Philippines staffing research
How should partial survey completions be reviewed?
A cautious framework for checking whether partial interviews differ from completed surveys before analysts discard them.
Published: September 2, 2026. Research question: what partial completions reveal about questionnaire position, respondent choice, and the limits of a completed-only analysis? This article describes an operational review method for OutsourcedCallers.com readers. It does not report a measured result for OutsourcedCallers.com or for Filipino callers.
Define the review unit before reading outcomes: one sampled contact, consent event, answered items, stop point, stop wording, call duration, and final disposition. Keep incomplete and disputed records when they belong to the released sample. Removing inconvenient cases after review would change the study population.
Code the difference between a completed interview, a respondent-led stop, a technical interruption, and an interviewer-led termination. Write a definition and one counterexample for each code. Reviewers should use unknown when the record does not support a decision.
Collect only the fields needed for the question: sample source, contact outcome, item order, responses, last completed item, stop code, duration, and any approved retry. Preserve timestamps and the version of the queue, script, or scorecard that existed when the event occurred.
Separate what the caller could observe from later owner action. The Filipino caller owns accurate capture and the released handoff. The survey method owner owns policy, authoritative system changes, and the interpretation used for a business decision.
Use independent review for a small sample before scaling the coding work. When reviewers differ, require both to cite the record field or call moment behind the code. Adjudication should preserve the first ratings so disagreement remains measurable.
Report the numerator and denominator beside any rate. Also show missingness by reason. A neat percentage can hide records that lacked audio, timestamps, owner decisions, or a reliable link between systems.
Test whether the result changes under a second reasonable definition. If a narrow and broad rule lead to different conclusions, publish both and explain the operational choice instead of selecting the tidier number.
The main limitation is that a partial record cannot supply the answers that the respondent did not give. Treat association as a prompt for process review, not proof that one caller action caused the later result.
A practical close gives the survey method owner three items: the field definition that failed most often, a short set of example records, and the next bounded review. The owner can then change the source, route, or instruction and test the same measure again.
Sources: https://www.nist.gov/privacy-framework | https://www.nist.gov/cyberframework | https://www.ftc.gov/legal-library/browse/rules/telemarketing-sales-rule | https://www.ftc.gov/business-guidance/advertising-marketing/telemarketing | https://www.ftc.gov/business-guidance/privacy-security/data-security | https://www.fcc.gov/general/telemarketing-and-telephone-consumer-protection-act | https://www.donotcall.gov/ | https://privacy.gov.ph/data-privacy-act/ | https://aapor.org/standards-and-ethics/best-practices/
Method note: Retain exclusions, unknown values, reviewer disagreements, and version dates. Do not publish personal contact details or infer protected or sensitive traits from a call record. FAQ: What is the unit? one sampled contact, consent event, answered items, stop point, stop wording, call duration, and final disposition. Who owns interpretation? The survey method owner. What happens when evidence conflicts? Preserve both codes and send the case to adjudication.