Demand for traditional and modern health care in Sub-Saharan Africa: Experimental evidence on beliefs about the origin of illnesses and treatment effectiveness

Participants will be randomly assigned to one of four groups: (1) a control group, (2) a priming group, (3) an information treatment on the biomedical origins of the disease (“origin” treatment), or (4) an information treatment on the effectiveness of modern medical care (“effectiveness” treatment). The two treatment arms provide scientifically validated information, focusing either on the biomedical etiology (i.e. the cause) of the disease or on the effectiveness of modern medical treatment, respectively.

2026-01-14

2026-02-10

Primary outcomes are grouped into belief-based outcomes, choice outcomes, and allocation outcomes:

(1) Beliefs and uncertainty
(a) Beliefs about the root causes (etiology) of epilepsy/schizophrenia and the associated subjective uncertainty;
(b) Beliefs about the effectiveness of modern or traditional medicine and the associated subjective uncertainty;
(c) Beliefs about the effectiveness of modern medical treatment for epilepsy/schizophrenia, elicited using an incentivized belief-elicitation task.

(2) Health care provider choice
(d) Hypothetical choice of health care provider (not incentivized).

(3) Donation allocation task
(e) Allocation of a real monetary endowment between two NGOs: one facilitating knowledge exchange among African traditional healers and one facilitating knowledge exchange among African medical doctors on best-practice treatment options (incentivized).

(a) Beliefs about the root causes (etiology) of epilepsy/schizophrenia and the associated subjective uncertainty (not incentivized):
Participants indicate on a scale from 0-10 how likely they perceive a certain disease (epilepsy or schizophrenia) to have a spiritual root cause or to have a biomedical root cause. To capture subjective uncertainty, participants additionally report the minimum and maximum perceived likelihood (0-10) for each root cause. Uncertainty is operationalized as the width of the reported belief interval.

(b) Beliefs about the effectiveness of modern or traditional medicine and the associated subjective uncertainty (not incentivized):
Participants report their beliefs about the effectiveness of modern and traditional medicine for treating epilepsy/schizophrenia on a scale from 0 to 10. Subjective uncertainty is elicited by asking participants to indicate the minimum and maximum perceived likelihood (0-10) that modern or traditional medicine is effective for treatment. Uncertainty is measured as the width of the reported interval.

(c) Beliefs about the effectiveness of modern medical treatment for epilepsy/schizophrenia (incentivized):
Participants are presented with a summary of a meta-analysis assessing the effectiveness of modern medical treatment for epilepsy/schizophrenia. They are then asked to estimate the proportion of patients in the study who were symptom-free after one year of receiving modern medical treatment. Participants receive a bonus payment of USD 3 if their estimate lies within ±3 percentage points of the true value.

(d) Hypothetical health care provider choice (not incentivized)
Participants indicate which type of health care provider a person affected by epilepsy/schizophrenia should consult to receive appropriate care. Response options include: pharmacist, traditional healer, religious leader, hospital/health center/health care facility, and medical doctor or health worker (e.g., nurse).

(e) Donation allocation task (incentivized)
Participants are asked to allocate an amount of USD 50 between two NGOs: one facilitating knowledge exchange among African traditional healers (ANHA) and one facilitating knowledge exchange among African medical doctors on best-practice treatment options (PAHPO). For a randomly selected subset of participants, the donation decision is implemented, with the corresponding amount transferred by the research team.

We conduct an online survey experiment with participants recruited via Facebook advertisements. Facebook users who view and click on the advertisements are redirected to the study platform, where they complete a 20-minute survey. The survey elicits participants’ beliefs about the causes of, and treatment options for, epilepsy and schizophrenia.

Participants are randomly assigned to one of four experimental groups: (1) a pure control group, (2) a priming group, (3) treatment group 1 (“origin”), or (4) treatment group 2 (“effectiveness”). In addition, participants are subject to a second randomization that determines the order in which the disease-specific modules are presented, with participants receiving questions on epilepsy first and schizophrenia second, or vice versa.

The overall experimental structure is as follows. First, we estimate within-individual relationships between beliefs about disease origins and treatment effectiveness for several diseases to study how widespread spiritual beliefs are. Second, we measure prior beliefs and the level of uncertainty about disease etiology and treatment effectiveness. Third, we implement an information experiment to identify causal effects of information on posterior beliefs and uncertainty about disease etiology and treatment effectiveness. Fourth, we use an incentivized belief-elicitation task to study belief updating and resulting changes in the demand for health care services. Finally, we employ an incentivized measure to assess donation decisions. The exact experimental flow for each of the four groups is outlined below and presented as graphical overview in Figure 1 (separately uploaded).

____________________________
Group 1 (pure control):
____________________________
(0) Within estimate of origin-effectiveness (for several diseases)
(1) Elicitation of beliefs and uncertainty about the root cause of illness #2 (epilepsy/schizophrenia)
(2) Elicitation of beliefs and uncertainty about effectiveness of traditional and modern medicine for treating illness #2 (unincentivized + incentivized)
(3) Health care provider choice
(4) Donation task

____________________________
Group 2 (priming group):
____________________________
(0) Within estimate of origin-effectiveness (for several diseases)
(1) Elicitation of beliefs and uncertainty about the root cause of illness #1
(2) Elicitation of beliefs and uncertainty about effectiveness of traditional and modern medicine for treating illness #1 (unincentivized)
(3) Health care provider choice
(4) Elicitation of beliefs and uncertainty about the root cause of illness #2
(5) Elicitation of beliefs and uncertainty about effectiveness of traditional and modern medicine for treating illness #2 (unincentivized + incentivized)
(6) Health care provider choice
(7) Donation task

____________________________
Group 3 (treatment group “origin”):
____________________________
(0) Within estimate of origin-effectiveness (for several diseases)
(1) Elicitation of prior beliefs and uncertainty about the root cause of illness #1
(2) Elicitation of prior beliefs and uncertainty about effectiveness of traditional and modern medicine for treating illness #1 (unincentivized)
(3) Health care provider choice
(4) Information treatment “origin”: participants read a text about scientifically validated information on the biomedical origin of the diseases
(5) Elicitation of posterior beliefs and uncertainty about the root cause of illness #2
(6) Elicitation of posterior beliefs and uncertainty about effectiveness of traditional and modern medicine for treating illness #2 (unincentivized + incentivized)
(7) Health care provider choice
(8) Donation task

____________________________
Group 4 (treatment group “effectiveness”):
____________________________
(0) Within estimate of origin-effectiveness (for several diseases)
(1) Elicitation of prior beliefs and uncertainty about the root cause of illness #1
(2) Elicitation of prior beliefs and uncertainty about effectiveness of traditional and modern medicine for treating illness #1 (unincentivized)
(3) Health care provider choice
(4) Information treatment “effectiveness”: participants read a text about scientifically validated information on the effectiveness of modern medical treatment
(5) Elicitation of posterior beliefs and uncertainty about the root cause of illness #2
(6) Elicitation of posterior beliefs and uncertainty about effectiveness of modern or traditional medicine for treating illness #2 (unincentivized + incentivized)
(7) Health care provider choice
(8) Donation task

Individuals are randomized via computer-generated random assignment within the online survey.

Individual respondent

No

Not applicable. No clusters.

There is no fixed sample size. We aim to reach as many participants as possible by recruiting participants via Facebook ads. Results from previous studies suggest that we might be able to reach about 6,000 participants with the determined budget.

There is no fixed sample size. We aim to reach as many participants as possible by recruiting participants via Facebook ads. Participants will be allocated to one out of four treatment arms with equal probabilities.

Given that the final sample size is unknown, power calculations or estimating the exact minimum detectable effects size renders difficult. Yet, previous studies suggest that we might be able to reach about 6,000 participants in total. For now, we assume that we will reach this number. Setting the significance level of alpha to 0.0167 (to adjust for the comparison of three groups against each other) and setting power (beta) to 80%, we would be able to detect the following MDEs for the two outcomes "donation task" (binary outcome) and "posterior beliefs" (continous outcome on a scale from 0-10): Donation task: MDE is equal to 0.072 percentage points Posterior beliefs: MDE is equal to 0.362 scale points (equal to 0.25 SD)