Data scavenger hunt
Introduction to Social Research Methodology
Aim
Today’s session argued that a great deal of the evidence Meridian’s board needs already exists — in official statistics, international databases, and published reports. This exercise tests that claim, and you. Working in pairs, you will locate real secondary data relevant to Meridian’s turnover problem, document each source properly, and then apply the fitness-for-purpose checklist to decide what your finds can — and cannot — tell the board. You will need a laptop or phone with an internet connection.
The scenario, briefly
Meridian is a mid-sized Polish retail-and-services company: about 450 employees, 32 stores, and a growing e-commerce arm. Annual staff turnover has risen from 14% to 27% in two years, and customer satisfaction is falling. The board wants evidence, not hunches. Before proposing any new data collection, your team has been asked to establish what existing data can contribute: how does Meridian compare with the sector, and is its problem unusual?
Task 1: the hunt (25 minutes, pairs)
Locate three sources of secondary data:
- One GUS or Eurostat indicator relevant to employment or wages in the retail sector in Poland — for example, employment levels, average wages, or vacancies in retail trade.
- One OECD or Eurostat indicator on job tenure or labour turnover — how long people stay in jobs, or how much movement there is in the labour market.
- One publicly available industry or consultancy report on employee turnover in retail (Polish or international) — from a retail federation, HR consultancy, job portal, or similar publisher.
For each of the three, record the following in the log below. “We couldn’t find out” is an acceptable entry — but only if you actually looked, and you should note where you looked.
| What to record | Meaning |
|---|---|
| Exact source | Publisher, dataset or report name, and table/indicator code if there is one |
| Years covered | First and last year available; note any gaps |
| Unit of analysis | What one row of the data describes: a country? a sector? a firm? a person? |
| How collected | Survey (which one, what mode?), administrative records, register data, or unknown |
Where to look. Do not use exact links from a search engine’s AI summary — navigate the sources yourself:
- Statistics Poland (GUS) — stat.gov.pl: try the English version, the “Labour market” topic area, or the Local Data Bank (Bank Danych Lokalnych). Retail trade sits in PKD/NACE section G (“Trade; repair of motor vehicles”).
- Eurostat — ec.europa.eu/eurostat: use the data browser and search terms like employment by economic activity, labour force survey, job vacancies, or earnings. Filter to Poland and NACE section G.
- OECD — data-explorer.oecd.org: search for job tenure, employment by job tenure intervals, or labour market statistics.
- For the report in (c), a normal web search works — try combinations like retail employee turnover report Poland, rotacja pracowników handel raport, or the publications pages of large HR consultancies and job portals. It must be publicly accessible (at most requiring a free email registration), and it must be a report with data in it, not a blog post.
Divide the labour: one of you drives the statistical databases, the other hunts the report — then swap to double-check each other’s log entries.
Task 2: the audit (20 minutes, same pairs)
Now apply the session’s fitness-for-purpose checklist to each of your three finds: who collected it, why, when, from whom, how, with what instrument, under what definitions?
For each source, answer in writing:
- Could it help diagnose Meridian’s turnover problem? If yes, state the specific job it could do — establishing context, providing a benchmark, or sharpening the question. Be concrete: “it tells us whether retail wages in Poland rose faster than Meridian’s” is an answer; “it gives useful background” is not.
- What can it NOT tell the board? Every source has at least one hard limit. Check in particular:
- Unit of analysis — does it describe sectors or countries when the board’s question is about stores and teams?
- Definitions — does its definition of “turnover”, “retail”, or “employee” match the way Meridian counts? If you cannot find the definition, say so — and say what that does to the comparison.
- Who and why — for the report in (c) especially: who published it, what do they sell, and would they be embarrassed by any particular result?
- Verdict: for each source, one sentence — use it, use it with stated caveats, or do not use it — with the caveat or reason named.
Task 3: best and most misleading (15 minutes, plenary)
Each pair nominates two finds to present to the class, in one minute each:
- Your best find — the source that would genuinely help the Meridian team, and the specific job it would do.
- Your most misleading find — the source that looks most useful but would mislead the board if used naively. Name the trap: wrong unit of analysis, incompatible definition, self-interested publisher, missing documentation, or something else from today’s session.
As pairs present, note how often the same official sources recur — and how differently the same source can be judged once the checklist has been applied.
Before you leave
Hand in (or photograph and submit) your log from Task 1 and your three verdicts from Task 2. One sentence to finish: name the single most important question the board’s evidence base still cannot answer after your hunt — that unanswered question is where primary research begins, and we will design instruments for exactly such questions when we come to survey design.