Who do we ask?

Introduction to Social Research Methodology

Author

Affiliation

Ben Stanley

Department of Social Sciences, SWPS University

Aim

You are Meridian’s junior research team, and the board has approved fieldwork on the turnover problem. Before a single questionnaire goes out, three sets of decisions must be made: the arithmetic of who gets selected, the choice of sampling strategy for each strand of the research, and a workable plan for actually recruiting the people you need. This exercise walks you through all three. Calculators (or phones) are allowed and needed for Task 1; nothing beyond division and multiplication is involved.

Background you will need throughout: Meridian employs 450 people: 330 in the retail stores, 70 in the e-commerce division, and 50 at HQ. Annual staff turnover has risen from 14% to 27% in two years.

Task 1: the arithmetic of selection (15 minutes, pairs)

Work through the three calculations below. Show your working — the numbers are simple, but you must be able to explain why each step is the right one, not just produce a figure.

1a. A proportionate stratified sample

The board wants a survey of current staff, and you have decided on a proportionate stratified design by division, with a total sample size of n = 90 drawn from the HR roster of 450.

1. What sampling fraction does n = 90 from N = 450 imply?
2. How many employees should be sampled from each stratum — retail (330), e-commerce (70), HQ (50)?
3. Check that your three stratum samples sum to 90.
4. In one sentence: what does stratifying by division guarantee that a simple random sample of 90 would not?

1b. What counts as a response?

The survey is fielded as a CAWI questionnaire. You send 300 invitations. When fieldwork closes you have 96 completed questionnaires and 12 partial completions (respondents who answered the first two sections and then stopped).

1. Calculate the completion rate counting completes only.
2. Calculate the rate if partials are also counted as responses.
3. Discuss briefly (2–3 sentences): should the partials count? What would be a defensible rule, and when should that rule have been decided?
4. One sentence: why is the more urgent question not “is this rate high enough?” but “who are the missing 200?”

1c. A systematic alternative

Suppose that instead of stratifying, you drew the 90 from the full alphabetical roster of 450 using systematic sampling.

1. Calculate the sampling interval k.
2. Describe the selection procedure precisely, step by step, including how the starting point is chosen and why that step matters.
3. Name one property of the roster’s ordering you would want to check before trusting this design, and give an example of an ordering that would cause trouble.

Task 2: choose a sampling strategy (20 minutes, pairs)

For each of the four research briefs below, decide on a sampling strategy — name a specific design from today’s session (probability or non-probability) — and justify it in 2–3 sentences. Your justification must address: (i) what population the brief is really about, (ii) whether a usable sampling frame exists for it, and (iii) what the chosen design can and cannot support by way of conclusions. There is more than one defensible answer for some briefs; a well-argued second-best choice beats an unjustified “correct” one.

Brief A — why do leavers leave? The board wants to understand the causes of turnover by hearing directly from people who resigned from Meridian in the past year. HR can find contact details for some, but not all, of them.

Brief B — testing the new rota app. IT has built a shift-scheduling app and wants structured feedback from store staff before rolling it out to all 32 stores next quarter.

Brief C — the courier experience. Meridian’s e-commerce deliveries are made by self-employed gig couriers. They are not on the HR roster, and no one at HQ has a complete list of them. You want to understand their day-to-day working experience in depth.

Brief D — the pulse check. The board meets in ten days and the CEO wants a quick read on current staff morale — “just a temperature check, nothing fancy” — by the end of this week.

Task 3: a recruitment plan for ex-employees (20 minutes, pairs)

Return to Brief A. You are surveying ex-employees who left Meridian in the past year. Selection is only half the problem — these people no longer read the intranet, no longer have company e-mail, and owe Meridian nothing. Draft a one-page recruitment plan with the following four components:

1. Channel. Which recruitment channel(s) will you use to reach leavers, and why? Be specific about how each channel connects to this population, and name one bias each channel introduces.
2. Invitation. Write the actual invitation text you would send — 60–90 words. It should do the four jobs discussed today: who is asking and why this person; topic and honest time estimate; what happens to the answers; what the study is for. Count your words.
3. Incentive. Decide whether to offer an incentive, and if so, which kind. Justify the decision in 2–3 sentences with reference to the trade-offs discussed in the session — including who this incentive will and will not attract.
4. One safeguard against hierarchy pressure. Leavers may fear consequences for references, rehiring, or former colleagues — and current managers have interests in what leavers say. Specify one concrete safeguard in your plan that keeps participation genuinely voluntary and answers frank, and explain in one sentence how it links to the ethical principles from session 4.

Wrap-up (5 minutes, plenary)

Be ready to report: your stratum sizes from 1a, your design choice for the brief your group found hardest, and your invitation text read aloud. As you listen to other groups’ invitations, ask yourself the respondent’s question: would I give these people ten minutes?