The operationalisation ladder: answer key
Introduction to Social Research Methodology
Task 1: Model ladders
There is no single correct ladder; mark for the discipline of the construction, not for matching the models below. A full-credit ladder has: a conceptual definition that distinguishes the concept from its neighbours; 2–3 dimensions that could genuinely move independently; two indicators per dimension; an item (or record extract) per indicator, verbatim, with response options; and a correct level-of-measurement note on every item. The decisive test is traceability in both directions — every item should climb back up to the concept, and no dimension should be left without items (that gap is a content-validity failure, and naming it as such deserves credit).
Model ladder: employee engagement
Conceptual definition. For this study, employee engagement is a positive, fulfilling, work-related state of mind characterised by vigour, dedication, and absorption (Schaufeli and colleagues) — deliberately excluding hours worked, tenure, and intention to stay, which are possible consequences of engagement, not the state itself.
Dimensions, indicators, and items:
| Dimension | Indicator | Draft item / extract | Level |
|---|---|---|---|
| Vigour (energy and resilience at work) | Self-rated energy | “At work, I feel full of energy” — 1 (strongly disagree) to 5 (strongly agree) | Ordinal; treated as interval by convention |
| Absence frequency | Number of absence days in the last quarter, from the HR system | Ratio | |
| Dedication (enthusiasm, pride, significance) | Pride in work | “I am proud of the work I do” — 1 to 5 agreement | Ordinal; treated as interval |
| Sense of meaning | “The work I do is meaningful to me” — 1 to 5 agreement | Ordinal; treated as interval | |
| Absorption (happy immersion in work) | Time distortion | “Time flies when I am working” — 1 to 5 agreement | Ordinal; treated as interval |
| Detachment difficulty | “I find it difficult to detach myself from my work” — 1 to 5 agreement | Ordinal; treated as interval |
Acceptable variation. Students may use a different nominal definition (discretionary effort, say) — fine, provided it is stated and the dimensions follow from it, not from the Schaufeli definition they did not choose. Two questionnaire items per dimension with no record-based indicator is acceptable; a record-based indicator used thoughtfully (absence frequency, with the caveat that it also reflects health and family circumstances) deserves extra credit. Watch for smuggled neighbours: “intends to stay at Meridian” as an engagement indicator contradicts a Schaufeli-style definition and should be challenged — unless the pair defined engagement that way at the top of the ladder, which is exactly the point.
Model ladder: customer loyalty
Conceptual definition. For this study, customer loyalty is a customer’s enduring commitment to continue buying from Meridian, comprising both repeat behaviour and a favourable attitude towards the company — distinct from satisfaction (a met expectation on one occasion), which loyalty may follow from but does not require.
Dimensions, indicators, and items:
| Dimension | Indicator | Draft item / extract | Level |
|---|---|---|---|
| Behavioural loyalty (repeat custom) | Purchase frequency | Number of transactions on the customer’s loyalty card in the last 3 months, from till records | Ratio |
| Share of wallet | “Of your total spending on groceries and household goods, roughly what share do you spend at Meridian?” — percentage 0–100 | Ratio | |
| Attitudinal loyalty (preference and commitment) | First-choice preference | “Meridian would be my first choice for this kind of shopping” — 1 to 5 agreement | Ordinal; treated as interval |
| Switching resistance | “I would keep shopping at Meridian even if a competitor offered slightly lower prices” — 1 to 5 agreement | Ordinal; treated as interval | |
| Advocacy (recommending to others) | Likelihood to recommend | “How likely are you to recommend Meridian to a friend or colleague?” — 0 to 10 | Ordinal; treated as interval (the NPS item) |
| Past recommendation | “In the past six months, have you recommended Meridian to a friend or family member?” — yes / no | Nominal |
Acceptable variation. Two dimensions (behavioural and attitudinal) with advocacy folded into attitude is fine. Self-reported purchase frequency instead of till records is acceptable but weaker — credit pairs who note that records beat recall here (a reliability point from the lecture). The most common fault: defining loyalty as satisfaction, or using satisfaction items as loyalty indicators. The lecture flagged the distinction explicitly (satisfied customers defect; dissatisfied ones stay out of inertia); challenge it in feedback.
Task 2: Sorting solution
| # | Variable | Level | One-line justification |
|---|---|---|---|
| 1 | Store ID (1–32) | Nominal | The numerals are labels of convenience; store 16 is not “twice” store 8, and the “average store ID” is arithmetic nonsense. |
| 2 | Department | Nominal | Categories are simply different; no ordering among sales floor, e-commerce, and head office. |
| 3 | Contract type | Nominal | Same/different is all we can say (and note the exhaustiveness lesson: the B2B category is there because contractors must be classifiable). |
| 4 | Tenure in months | Ratio | True zero (no tenure at all), so 24 months is genuinely twice 12. |
| 5 | Monthly sales in PLN | Ratio | True zero; a store selling 400,000 PLN sold twice as much as one selling 200,000. |
| 6 | Number of absences last quarter | Ratio | A count with a real zero; the full toolkit of arithmetic applies. |
| 7 | Satisfaction rating 1–5 | Ordinal — the classic ambiguous case | Ranked categories with undefined gaps; routinely treated as interval by convention, especially in multi-item averages — credit either answer only if the student flags the convention explicitly. |
| 8 | NPS category (detractor/passive/promoter) | Ordinal | Three ranked categories; also a nice example of downward conversion — collapsed from a 0–10 rating, and the finer information cannot be recovered. |
| 9 | Age bracket | Ordinal | Ordered ranges with unequal, undefined widths (the open-ended “45 and over” gives the game away). |
| 10 | Seniority band (junior/mid/senior) | Ordinal | Ranked, but the “distance” from junior to mid is undefined. |
| 11 | Year the store opened | Interval | Equal intervals (one year is one year anywhere on the scale) but the zero point is an arbitrary anchor — a store opened in 2010 is not “twice as recent” as anything. |
| 12 | Standardised aptitude test score | Interval | A 10-point gap means the same across the scale, but zero does not mean “no ability”, so ratios are meaningless. |
Ambiguities worth staging in discussion. Item 7 is the deliberate trap and the room should divide on it; the right resolution is the lecture’s — strictly ordinal, conventionally treated as interval, and the convention must be adopted knowingly, with the ordinal-robustness check (do medians tell the same story?) when a decision hangs on the result. Item 12 sometimes attracts “ratio” — push back with the zero test. Items 8 and 9 sometimes attract “nominal” — push back with the ordering test (a promoter is more favourable than a passive; 35–44 is older than 25–34).
Task 3: Diagnoses
Draft measure A — a validity problem. Hours worked from payroll is the lecture’s “reliable but not valid” archer’s target made flesh: apply the technique twice and you get the same answer twice, so reliability is excellent — but hours worked reflect contracted schedules, overtime chased for money, and presenteeism born of fear at least as much as engagement. The measure consistently measures the wrong concept. The sting is in the last sentence of the draft: “error rates are near zero, so the measure is fully trustworthy” uses reliability rhetorically as if it certified validity. Reward students who name that move — consistency is necessary for trust, never sufficient.
Draft measure B — a reliability problem. The item is at least aimed at roughly the right concept (how the customer feels about the experience), so validity is not the central disease. The fault is consistency, and it enters twice: the question is so vague that the same customer would answer differently on different days (“things these days” invites answers about the weather, the government, or their knee); and the coding is done “as they judge best” by whichever staff member is on shift, so the recorded value measures the coder as much as the customer — the mutual-exclusiveness lesson about different people making different calls, resurfacing as unreliability. Full credit requires naming reliability as the main disease; a student who also notes secondary validity leakage (the vague wording lets other attitudes contaminate the answer) is reading carefully, but must still commit to the primary diagnosis.
Timing and discussion guidance
- In Task 1, the hardest rung is usually dimensions: pairs either produce synonyms (three names for one aspect) or neighbours (satisfaction as a “dimension” of loyalty). The test to offer them: could the two dimensions plausibly move in opposite directions for the same person? If not, they are one dimension.
- Circulate during Task 1 step 4 and check that every item carries a level note — students happily draft items and forget the levels, which is precisely the design-stage carelessness the lecture warned becomes permanent at the analysis stage.
- In the wrap-up, connect Task 3 back to the ladder: draft measure A went wrong at the operationalisation rung (an operational definition that wandered from the concept), while draft measure B went wrong below it, at the measurement rung. A good measure has to survive every rung.
- If time runs short, cut the wrap-up to five minutes but do not cut the item 7 discussion — the Likert convention is the single point most likely to matter in students’ own projects, and again in sessions 11 and 13.