Preparing a Public Work Sample for Research Consulting
Choose a work sample that shows your research method and contribution while respecting ownership, publication permission and confidentiality.
A public work sample for research consulting is a short, disclosable account of a single piece of research that sets out the question, what you personally did, the method, what can be attributed to you in public, and where your permission to describe the work stops. It is not a shortened client report and not a second CV. Its purpose is to let a reader who has never met you test two things: how you frame a research problem, and how accurately you describe your own part in addressing it.
The difficulty is that a sample has to pass two tests at once. It must be substantive enough to be assessed, and disclosable enough to circulate without a follow-up conversation with a client, an employer or a data protection officer. Most of the work in preparing one lies in the second test, and that work is best done before the sample is needed.
What a reader is assessing
A reviewer looking at a work sample is usually trying to establish three things, and only the first is about subject knowledge.
Framing. Can you state a researchable question in the form it actually took, including the decision it was meant to inform, rather than the tidier version that appears in a final report?
Method fit. Does the design you chose match the question and the access you had? A methods paragraph that names an approach without explaining case selection, instrument design or analytical logic gives a reader little to work with.
Accuracy about role. Do your verbs survive contact with the other people on the project? This is where samples most often fail, because inflated claims are easy to write and difficult for a reader to challenge.
A five-field contribution statement
The format below is a suggested working structure. It is not a validated instrument, not an institutional requirement, and not a template that any client has endorsed. It is simply a way of keeping the disclosable and the non-disclosable parts of a project distinct.
1. Question. State the question as it was posed at the time, and the decision it was intended to inform if that decision can be described publicly. Where the decision cannot be named, describe the type of decision (a funding renewal, a targeting change, a programme redesign) without identifying the organisation.
2. Personal role. Use task verbs that are unambiguously yours: designed, drafted, conducted, coded, analysed, supervised, advised, reviewed. Avoid team verbs in the first person singular. If a sentence would still be true with another team member's name substituted, it does not describe your contribution.
3. Method. Give enough detail for a methodologist to judge fit: design, sampling or case-selection logic, instruments, analytical approach and the principal limitation. Keep participant-identifying detail out, including combinations of role, location and timing that could re-identify someone in a small study.
4. Attributable output. Point to what exists in public with your name on it, or to what you may describe with permission. A co-authored paper, a published policy brief, a conference note, a methods annex you are allowed to reference. Where nothing is public, say so; an honest absence is more useful than a vague gesture towards a report the reader cannot see.
5. Permission limits. One or two sentences on who owns the material, what you checked, and what you are deliberately not describing. This field does more for your credibility than an additional paragraph of findings, because it shows that you treat disclosure as a decision rather than an oversight.
Laid out under five labels, this runs to roughly half a page. Adapt the headings to your field; the sequence matters more than the wording, because it moves from the research problem to the boundary of what may be said.
A fictional worked example
The following is invented for illustration. It does not describe a real researcher, a real client or a real engagement.
A researcher is preparing a sample for project-based policy and social research. She worked on a mixed-methods evaluation of a regional employment programme. Her first draft read:
Led a mixed-methods evaluation of a regional employment programme, including survey design, interviews and final reporting.
Revised into the five fields:
- Question. How far did the programme's coaching offer change participants' job-search behaviour in its first year, and what would need to change in delivery? The client's decision concerned whether to extend coaching to a second cohort.
- Personal role. Designed the participant interview topic guide and the session observation schedule; drafted the methods section of the interim report. Did not lead the evaluation, conduct fieldwork or write the final report.
- Method. Two-stage design: semi-structured interviews with participants and coaches, sampled for variation in time on the programme, plus observation of coaching sessions; thematic analysis against a framework agreed with the client before fieldwork. Limitations: self-report, single region, no comparison group.
- Attributable output. Nothing from the evaluation itself is publicly attributable. The topic guide may be described, not shared, following written confirmation from the client. A separate co-authored methods note on framework analysis is published and can be linked.
- Permission limits. The client owns the findings and has not agreed to their release. This sample describes design and method only; it contains no participant detail and no client-identifying information.
The revision is longer, and it is the length that makes it useful. The original sentence claims a leadership role the researcher did not hold, compresses design and delivery into one verb, and gives a reader no way to judge the method. The revision leaves a reader with a specific picture of her judgement and a clear statement of what she is not claiming.
Relationships, authorship and third-party material
Attribution and disclosure are easier to settle at the start of a project than to reconstruct afterwards. The UK Government Office for Science's Academic Engagement Toolkit, which addresses engagement between UK policy makers and academics, recommends that written arrangements clarify purpose, outputs, time, roles, attribution and information handling. That guidance concerns UK government engagement and does not establish the terms of any particular project, but the ordering it suggests is sound: roles and information handling are agreed before work begins, not negotiated when a sample is due.
A practical habit follows from this. At project kickoff, record a short disclosure note in your own files: who owns the outputs, which parts of the design you are free to describe later, whether instruments may be shared or only described, and who to ask for permission. Two or three sentences are enough. When a sample is needed months later, you are checking a record rather than testing your memory of an oral agreement.
Deciding what may be disclosed
It helps to think in layers, because restrictions rarely apply uniformly across a project.
| Layer | Typical disclosure position | What to check |
|---|---|---|
| Published outputs | Citable with a link and a role statement | Authorship order and whether your contribution is already described |
| Method and design | Often describable; sometimes commercially sensitive | Contract terms, funder conditions and whether the design itself was confidential |
| Instruments | Shareable, describable only, or withheld | Client permission and whether the instrument contains third-party copyrighted items |
| Findings | Usually the most restricted | Client ownership, embargoes, ethics approvals and participant risk |
| Reconstructed example | Permissible if clearly labelled as illustrative | That no real client, participant or unpublished finding is implied |
For each layer, the relevant conditions are contractual, institutional and ethical rather than stylistic. Employment contracts, funder terms and ethics approvals may each constrain what can be said about a study even in aggregate, and data protection obligations towards participants do not lapse when a project ends. Where a client holds the findings, permission is normally a specific request about a specific sentence, not a general clearance.
If nothing can be disclosed, a sample can still be built from a hypothetical: a short memo setting out how you would approach a stated question, with design, sampling, instruments and limitations. A hypothetical demonstrates reasoning. It does not demonstrate delivery, and it should be labelled so that no reader mistakes it for a completed engagement.
Where samples tend to go wrong
Four failures account for most weak samples. A team claim in the first person singular is the most damaging, because it invites a reader to wonder what else has been rounded up. A method described only by label ("mixed methods", "qualitative research") tells a reader the category of work but nothing about judgement. A sample with no permission line leaves the reader unable to tell whether disclosure was considered at all. And findings smuggled in behind a description of a "de-identified" case can undo the credibility the rest of the sample has built, particularly in small policy fields where the programme, the region and the year together identify the client.
A workable sequence
Write the permission line first. If you cannot state who owns the material and what you are not describing, the sample is not ready, and the missing step is a conversation with the client or your institution rather than more drafting. Then write the method, because it is usually the layer that can be described most fully at least risk. Then write your role, checking each verb against what you would be willing to defend to a co-author. Leave the question until last, when you know what the sample can support.
Before circulating a sample in an expert consulting context, the guidance on confidentiality in expert consulting sets out how information handling is approached, which is a useful check against your own disclosure note.