual corn shelling by hand, which requires no preparation and can begin immediately; or (B) a mechanical corn sheller, which yields higher piece-rate output but requires self-assembly after watching a five-minute instructional video on a tablet. The machine is provided at zero monetary cost. Workers may switch back to manual shelling at any time; pay depends solely on total output regardless of method. This design holds liquidity, information, and compensation constant across arms, isolating the cognitive margin of technology adoption.

2027-04-01

2027-05-31

H1 (confirmatory, primary): D_i — binary indicator equal to 1 if the worker chooses the mechanical corn sheller (machine adoption attempt). Tested at alpha = 0.05, Step 1 of fixed-sequence Holm-Bonferroni procedure.

H2 (confirmatory, secondary): D*_i — binary indicator equal to 1 if the worker both chooses the machine (D_i = 1) and successfully completes assembly without abandoning (A_i = 0). This is the composite successful adoption indicator. Tested at alpha = 0.05, Step 2 of fixed-sequence Holm-Bonferroni, conditional on H1 being confirmed.

D_i is recorded by a trained enumerator at the technology choice station on the session day. Workers indicate their choice on a private paper form; enumerators record the choice and the time. D*_i is the product of D_i and (1 - A_i), where A_i is the abandonment indicator. A_i = 1 if the worker chose the machine but stopped the assembly process before completion and switched to manual shelling. Assembly completion is confirmed by the enumerator upon successful operation of the machine. The confirmatory analysis uses a probit regression specification: D_i = alpha_0 + alpha_1*Cash_i + alpha_2'X_i + error, where Cash_i = 1 for cash-relief workers. Confirmation of H1 requires alpha_1 > 0 at alpha = 0.05. Complementary logistic regression and chi-square tests of proportions are also reported. H2 mirrors this specification with D*_i as the outcome.

H3 (exploratory, no pre-specified direction): A_i — binary assembly abandonment indicator, conditional on D_i = 1.

Additional secondary outcomes:
- Total piece-rate earnings ln(Y_i) over the four-hour session
- Assembly time in minutes (conditional on D*_i = 1)
- Pre-choice subjective probability of successful assembly, p-hat_i (0-100 scale)
- Pre-choice expected assembly time, t-hat_i (minutes)
- Pre-choice expected production, E-hat_i (earnings in USD)
- Morning Perceived Stress Scale (PSS-3) score — financial stress mediator
- Sleep quality, duration, and latency — sleep mediator

Ai ∈ {0,1}: Assembly abandonment, conditional on Di = 1 — whether the worker who attempted the machine failed to complete assembly and reverted to manual shelling (exploratory hypothesis H2).
ln(Yi): Natural log of total piece-rate earnings from the four-hour production session (secondary confirmatory hypothesis H3).
pi ∈ [0,1]: Worker's self-assessed probability of successfully completing assembly, elicited after viewing the instructional video but before the technology choice form is distributed (belief channel — auxiliary exploratory test).
t̂i: Worker's expected time (in minutes) to complete assembly, elicited after viewing the instructional video but before the technology choice form is distributed (belief channel — auxiliary exploratory test).
Êi: Worker's expected earnings from choosing the machine (in USD), elicited after viewing the instructional video but before the technology choice form is distributed (auxiliary exploratory test).
Sleep quality: Subjective sleep quality (0–10 scale), sleep duration (hours), and sleep latency (minutes), collected in the morning survey battery (mechanism analysis).
Morning PSS-3 score: Three-item Perceived Stress Scale sum (0–12) measuring financial preoccupation on arrival (first-stage manipulation check, H4).

H3 (abandonment) is estimated via linear probability model and logistic regression, conditional on D_i = 1, with inverse probability weighting (IPW) to correct for selection into machine adoption. No pre-specified rejection threshold; results in either direction are treated as theoretically informative, adjudicating between bandwidth-depletion (Mani et al., 2013) and distraction-facilitation (Dang et al., 2016) accounts of poverty and procedural learning.

Earnings are measured by weighing individual output at the end of the session and multiplying by the piece-rate. Assembly time is recorded from the start of the instructional video to confirmed machine operation. Pre-choice beliefs are elicited verbally by enumerators immediately after the instructional video, before workers submit their technology choice forms. PSS-3 is the three-item financial preoccupation subscale of the Perceived Stress Scale, administered on the morning of the session day and again post-session. Sleep measures are self-reported (quality 0-10, hours slept, minutes to fall asleep).

Causal mediation analysis (Imai et al., 2010) is used to estimate the average causal mediation effect (ACME) of financial stress and sleep on adoption outcomes, with 1,000 bootstrap iterations and sensitivity analysis via the rho-star critical value approach.

Two-arm between-subject randomized controlled trial. Workers are randomized at the individual level to one of two arms with equal allocation (1:1 ratio). No clustering; randomization is by individual. All sessions take place during the lean season (April-May 2027) in rural northern Haiti, ensuring that confounds of seasonal comparison designs (temperature, agricultural labor demand, nutritional stock, worker composition) are held constant by construction.

Sample size: 400 total (200 per arm). Planned attrition rate: 15%. Required sample per arm for 80% power at the primary moderate effect (delta-p = 0.15, p_C = 0.30, p_T = 0.45, alpha = 0.05) is 163; 200 per arm provides a buffer. Sessions take place in groups of 20-25 workers at the firm's premises.

Eligibility: Adults aged 18-65 residing in study communities, primarily dependent on farming or farm-labor income, physically capable of four hours of manual labor, and available for a full-day session. Screening occurs at least one week before the session during a routine pre-employment visit.

Treatment assignment: Individual-level randomization using computer-generated assignment. Workers are informed of neither the study nor their arm assignment. The session is delivered by the company supervisor under normal operating conditions.

Multiple testing: Confirmatory hypotheses H1 and H2 are tested using a fixed-sequence Holm-Bonferroni procedure in the order H1 -> H2. H1 is tested at alpha = 0.05 (unadjusted). H2 is tested at alpha = 0.05 conditional on H1 being rejected. H3 is pre-classified as exploratory with no rejection threshold.

Not available

Individual-level randomization using a computer-generated random assignment list, produced in advance of any data collection. Assignment is conducted in the study office prior to worker contact. Workers are assigned sequentially as they are enrolled through the firm's labor roster. The randomization list is generated using statistical software (Stata or R) with a fixed seed documented in the pre-analysis plan. Enumerators delivering the cash advance are aware of arm assignment for logistical purposes only; session supervisors are not told which workers received an advance.

Individual worker. No clustering. Each eligible worker is independently assigned to the cash-relief arm (Arm 1) or control arm (Arm 2) with equal probability (1:1 allocation).

No

Not applicable. Randomization is at the individual level; there is no clustering. 400 individual workers total (200 per arm).

400 workers (individuals). An observation is considered complete only if the worker participated in the full four-hour production task and consented to the post-session survey. Data collection continues until 400 completed post-session surveys are reached or at the end of the scheduled lean-season field period (April-May 2027).

Arm 1 (Cash-relief): n = 200 workers
Arm 2 (Control): n = 200 workers
Total: N = 400 workers

Primary hypothesis H1 (entry margin, binary outcome D_i): - Control-arm baseline attempt rate (p_C): 0.30 - Treatment-arm attempt rate (p_T): 0.45 (moderate scenario, delta-p = 0.15) - Required sample per arm at 80% power, alpha = 0.05 (two-sided): 163 - Power at n = 200 per arm (moderate scenario): 0.91 - Power at n = 200 per arm (small effect, delta-p = 0.09): 0.48 (underpowered; reported as lower bound) - Power at n = 200 per arm (large effect, delta-p = 0.19): 0.98 - Attrition adjustment: 15% assumed; 163/0.85 = 192 required; 200 recruited Standard deviations are mechanically determined by the binary outcome: SD = sqrt(p*(1-p)). At p_C = 0.30: SD = 0.458. At p_T = 0.45: SD = 0.497. H2 (perseverance margin): Secondary confirmatory hypothesis. Power depends on attempt rate from H1. At baseline attempt rate of 0.50 and delta-p = 0.19 in perseverance, power at n = 100 attempters per arm is approximately 0.80.