Diagnosing and correcting household waste-segregation failure in Ward K/E, Mumbai

Introduction

1.1 Background and context

Ward K/E generates an estimated 42 tonnes of household waste per day. Municipal policy mandates wet and dry segregation at source, and awareness campaigns have run for years, yet mixed waste remains the norm and most dry recyclables reach landfill.

India's Solid Waste Management Rules, 2016 place segregation at source at the centre of municipal waste policy, making the household the first and most important sorting point in the entire system. When segregation fails at the home, every downstream stage, collection, recovery, composting and recycling, inherits contaminated, mixed waste that is expensive or impossible to recover. The cost of that first failure is therefore large and compounding.

1.2 The knowledge and action gap

Municipal and civil-society responses to low segregation have overwhelmingly taken the form of awareness campaigns, on the implicit assumption that residents do not segregate because they do not know how. Yet segregation rates across many Indian cities have remained stubbornly low despite sustained awareness effort. This pattern suggests a gap between knowledge and action that information alone does not close. This study was designed to test, in one ward and with primary data, whether the binding constraint on segregation is informational or structural.

1.3 Research questions and objectives

1. What actually determines whether a household in Ward K/E segregates its waste, knowledge, means, or motivation?
2. Does a structural intervention that supplies the missing means and a feedback loop outperform the awareness-first approach?
3. Is such an intervention affordable and plausibly scalable at ward level?

The work was conducted over 22 weeks under FIELDWORK's four-phase method, with each phase independently assessed before the next could begin.

Study area and methods

2.1 Study area and sampling

Ward K/E spans high-rise apartment complexes, chawls and low-rise housing, and a small share of informal settlements. To avoid a convenience sample, 140 households were surveyed under a stratified design reflecting the ward's housing mix (Table 2). Stratification mattered because barriers to segregation differ sharply by housing type, and an unstratified sample would have over-represented high-rises and understated the problem elsewhere.

2.2 Instruments and data collection

The survey was itself piloted with eight households and revised before full administration, to remove leading questions and ensure the barrier options reflected residents' own language rather than the researcher's assumptions.

2.3 Measurement and analysis

Segregation compliance was defined as a household correctly separating wet and dry waste on inspection. It was measured by direct bin inspection supported by a daily tracking tool, not by self-report alone, at baseline, during the pilot, and weekly across deployment. A subsample of inspections was double-checked by a second observer to guard against single-observer bias. Site-level figures were combined as household-weighted averages. Given the applied, single-ward design, results are reported as effect sizes and directional change rather than inferential tests.

2.4 Ethics, consent and data handling

Participation was voluntary and based on informed consent. Households were told the purpose of the study, that they could decline or withdraw at any time, and that their data would be anonymised. No personally identifying information is retained in the dataset; households are referenced by anonymous codes. Only adult respondents were surveyed. Building committees approved the work in each participating society before deployment.

2.5 Study timeline

Baseline results

3.1 Compliance and the knowing and doing gap

Baseline segregation compliance was 28%. Critically, 78% of surveyed households correctly described what belongs in wet versus dry waste. A 50-point gap between knowing and doing (Table 3) is the central empirical finding of the baseline and the fact that reframed the entire study.

3.2 Waste composition

3.3 Reported barriers

3.4 Variation by housing type

3.5 Knowledge versus behaviour

Root-cause analysis

The obvious response, more awareness campaigns, was rejected across three rounds of mentor review because it did not fit the data. Knowledge was high; behaviour was low; and the barriers households actually reported were structural. The two leading constraints, no separate bins at home and no feedback that segregation mattered, are both about means and motivation rather than information. Visible collection-side mixing compounded the second: residents who did segregate saw their effort undone at the truck and concluded, rationally, that it was pointless.

The research problem was therefore reframed: segregation fails because the home lacks the means (bins) and the motivation loop (feedback), not the knowledge. This reframing defined the design brief for the intervention and set the criteria against which it would be judged.

Intervention design

5.1 Design principles

Three principles followed directly from the root-cause analysis: remove the physical barrier, close the feedback loop, and keep unit cost low enough to be scalable. Each element of the intervention maps to one of these.

• Means. A subsidised two-bin kit placed in each participating home, directly removing the leading barrier.
• Signal. A lightweight tracking tool logging daily household and building-level compliance, built with an AI stack and applied data science.
• Loop. Weekly feedback to each building showing its own segregation trend, plus visible confirmation that segregated waste was collected separately, countering the "it gets mixed anyway" barrier.

5.2 The tracking tool

The tool recorded daily compliance at household level from bin inspections logged by building volunteers, aggregated it to a single weekly building score, and generated the feedback message automatically. The applied data-science component handled noisy, incomplete daily logs and smoothed them into a trend the building could act on, so a single missed inspection did not distort the signal.

5.3 Pilot and iteration

The tool passed through three iterations during the pilot. Version one was too complex for daily volunteer use; version two logged data but returned no feedback and changed nothing; version three, which surfaced a single weekly building score, was the version households engaged with (Table 4).

5.4 Unit cost

Deployment and results

6.1 Design and headline results

The validated intervention was deployed across four housing societies chosen to represent the ward's housing mix, reaching 615 households over six weeks. Compliance was measured weekly by bin inspection and the tracking tool.

6.2 Results by society

6.3 Weekly trajectory

6.4 Participation and retention

6.5 Downstream effects

Correctly segregated dry waste reaching the recycling stream rose 2.4 times over the six weeks. Contamination of the wet-waste stream, which blocks composting, fell from 44% to 19% of sampled loads. The intervention therefore improved not only the headline segregation rate but the usability of both output streams, which is where the environmental value is realised.

Cost-effectiveness and scalability

At roughly ₹250 per household and a 43-point average gain, the implied cost of moving one additional household into reliable segregation is approximately ₹580. The largest cost item, the bin kit, is a one-time expense, so the marginal cost of sustaining compliance in later periods is far lower than the first-year figure.

Two factors support scalability: the bin kit cost falls with bulk procurement, and the tracking tool has near-zero marginal cost per additional building. The principal constraint on scale is not money but the volunteer capacity to run weekly inspections, which the next section addresses.

Stakeholder engagement

Segregation is a system that spans several actors, and the intervention only worked because each was engaged deliberately rather than assumed.

• Residents. The primary actors. Engaged through the bin kit and, crucially, the weekly feedback that made their effort visible and worthwhile.
• Building committees. Approved the work, nominated the volunteers who logged daily inspections, and owned the weekly score. Their buy-in was the single biggest predictor of a society's retention rate.
• Municipal collection workers. Walked the routes with the fellow during mapping and confirmed separate collection of segregated waste, which directly countered the most corrosive barrier, the belief that it all gets mixed anyway.

The society with the weakest committee engagement (Society B) recorded both the lowest gain and the lowest retention, underlining that the human system, not the tool, was the limiting factor.

Discussion

The results support the reframed hypothesis. Where awareness was already high, adding knowledge would not have moved behaviour; supplying the missing means and closing the feedback loop did. The pilot makes the mechanism explicit: version two, which logged data but returned no feedback, barely moved compliance, while version three, which returned a weekly score, produced the jump. Measurement is not motivation; the returned signal is.

The timing in the full deployment tells the same story. The largest weekly gains coincided with the start of building-level feedback in week two rather than with the distribution of bins alone, suggesting the motivation loop, not the bins by themselves, drove sustained adoption.

Variation between sites is also instructive. Society B, where occasional collection-side mixing persisted and committee engagement was weakest, gained least. This points to segregation as a system property spanning household, building and collection, not a household behaviour in isolation. It also implies that interventions targeting only the household, as awareness campaigns do, are structurally incomplete.

Limitations

• Durability. Six weeks demonstrates adoption, not permanence. Retention at 6 and 12 months is untested and flagged for follow-up.
• Measurement. Compliance combines bin inspection with tool logs; both may overstate, so figures are treated as directional, not exact.
• Generalisability. One ward and four societies. Replication is required before any wider claim.
• System boundary. Persistent collection-side mixing in some sites likely capped the achievable ceiling.
• Volunteer dependence. The feedback loop relies on volunteer inspections, whose availability may not hold at larger scale.

Policy implications

If the finding holds under replication, it carries a direct message for municipal practice: the marginal rupee spent on further awareness campaigns is likely to underperform the same rupee spent on means and feedback. Specifically:

1. Pair any segregation mandate with the physical means to comply, subsidised bins, rather than information alone.
2. Protect segregation integrity at collection; visible mixing at the truck destroys household motivation faster than any campaign can build it.
3. Give buildings a simple, regular feedback signal on their own performance, which is cheap to produce and appears to be the active ingredient.

Conclusion

Household waste segregation in Ward K/E failed for structural reasons, not for lack of knowledge. A low-cost bin-and-feedback intervention raised weighted compliance from 28% to 71% across 615 households in six weeks, retained 90% of participants, and materially increased recovery of dry waste, at roughly ₹250 per household. The finding is bounded by its duration and single-ward scope, and the honest next step is a longer, multi-ward replication that tests whether the gains persist and whether the volunteer-dependent feedback loop holds at scale.

Reflection and learning outcomes

Beyond its result, the project was a structured learning experience. The most important lesson was epistemic: the first, obvious explanation, that people needed more awareness, was wrong, and only primary field data revealed it. Learning to reject a comfortable hypothesis on evidence is a transferable research skill.

• Inquiry and evidence. Designing an instrument, gathering primary data, and letting it overturn an assumption.
• Real-world engagement. Working with residents, building committees and municipal workers, and learning that the human system determines whether a technical solution survives.
• Perseverance. Two failed tool versions before one worked, and the discipline to document the failures rather than hide them.
• Communication. Turning noisy field data into a signal a building committee could act on.

These map onto the inquiry, action and reflection expected of extended, self-directed student research, and the record above is the audit trail of that process rather than a summary written after the fact.

Data and evidence availability

• Baseline survey dataset (140 households, 18 fields)CSV
• Waste-audit weights and collection-route mapPDF · 12 pp
• Root-cause analysis with mentor review notes (3 rounds)PDF · 9 pp
• Segregation-tracking tool (v3) and pilot resultsRepository
• Six-week deployment log, per-society weekly measurementsCSV · 4 sites
• Final field report with full limitationsPDF · 24 pp

On a fellow's live record, each item links to the underlying file so the data can be inspected directly.

Independent assessment and verification

Each of the four phases was assessed by an independent reviewer against a fixed rubric before the fellow could proceed. Nothing on this record is self-reported; every phase was gated.

Record ID: FW-2026A-0147   Status: Verified   Rubric: FIELDWORK four-phase v1
Permanent link: fieldwork.earth5r.org/fellows/2026a/0147 (scan the code above to verify)
This is a representative FIELDWORK record. On a fellow's live record, each figure and evidence item links to its underlying data so any university can verify it directly.

Acknowledgements

This work was carried out with Earth5R and its field programme, under FIELDWORK's mentored track. The fellow thanks the participating housing societies and their committees, the municipal collection workers who gave their time during route mapping, and the mentor and independent assessors who held the work to a standard at every phase.

References

1. Ministry of Environment, Forest and Climate Change, Government of India. Solid Waste Management Rules, 2016.
2. Municipal Corporation of Greater Mumbai. Solid Waste Management Department, ward-level waste generation data.
3. Central Pollution Control Board. Annual Report on Implementation of Solid Waste Management Rules.
4. Swachh Bharat Mission (Urban). Guidelines on source segregation and municipal solid waste management.
5. Earth5R. Community waste-management methodology and field programme documentation (primary source).

Appendix A: Survey instrument

The 18-item structured survey administered to 140 households, grouped by theme.