Shifting Beliefs Changing Behavior: Experimental Evidence on Sanitation from India

Alex Shpenev; Cristina Bicchieri; Sania Ashraf; Upasak Das

Shifting Beliefs Changing Behavior: Experimental Evidence on Sanitation from India

T0 review · 1 major / 2 minor · reviewed 2026-05-24 · grok-4.3

Challenge this review Download PDF Read on arXiv ↗

Pith's one-line read A randomized intervention in Tamil Nadu raised household toilet access by 8 percentage points and consistent usage by 6-7 points by updating beliefs about others' sanitation practices.

desk verdict The RCT reports modest gains in toilet access and use from a bundled intervention, but the mediation analysis does not cleanly separate belief shifts from other channels. read the letter →

arxiv 2312.16205 v2 submitted 2023-12-23 econ.GN q-fin.EC

Sania Ashraf , Cristina Bicchieri , Upasak Das , Alex Shpenev This is my paper

classification econ.GNq-fin.EC

keywords sanitationopendefecationsocialnormsempiricalexpectationsnormativerandomizedtrialIndiabehaviorchange

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper examines whether beliefs about how common and socially approved toilet use is can drive actual sanitation choices in settings where one household's returns depend on community-wide compliance. It reports results from a randomized trial that delivered descriptive norm messages, community engagement, and information to raise perceived prevalence and approval of toilet use. The intervention produced an 8-point gain in toilet access and 6-7-point gain in consistent usage, with mediation analysis attributing a substantial share of these gains to updated empirical and normative expectations. Households also showed higher willingness to pay for living in open-defecation-free areas. These patterns indicate that social expectations function as a distinct channel for changing private behavior when externalities are present.

What carries the argument

The randomized community intervention that combines descriptive norm messaging, engagement activities, and information provision to shift empirical expectations (beliefs about prevalence) and normative expectations (beliefs about approval) of toilet usage.

What would settle it

A replication that uses direct observation or sensor-based measures of usage and finds no difference in consistent toilet use between treatment and control households after belief changes are controlled for would falsify the central claim.

Watch

Extended reading notes

Core claim

The randomized intervention raised toilet access by 8 percentage points and consistent toilet usage by 6 to 7 percentage points relative to control areas. These increases were substantially driven by rises in empirical expectations about the prevalence of toilet usage and normative expectations about its approval. The study further records an increase in willingness to pay for residing in open-defecation-free surroundings, consistent with an elevated perceived value of cleaner sanitation environments.

Load-bearing premise

The measured gains in toilet access and usage are caused by the shifts in social expectations rather than by other features of the community engagement, information provision, or external events during the study period.

Editorial extensions

If this is right

Empirical and normative expectations about toilet use can be altered through targeted messaging and community activities.
These belief shifts translate into higher rates of private toilet adoption and consistent usage.
Households assign greater value to living in areas with lower open defecation after expectations change.
Norm-focused interventions can produce measurable sanitation improvements even when infrastructure is already available.
The effects remain after accounting for local protests and pandemic disruptions.
pith_inferences=[
Similar belief-updating approaches could be tested for other behaviors with community externalities such as mask-wearing or water treatment.
If expectations are the operative channel, lighter-touch messaging versions might achieve comparable results at lower cost.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Desk Editor's Note

read the letter

The paper runs a field RCT in peri-urban Tamil Nadu that combines descriptive norm messages, community meetings, and basic information on sanitation. It finds an 8 percentage point rise in toilet access and 6-7 points in consistent use, with some robustness checks against local protests and COVID timing. Those numbers are the concrete new data point; the rest largely applies existing social-norms framing to this setting. The willingness-to-pay result for cleaner surroundings is a secondary finding that lines up with the main pattern but does not change the picture much. The mediation claim that expectation shifts drive most of the behavior change is the soft spot. The intervention is explicitly a package, so any direct effect from the engagement sessions or the subsidy information could be picked up by the measured belief variables. Without a design that holds the other pieces fixed, the mediation numbers overstate how much the belief channel is doing on its own. The paper is straightforward empirical work with primary data and no obvious circularity in the reporting. It will interest development economists who already follow norms interventions in health or sanitation, but the mechanism evidence is not sharp enough to shift priors on how to design future programs. A serious referee should see it; the design and sample are real, the robustness checks are reported, and the questions about channel separation are fixable with additional analysis or clearer caveats. I would send it out rather than desk-reject.

Referee Report

1 major / 2 minor

Summary. The paper reports results from a randomized controlled trial in peri-urban Tamil Nadu evaluating an intervention that bundles descriptive norm messaging, community engagement, and information provision to shift beliefs about toilet usage. It finds an 8 percentage point increase in toilet access and 6-7 percentage point increase in consistent usage relative to control, with effects robust to local protests and COVID-19 disruptions. Mediation analysis indicates these gains are substantially driven by increases in empirical expectations (perceived prevalence) and normative expectations (perceived approval) of toilet use. The intervention also raises willingness to pay for open-defecation-free surroundings.

Significance. If the attribution to belief shifts holds, the study supplies field-experimental evidence on how social expectations can drive sanitation behavior in a high open-defecation setting, with direct relevance to public-health policy. The randomized design, primary data collection, and checks for external disruptions are strengths that would support the contribution even if channel identification requires further work.

major comments (1)

[Mediation analysis (results section)] Mediation analysis (results section): The headline claim that improvements in access and usage are 'substantially driven' by shifts in empirical and normative expectations rests on mediation through measured belief variables. Because the intervention is explicitly a bundle of norm messaging plus engagement plus information provision, and the design contains no separate arms or instruments to isolate channels, the mediation risks attributing direct effects of engagement (e.g., lower transaction costs) or information (e.g., subsidy details) to the belief measures. This is load-bearing for the central interpretation and requires either additional design features, bounding exercises, or explicit discussion of the limitation.

minor comments (2)

[Abstract] Abstract: The robustness statement mentions protests and COVID but does not indicate which specifications or sample restrictions were used; a one-sentence summary of the checks would improve transparency.
[Tables/figures] Tables/figures: Ensure that all mediation tables report both the reduced-form treatment effects and the mediated proportions with standard errors so readers can assess the precision of the 'substantially driven' claim.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive comment on the mediation analysis. We agree that the bundled intervention design limits causal isolation of channels and will revise the manuscript to explicitly discuss this limitation while tempering interpretive language.

read point-by-point responses

Referee: Mediation analysis (results section): The headline claim that improvements in access and usage are 'substantially driven' by shifts in empirical and normative expectations rests on mediation through measured belief variables. Because the intervention is explicitly a bundle of norm messaging plus engagement plus information provision, and the design contains no separate arms or instruments to isolate channels, the mediation risks attributing direct effects of engagement (e.g., lower transaction costs) or information (e.g., subsidy details) to the belief measures. This is load-bearing for the central interpretation and requires either additional design features, bounding exercises, or explicit discussion of the limitation.

Authors: We agree that the bundled intervention precludes clean separation of channels, and the mediation analysis cannot definitively attribute effects solely to belief shifts versus correlated impacts from engagement or information components. We will revise the results section (and abstract/conclusion) to include an explicit discussion of this limitation, noting that the analysis provides suggestive evidence on the role of expectations but does not isolate mechanisms. Language claiming effects are 'substantially driven' by belief shifts will be tempered to 'consistent with mediation through' or similar. We do not have separate arms or valid instruments for further decomposition, but will add this caveat as requested. revision: yes

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: empirical RCT with standard mediation on primary data

full rationale

This is a randomized field experiment that collects primary survey data on toilet access, usage, empirical expectations, and normative expectations. The central results are intent-to-treat effects (8pp access, 6-7pp usage) plus a mediation decomposition attributing part of those effects to measured belief changes. No equations, fitted parameters, or derivations appear in the reported chain; the mediation uses observed post-treatment variables in a standard Baron-Kenny or similar framework and does not reduce the outcome to the inputs by construction. No load-bearing self-citations or uniqueness theorems are invoked to justify the design or interpretation. The analysis is therefore self-contained against external benchmarks and receives the default non-circularity finding.

Assumptions & free parameters 0 free parameters · 1 assumptions · 0 invented entities

The central claim rests on standard assumptions of randomized experiments and mediation analysis rather than new free parameters or invented entities.

assumptions (1)

domain assumption Randomization balances treatment and control groups on observables and unobservables
Invoked for identifying causal effects of the intervention.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Shifting Beliefs Changing Behavior: Experimental Evidence on Sanitation from India." pith.science (2026). https://pith.science/paper/2312.16205

@misc{pith2026231216205,
  author       = {Pith},
  title        = {Pith review of: Shifting Beliefs Changing Behavior: Experimental Evidence on Sanitation from India},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2312.16205}},
  note         = {Machine review of arXiv:2312.16205}
}

read the original abstract

Open defecation, which is linked to poor health outcomes and lower cognitive ability has been widespread in India. Improved sanitation practice generates local health externalities, which implies that the returns to private toilet usage depend on community wide compliance. Therefore, beliefs about others toilet usage behavior may influence both sanitation choices and the perceived value of residing in areas with lower prevalence of open defecation. This paper evaluates a randomized, intervention in peri-urban Tamil Nadu designed to shift social expectations regarding sanitation practice. The intervention combines descriptive norm messaging, community engagement, and information provision to increase perceived prevalence and approval of toilet usage. The findings indicate a significant increase in toilet access in the intervention areas by 8 percentage points and consistent toilet usage by 6 to 7 percentage points relative to control areas. The effects are robust to potential bias due to disruption due to local socio-political protest and the COVID 19 pandemic during the study period. We also find significant increase in empirical expectations (beliefs about the prevalence of toilet usage) and normative expectations (beliefs about its approval). Mediation analysis indicates that the improvement in toilet access and usage is substantially driven by changes in these social expectations. We further observe significant increase in the willingness to pay among respondents for residing in open defecation free surroundings, suggestive of an increased perceived valuation of cleaner sanitation environments because of the intervention. The findings underscore the need for norm-centric interventions to propel change in beliefs and achieve long-term and sustainable sanitation behavior.

Discussion (0). Sign in to comment.

Pith tools

Reviewed May 24, 2026 · model on record in the stance chip above.