Population, unit of observation, and responsible contact
6 min read
Lesson objectives
- Specify an eligibility rule and contact frame aligned with the question while avoiding misleading recruitment and excess data collection.
Please sign in to use all the tools
Once you sign in, the reading tools are yours: larger or smaller text, typeface, line spacing, focus mode, highlighting and bold, and notes saved to your account and gathered in My study.
Who does the result describe?
'The store's customers' is broader than most available datasets. Before drafting the first survey field, define the population to which the decision applies: perhaps people who completed a first order in April, live within a specified service area, and have had 30 days in which to return. Then define the unit of observation: person, account, or order? One person may use two accounts, and a household may share one. If the decision concerns individual experience, a single order is not automatically an independent person. An inclusion rule says who enters the study: a completed order, no staff test account or cancellation, and sufficient follow-up time. Give the reason for each exclusion because it changes the scope of inference. Clear units and population make a denominator interpretable rather than a number without a source.
A contact frame is the list or channel through which this population can be invited. A store email file misses customers who declined messages; an in-app notice misses people who deleted the app; door interviews do not automatically describe online buyers. Map the coverage gap: which eligible people cannot be reached? Do not treat everyone who received an invitation as a respondent. Nonresponse may relate to the experience itself: unhappy customers might answer more readily, so mean satisfaction among respondents need not describe everyone. Record eligible, invited, and responding counts under explicit definitions, and compare available characteristics without claiming bias is fully corrected. For a probability sample, record selection chances and frame. For an opt-in sample, disclose the voluntary method and avoid an unsupported probability margin of error.
Participant contact is part of both research quality and ethics. Explain who conducts the study, its purpose in plain language, the data requested, approximate time, how answers will be used, and who can see them. Make participation and withdrawal clear; do not imply service depends on replying when it does not. Do not collect an identity number or precise address if a broad area is enough; each extra field needs a purpose and retention boundary. Keep research separate from promotion, and do not send an individual's answers to a service employee who will make decisions about them without a declared basis. Professional principles inform responsible design, but a real project still requires checking applicable local privacy rules rather than treating an international code as local law.
Run a consistency check before sending. Will the question reach someone who actually saw the policy? Will the result describe accounts or people? Can staff accounts and duplicates be removed from the frame? Can the shown version be logged so that a participant is not asked about wording they never saw? If two languages are compared, do the question and options carry the same meaning rather than merely translated words? When coverage is poor, change the inference or recruitment, not the sample description after a favorable result appears. 'Among 90 respondents invited by email from eligible accounts' is honest; '90 representative a fictional city users' is unsupported by these counts alone.
Lesson summary
Specify an eligibility rule and contact frame aligned with the question while avoiding misleading recruitment and excess data collection.