Philippines staffing research · Updated
What evidence is missing when exceptions reopen?
Compare first closure records with the later event that returned work to the queue.
Research question: Which observable evidence gaps in an initial exception record are associated with reopening during the registered follow-up window?
Scope: ten weeks of closed exceptions plus a fixed fourteen-day follow-up in one queue, with eligibility, fields, reopening definition, and cutoff handling registered first.
Methodology: Freeze the first closure packet, then link later events by stable ID. Reviewers code evidence without seeing reopening status. Separate true reopenings from duplicate contacts and unrelated new cases.
Unit of analysis: one eligible exception at its first recorded closure. Stable identifiers keep replies, edits, and repeated checks from inflating the denominator.
Classification: remains closed, reopened for missing source, conflicting evidence, incomplete action, owner reversal, customer reply, system correction, duplicate, or unclassifiable.
Measures: closure and reopening denominators, gap prevalence, reopening proportions by gap, time to reopening, correction direction, and loss to follow-up. Report raw counts beside proportions and describe missingness.
Validation: Independently reconstruct sampled histories from raw events, oversampling rare classes without using that sample to estimate prevalence.
Operating boundary: A Philippines-based specialist may collect permitted records and apply the frozen codebook. Named internal owners retain privacy, security, legal, financial, employment, policy, and customer-remedy decisions.
Data handling: Use named accounts, minimum necessary fields, documented retention, and de-identified reporting where practical. Register exclusions before reviewing outcomes.
Inference boundaries: A repeated association can guide a closure-checklist test. It cannot establish that a recorded gap caused reopening.
Limitations: Events outside the queue may be missed, the follow-up can undercount late returns, coders may infer evidence from notes, and complex cases may carry more gaps.
Conclusion: Pilot checklist changes for reproducible gaps and measure a new cohort.