Proxy response rules for outsourced survey calling

Define why the survey selected this respondent

Proxy decisions begin with the survey’s unit of analysis. A customer may be selected because they made a purchase, a staff member because they use a process, or a household because it experienced a service. Another person can be knowledgeable yet still represent a different unit. The calling brief should explain whose experience each question intends to measure.

Set eligibility rules before fieldwork. Identify whether proxies are prohibited, allowed for the whole survey, or allowed only for factual sections. Define acceptable relationships, required knowledge, age or capacity considerations, and any additional consent. A caller cannot construct these rules fairly while someone waits on the line.

Separate gatekeeper assistance from proxy participation. A receptionist can route a call, an interpreter can relay language, and a caregiver can help with communication without becoming the respondent. The record should distinguish assistance, joint response, and substituted response because each has different meaning.

Verify role and permission neutrally

Ask the minimum questions needed to establish the proxy rule. Do not disclose sensitive survey topics to an unverified person just to explain why the selected respondent was called. Use a neutral purpose statement and the approved identity process. If the relationship cannot be established, stop or schedule another attempt according to protocol.

Permission must be appropriate to the study. A colleague saying the participant is busy does not necessarily authorize them to answer. Prior household consent may not cover a new topic. Record who consented, the scope, channel, and time. If the selected person can participate directly later, offer that route without pressuring the proxy.

Avoid coaching eligibility. Reading a list of acceptable relationships can invite someone to choose the answer that opens the survey. Ask open questions where the protocol allows, then apply the rule consistently. Ambiguous cases go to the research owner rather than being resolved by interviewer instinct.

Control which questions a proxy can answer

A proxy may report observable facts but cannot reliably report another person’s feelings, satisfaction, intentions, or private experiences. Tag questions by proxy eligibility and build skip logic that follows the tag. The caller should not paraphrase an ineligible subjective question into something the proxy feels able to answer.

Attribute every accepted answer correctly. The dataset should preserve that a proxy responded, their approved relationship category, whether the selected person was present, and which items used proxy evidence. Do not put the selected respondent’s name beside answers they did not give. This provenance matters for analysis and later quality review.

Use a dignified stop script. The proxy may be trying to help and should not be accused of interfering. Explain that the study requires a different respondent for the remaining questions, thank them for the permitted information, and follow the protocol for callback, partial completion, or final disposition.

Protect neutrality and analysis quality

Interviewers should not signal that proxy answers are less valued or encourage agreement with what they think the selected person would say. Read approved wording, capture uncertainty when allowed, and avoid converting “I do not know” into a negative response. Quality review should listen for subtle leading prompts around eligibility.

Analysis plans need a proxy treatment before data collection ends. Researchers may include, weight, stratify, sensitivity-test, or exclude proxy answers depending on design. The calling team should deliver the required flags and counts, not make analytical decisions from response rates.

Audit proxy incidence by interviewer, question, relationship, sample group, and outcome. High rates can indicate a population need, a list problem, calling-time mismatch, or eligibility coaching. Review recordings or approved transcripts beside the questionnaire version and skip path.

Test the protocol with distinct cases: a spouse who knows factual service dates, an assistant offering opinions for an executive, an interpreter relaying exact answers, a caregiver helping communication, and a colleague who only knows the participant moved. Reviewers should agree on eligibility, consent, allowed questions, attribution, and final disposition for each.

Plan partial interviews and later contact explicitly

If a proxy can answer factual screening items but not the substantive questionnaire, decide whether those facts form a valid partial, a contact update, or no interview. Do not inflate completion by counting routing information as research answers. Save the permitted items with their respondent type and leave unanswered questions genuinely missing rather than filling them from inference.

A callback to the selected participant should not reveal what the proxy said unless the protocol permits it. Provide the neutral survey purpose, approved appointment context, and contact preference. The selected person must remain free to decline even if the proxy was enthusiastic. Record their own consent and responses as a separate respondent event linked to the original sample record.

When a proxy reports that direct participation is impossible, apply the study rule for final disposition. The interviewer should not ask for medical, employment, or family details beyond what eligibility requires. A broad reason category can support fieldwork analysis while respecting privacy. Escalate novel circumstances to the research owner before collecting substantive answers.

Version the proxy rule with the questionnaire. If an eligibility change takes effect during fieldwork, do not silently apply it to earlier interviews or mix response types without an analysis note. The call record should identify the questionnaire and proxy-rule versions used. Retraining must include examples of newly allowed and newly prohibited cases, and quality review should sample the transition period to confirm that callers did not continue using remembered rules.

Report proxy operational outcomes separately from substantive answers. Useful measures include proxy offered, eligible, consented, partial, stopped, converted to direct callback, and unresolved. These counts help researchers diagnose access without treating proxies as failed respondents. They also reveal callers who avoid the eligibility work by marking every gatekeeper unavailable or who overuse proxies to improve completion rates.