Sexual Harassment in Public Spaces and Police Patrols: Experimental Evidence from Urban India
CASE STUDY 46
A randomised evaluation of visible and undercover patrols in Hyderabad
Prepared by Digital Catalyzt | Research and product-design draft | 29 October 2026
1. Abstract / Executive Summary
This case study reviews experimental research evaluating a police-patrol intervention at 350 public-space hotspots in Hyderabad. The study reports no effect on overall observed street harassment, while uniformed patrols reduced severe forms such as forceful touching and intimidation by 27% and reduced the likelihood that women left a hotspot because of harassment. The findings show why intervention design and officer conduct matter, while also cautioning against claiming that patrols eliminate harassment.
2. Keywords
Hyderabad; SHE Teams; street harassment; policing; randomised controlled trial; public safety
3. Introduction and Research Question
What did an experimental evaluation find about visible police patrols, and how should digital safety tools complement rather than replace institutional response?
4. Methods and Evidence Selection
Secondary review of the published article abstract and NBER working-paper summary. The underlying study randomised exposure to uniformed and undercover officers across 350 hotspots and used repeated direct observations. This report does not independently reanalyse the trial dataset.
5. Case / Research Context
The research was conducted in collaboration with Hyderabad City Police and evaluated a hotspot-policing programme associated with SHE Teams. Researchers observed harassment and women’s responses at public-space hotspots under different patrol conditions.
6. Findings and Evidence
- The study covered 350 hotspots and compared conditions involving uniformed and undercover officers.
- Visible uniformed patrols reduced severe forms of observed harassment, including forceful touching and intimidation, by 27% according to the reported results.
- The study found no effect on overall street harassment, so the outcome was specific rather than a general elimination of harassment.
- Patrol visibility and officers’ attitudes toward harassment were identified as important mechanisms affecting outcomes.
7. Discussion and Interpretation
The experiment provides stronger causal evidence than a simple before-and-after comparison for the intervention and outcomes studied. Yet the reported effect is not proof that all forms of harassment, locations or time periods respond similarly. A technology layer cannot substitute for trained responders, accessible reporting, accountability or safe public infrastructure.
8. Legal / Policy Context
Police programmes should define the behaviours being targeted, train officers to respond consistently, measure outcomes beyond arrest or complaint counts, and maintain independent oversight. Evaluations should monitor displacement to nearby areas and unintended consequences.
9. Implications for VETRIPEN
Potential product hypotheses: a clearly labelled emergency action; a user-controlled way to share location with chosen contacts; plain-language guidance on available official support; and a transparent status indicator distinguishing message queued, sent and confirmed. Any direct police integration must be formally authorised and tested; do not advertise an unapproved integration.
10. Recommendations and Evaluation Metrics
Measure response-time distributions where data are legitimately available, delivery failures, escalation completion, user experience, complaint follow-through and potential displacement. Any linkage to police data requires legal authority, privacy assessment and independent evaluation.
Recommended implementation safeguards: consent by default; minimal data collection; clear sharing status; no unapproved emergency-service integration; accessibility testing; incident-response procedures; and a documented process for reviewing failures.
11. Limitations
This is a secondary summary, not an independent trial audit. The reported 27% effect applies to specified severe observed behaviours in the study setting, not all harassment or all Indian cities. Results depend on intervention implementation and the study’s observation design.
12. Conclusion
The evidence reviewed here supports a cautious, user-centred approach to women’s safety: make risks and support pathways understandable, preserve user choice, and evaluate actual outcomes rather than assuming that technology or a single intervention guarantees safety. Findings must remain tied to the study populations, definitions and methods cited.
13. References and Source Notes
The links below are the source records consulted for this report. A source link is not a claim that every detail has been independently verified against the complete underlying dataset or full text.
- Quarterly Journal of Economics — published article abstract
- NBER Working Paper 31734
- World Bank Women, Business and the Law research PDF
14. Declarations
Research status: secondary-source synthesis; not a new field study. Funding and conflicts: no external research funding or formal peer review is represented in this document. Product relationship: VETRIPEN implications are design hypotheses for evaluation, not validated intervention outcomes.
15. Disclaimer
DISCLAIMER — INDEPENDENT RESEARCH DOCUMENT. Prepared privately by Digital Catalyzt as part of an independent research and product-design programme related to women’s safety and the proposed VETRIPEN service. For research, education, discussion and preliminary product planning only. Unless expressly stated and documented, this document has not undergone independent peer review and is not a journal publication, government statement, legal opinion, medical or psychological assessment, or formal investigation. Claims are limited to the sources and scope stated; sources may contain limitations, errors, incomplete data or later revisions. Hypothetical examples must not be represented as real events or interviews. Legal and policy discussion is general information, not legal advice. Proposed VETRIPEN features are design hypotheses unless independently evaluated. This document does not guarantee prevention of harm, emergency response, message delivery, confidentiality in every circumstance or any particular outcome. Readers should assess cited evidence, limitations and local context independently.