Sampling · ResearchCall

4. Sampling

11 settings shown · an agent asks 2 · 3 are recorded without effect

Some settings on this station are kept but not yet read by anything. They are marked, because a control that quietly changes nothing is worse than one that is missing.

sample.sourceSampling frame *recorded only

A file or a database. Whatever is missing there is missing from the study.

Nothing reads this value yet: Named for the record; the dry run draws from a generated fictitious frame.

sample.sizeSample size *steers the run

The size is fixed before the draw.

How many records are drawn.

sample.methodSampling methodsteers the run

Stratified only if the sampling frame carries the stratifying attributes.

Random draw or census. Stratified is refused rather than silently drawn at random, because the dry-run frame carries no stratifying attributes.

sample.time_windowsTime windowssteers the run

MorningAfternoonEvening

The windows must be fixed before fieldwork so that nonresponse and answers stay comparable later.

The windows a record can be assigned to.

sample.assign_windows_randomlyAssign time windows at randomsteers the run

Random assignment turns time of day into a controlled variable instead of a silent preselection.

Off assigns windows in equal blocks instead of by lot.

contact_rules.attempts_per_personAttempts per personsteers the run

Follow-up raises the completion yield but biases the sample towards the reachable; the actual number appears in the report.

Extra attempts after no answer, busy or voicemail. The report states the number reached this way.

contact_rules.spread_attemptsSpread attempts over the daysteers the run

Relevant when more than one attempt is allowed; otherwise the same time-of-day bias is repeated.

Sends a repeat attempt into a different time window.

contact_rules.callback_after_refusal_maxAgreed callbackssteers the run

A no without an explicit callback agreement is final.

Upper bound for people who refused but accepted a later call.

contact_rules.daily_quotaDaily quotasteers the run

Recurrence is triggered from outside; the tool does not run itself across days.

Records worked per window in one run.

contact_rules.concurrent_callsConcurrent callsrecorded only

Conservative default: the concurrency limit of the service is unverified.

Nothing reads this value yet: The dry run is serial by design; the API's concurrency limit is unverified.

contact_rules.calling_hoursCalling hoursrecorded only

The range bounds the externally started daily runs.

Nothing reads this value yet: Needs a scheduler with a clock; this tool has no daemon and no timer.