This article proposes an operational study of how housing customer-service teams recognize and route potential accommodation inquiries. It reports no measured rates, resident outcomes, or legal conclusions. The study would ask whether a resident's request reaches a responsible owner, whether acknowledgment and next steps are documented, whether maintenance and accommodation work remain connected, and whether frontline staff collect more sensitive information than routing requires.
The unit would be one inquiry beginning with the first observed contact about a change, exception, assistance, or adjustment connected to a stated disability-related need. The resident would not need to use formal legal language. Later calls, messages, and management notes about the same request would remain attached to that first case so repeat contact can be measured without turning one inquiry into several observations.
The codebook would distinguish potential accommodation inquiries, maintenance-only requests, general service complaints, communication-access requests, parking or physical-access requests, policy exceptions, service or assistance animal inquiries, and insufficient-information contacts. A code would describe the observed request. It would not decide that the request must be granted or that the resident has established a legal entitlement.
The Department of Justice overview of the Fair Housing Act provides primary federal context for the study boundary. The protocol would use it to explain why a recognition and routing process matters. It would not ask service agents or research reviewers to make legal findings from a ticket.
Before extraction, the team would set the study window, source systems, inclusion and exclusion rules, sampling method, and final outcome cutoff. Cases still open at cutoff would remain unresolved. Silence after the last message would not be treated as evidence that the resident withdrew, received a solution, or pursued another channel. Tests, spam, and cases with no recoverable history would appear in an exclusion table.
Sampling should include routine inquiries and less common categories deliberately. A purely random sample may contain many maintenance contacts and too few communication-access, parking, animal, or policy-exception requests to test routing. The report would state the random component, any oversample, and the weight or limitation created by that design.
The chronology would capture the requested change, barrier described, property or policy involved, contact channel, initial classification, acknowledgment, transfers, explicit owner acceptance, requests for information, maintenance actions, accommodation-review actions, resident updates, and the last documented next step. Each event would retain a source and time. A later outcome would not overwrite what the first agent could see.
Reviewers would extract only information needed for the routing question. Names, full addresses, diagnoses, clinical attachments, credentials, unrelated household details, and unrestricted free text should not enter the analytic dataset. A coded summary could describe the request while an authorized pointer permits validation against the protected source.
The study would code whether support requested medical or other sensitive detail before the designated owner accepted the case. It would also record whether the approved workflow required that information at that stage. This is a review of process and data minimization, not a conclusion about what evidence may ultimately be appropriate for an individual request.
Acknowledgment would mean that the organization confirmed receipt and explained what would happen next. It would not mean approval. Owner acceptance would require an explicit event showing that a responsible person or team took the case or performed the defined next action. A transfer into a queue with no acceptance evidence would remain unowned.
24 CFR Part 100 supplies primary regulatory text concerning discriminatory housing practices. The study would use it as legal context and a boundary on interpretation, not as a frontline checklist for deciding individual cases. An organization applying this protocol should have qualified legal or compliance owners approve its operational definitions.
The HUD and DOJ Joint Statement on Reasonable Accommodations discusses accommodation requests and the interactive process. The proposed coding scheme would use that source to distinguish receipt, procedural exchange, and a documented decision. It would not convert an incomplete service history into a legal conclusion about whether every obligation was satisfied.
Maintenance and accommodation work can coexist. A resident may report a broken entrance and also request a change connected to access. The maintenance team can own the repair while an accommodation coordinator owns the requested adjustment. The chronology would show both paths. Repair completion would not automatically close the inquiry, and accommodation review should not delay urgent repair.
The method would separately flag reports of immediate danger, loss of essential service, inability to enter or leave, or another urgent condition. Those cases should follow the organization's emergency or habitability route. The research code would describe the signal and routing action without deciding legal severity or claiming that routing caused or prevented harm.
Assistance-animal inquiries may combine an accommodation request, documentation question, property-rule issue, and ordinary animal complaint. Reviewers would code the request made and the owner selected. They would not infer credibility from animal type, appearance, or wording. The analysis would focus on recognition, sensitive-information collection, ownership, and next steps.
Communication-access inquiries would receive a separate code when a resident requests interpretation, an alternate format, a relay method, or another communication change. The chronology would record the method requested, temporary method offered, and responsible owner. A generic language-service transfer would not automatically complete the accommodation question.
Outcome categories would include accepted for review, information requested through an approved process, alternate operational solution accepted, qualified-owner decision documented, resident withdrawal supported by evidence, unresolved at cutoff, and unknown. A note saying "sent to management" without a named owner or timing would not count as a documented next step.
HUD's housing discrimination complaint guidance describes a public reporting channel. The study may use this primary source to review whether internal scripts accurately describe public information. It would not treat a filed complaint, or the absence of one, as evidence that internal handling was correct.
Two trained reviewers would independently code a declared subset for eligibility, request type, sensitive-data collection, first route, owner acceptance, urgent signal, maintenance overlap, and outcome. Agreement would be reported for each field. Both original labels would remain available after adjudication so analysts can distinguish hard cases from unstable rules.
Repeated disagreement between maintenance and accommodation codes would trigger codebook revision before full review continued. If explicit owner acceptance cannot be found reliably in the systems, that absence would be reported as a limitation and possible operating gap. Reviewers would not infer ownership from a later vague note merely to complete a timeline.
The primary table would report eligible inquiries by request type and final process state with denominators, exclusions, unresolved cases, and unknowns. Supporting tables could show acknowledgment time, time to owner acceptance, rerouting, repeat contact, maintenance overlap, information requests, and missed promised checkpoints.
Timing would be shown as a distribution, with open cases visible at cutoff. Small groups would not be presented with false precision. Sensitivity analysis would test whether findings change when ownership requires an explicit acceptance event, cases missing event times are excluded, oversampled categories are removed, or the outcome window is extended.
The analysis should also compare the initial route with the adjudicated request category. That table would show whether frontline recognition problems cluster around particular wording or channels without assuming that the reviewer classification is a legal judgment. Cases reclassified only after additional resident information should be separated from cases where the original message already contained the relevant request. This distinction helps an operations team decide whether it needs a clearer intake rule, a better form, or faster access to a qualified owner.
Case-level review should examine promises made while ownership was unresolved. A resident may receive a response deadline, a maintenance visit, or a request for documents before an accommodation owner accepts the inquiry. The protocol would record whether each promise was kept and whether it addressed the actual request. A quick acknowledgment followed by repeated rerouting should remain visible rather than being summarized as timely service.
The report should discuss system limitations directly. If timestamps are overwritten, queue acceptance is not logged, or maintenance and accommodation records cannot be joined reliably, those gaps restrict the conclusions. Analysts should provide missingness by field and source. They should not fill chronology gaps from narrative tone or assume that a completed status means the resident received the documented next step.
The protocol can locate recognition failures, unowned transfers, overcollection, and missing next steps. It cannot determine legal merit, prove that staffing caused an outcome, estimate prevalence outside the observed records, or establish that one workflow is universally compliant.
The study package would retain the protocol, source inventory, privacy approval, extraction logic, sample construction, codebook, reviewer training, original and adjudicated labels, exclusion log, denominators, missing-data counts, sensitivity outputs, analysis code, and dated source checks. A second analyst should be able to rebuild every table.
Any conclusion should remain operational: where inquiries entered, whether the request stayed intact, whether a responsible owner accepted it, which next steps were documented, and what remained unresolved. Until authorized records are reviewed under this protocol, this article presents a proposed method only and makes no empirical claim.