---
name: incentive-and-participation-design
description: >
  Sets the incentive level, structure and timing for a study, and states what that
  choice does to who takes part and what they say. Use when someone says "what
  incentive should we offer", "how much should we pay respondents", "prize draw or
  fixed payment", "can we incentivise doctors", "our completion rate is too low",
  "is this incentive too high", or asks how to describe the incentive in the
  methodology.
category: 03 Fieldwork and Data Collection
ref: 03.05
tier: 2
inherits: [K2, K3, K4, K5]
---

# Incentive and Participation Design

## 1. One-line description
Sets what participants receive, in what form and when, matched to the burden the study imposes and the population it asks, and makes explicit the composition and response effects that follow from getting it wrong in either direction.

## 2. What this skill is used for

**The research problem it solves.** Incentive is usually treated as a line in a budget rather than a design parameter, and it is decided by precedent: what was paid last time, what the finance system can process, or what a supplier quoted. That is a mistake with a specific and asymmetric shape. Under-incentivise and you do not get a smaller version of the same sample. You get a differently composed one: participation concentrates in people who have their own reason to take part, which means the highly interested, the strongly opinionated at either extreme, and those with time to spare. Over-incentivise and the composition shifts the other way, toward people for whom the payment is the point, which raises misqualification, reduces effort, and in some populations raises the ethical question of whether consent was freely given. Neither failure appears in the dataset as an incentive problem. Both appear as findings.

**Where it sits in the research lifecycle.** After sourcing, because the appropriate incentive depends on where the sample comes from, and after the instrument, because the appropriate incentive depends on the burden. Before fieldwork launches, and revisited at soft launch, where completion and drop-off will show whether the level was right.

**Typical use cases.**
- Setting the incentive for a new study where no precedent applies.
- Deciding between a fixed payment and a prize draw, and being able to say why.
- Designing incentives for a long, intrusive or multi-stage commitment (diaries, ethnography, longitudinal panels).
- Setting incentives for professional, clinical and B2B audiences, including where payment is restricted or prohibited.
- Diagnosing a completion or drop-off problem that may be an incentive problem.
- Designing non-monetary participation value, including results-sharing, where money is unavailable or inappropriate.
- Assessing whether an incentive is high enough to constitute undue inducement in a low-income or dependent population.

**Who uses it.** Research managers and directors specifying fieldwork; client-side insight managers approving budgets and needing to know what they are buying; academic and public sector researchers writing incentive justifications for ethics committees; UX and product researchers recruiting their own users.

## 3. When to use it

- A study is being specified and the incentive has to be decided rather than inherited.
- The burden is unusual: long, repeated, intrusive, requiring travel, equipment, record-keeping or the participation of others.
- The population is one where the standard approach does not apply: professionals, clinicians, senior executives, children, or people in a dependent relationship with the sponsor.
- Completion or drop-off in field suggests the offer is not matched to what is being asked.
- The sample source contains people who take part in research often, so the incentive interacts with misqualification risk.
- The study will pass an ethics committee, an institutional review, or a client's own governance, all of which will ask what participants receive and why.
- The study runs across markets where the same nominal value means very different things.

## 4. When NOT to use it

- **The incentive is being used to influence what respondents say, rather than whether they take part.** A payment conditional on a particular answer, a completed purchase, a positive review, or continued eligibility that depends on the answer given, is not an incentive. It is a purchase of data, and the resulting responses cannot be interpreted. This is not a design problem to be optimised; it is a study that should not run.
- **The population is vulnerable, dependent, or in circumstances where the payment could reasonably override their judgement about participating.** People in serious financial hardship, people whose access to a service depends on the sponsor, patients recruited by their clinician, employees by their employer, or anyone for whom the amount is large relative to their means. The question is no longer methodological. Route to **13.05 Research Ethics and Consent Design**, which owns undue inducement, before the level is set.
- **The real problem is the instrument, the sample source or the invitation.** A study with a 40% break-off at question 14 does not have an incentive problem. Raising the incentive to compensate for a defective instrument buys completions from people who are pushing through something they dislike, which produces worse data at higher cost. Diagnose first with **03.04 Fieldwork Monitoring and Response Quality**.
- **The offer would be illegal, prohibited or a compliance risk in this population or jurisdiction.** Payments to public officials, healthcare professionals in some systems, employees of organisations with gift policies, or anything falling under anti-bribery, transfer-of-value or professional conduct rules. The constraint is not negotiable and the design has to work within it, which usually means non-monetary participation value or a charitable route.
- **The study is already in field and the proposed fix is to change the incentive mid-stream.** Respondents recruited under different offers are not exchangeable, and the change partitions the sample in a way that correlates with motivation. If it must be done, it is a logged amendment handled by **03.04**, not an operational tweak.
- **The incentive will not be disclosed in the methodology.** Level, structure and timing all bear on interpretation, particularly on stated intent, satisfaction and advocacy measures. An undisclosed incentive is a hidden variable, and a study that will not report it is one whose composition cannot be assessed.
- **The task is deciding where the sample comes from.** Source selection belongs to **03.01 Recruitment and Sample Sourcing**. Incentive interacts with source and does not substitute for it: no incentive level rescues a source that cannot reach the population.

## 5. Required inputs

**Required. Without these the skill cannot run.**
- **The full burden of participation**, not just the interview length: total time including any screening and set-up, number of contacts, effort type (recall, record-keeping, installing something, being observed), intrusiveness, travel, and anything the participant must do outside the session.
- **The population and its relationship to the topic and the sponsor.** Determines what motivates participation beyond payment, and whether payment is appropriate at all.
- **The sample source.** Incentive expectations, and the risk of incentive-driven misqualification, differ sharply by source.
- **Constraints**: budget, what can actually be paid and how, and any professional, legal or organisational restriction on payment to this population. Without knowing what is payable, the design is theoretical.

**Optional, and what each one adds.**
- **Comparable studies' incentive levels and their achieved completion rates.** Turn the level from an assertion into a calibration. Only useful where burden, population and source match, and that match should be checked.
- **Ethics committee requirements or institutional policy.** Often prescribe both a ceiling and a justification format, which changes the output rather than the reasoning.
- **Local wage or cost data for the population.** Lets the level be expressed as compensation for time on a defensible basis rather than as a round number, which is the strongest justification available to an ethics committee.
- **Previous wave's incentive.** In a tracker, a change in incentive is a change in the composition of the responding sample, and therefore a candidate explanation for any movement in the results.
- **What participants would value that is not money**: early access, results, influence, recognition, professional development credit. Often more effective than cash in professional populations, and sometimes the only permissible option.
- **Payment mechanics and timing available.** Reliability and speed of payment affect completion and future participation more than most researchers expect.

## 6. Questions to ask before starting

1. **What is the total burden, in time and in effort, from first contact to final obligation?** Almost every under-incentivised study underestimated this, usually by ignoring screening, set-up and between-session tasks. Default if unanswered: sum the design's stated components, add the screening and set-up time, and state the total as an estimate.
2. **What is this population's realistic alternative use of that time, and what does money mean to them?** The same amount is trivial to one population, meaningful to another, and potentially coercive to a third. Default: express the incentive as compensation for time at a defensible rate for that population, and check the result against the ethics test in Step 8.
3. **Is payment permitted for this population at all?** Some professions, employers and jurisdictions restrict or prohibit it. Default: assume a restriction exists for clinicians, public officials and employees of large regulated organisations, and confirm before offering.
4. **What is the participant's own reason to take part, apart from payment?** Interest in the topic, a stake in the outcome, a desire to be heard, professional curiosity. Where this is strong, less money is needed, and where it is strong and one-sided, it is a composition risk in itself. Default: assume topic interest is a motivator and design the invitation not to amplify it (**03.01**).
5. **How likely is misqualification at this level, with this screener?** The risk rises with incentive value and with how obvious the eligibility criteria are. Default: assume some misqualification wherever the incentive is meaningful and the screener is guessable, and specify the detection route (**03.04**).
6. **When and how will people actually be paid, and how reliably?** Payment delay and failure are the most common cause of complaints and of refusal to take part again. Default: assume prompt payment is required and treat any delay longer than a few days as a design flaw to be justified.
7. **Will this be disclosed to an ethics committee, a client or a public audience?** Determines the justification format. Default: write the justification as though it will be reviewed.

## 7. Step-by-step methodology

**Step 1. Fix the purpose in writing before choosing a number.** The incentive compensates a person for time, effort and inconvenience. It does not buy a particular answer, it does not reward a particular kind of respondent, and it is not a reward for having the right opinion. Two design consequences follow immediately and are frequently violated. **Payment must not be conditional on the content of any answer**, including on qualifying for a follow-up stage that depends on what someone said, since that creates an incentive to say it. **Payment must be honoured for participation, not only for completion**, or at least a partial-completion policy must exist and be stated, because paying only completers makes the burden of a long or difficult instrument fall entirely on the participant. A correct result is a one-sentence purpose statement that any subsequent decision can be tested against.

**Step 2. Measure the burden properly.** Build the burden inventory: screening time, set-up or installation, each session's length, the number of separate contacts, any between-contact tasks (diary entries, photographs, record-keeping), travel and waiting, intrusiveness (being observed, discussing something private, involving family or colleagues), cognitive load (recall over long periods, technical judgement), and any equipment or data cost the participant bears. Convert this into a total time equivalent and a separate qualitative note on effort type, because effort is not linear in time: ten minutes of recalling a distressing event is not ten minutes of rating adverts. A correct result is a burden statement that a participant would recognise as accurate if they read it. This inventory is also the honest basis for the invitation, and understating burden in the invitation to raise the response rate produces drop-off at exactly the point the participant discovers the truth.

**Step 3. Set the level from the burden and the population, not from precedent.** The defensible construction is compensation for time at a rate appropriate to that population, adjusted upward for effort type, intrusion and out-of-pocket cost, and adjusted for how scarce the population's time is. This produces a number with a rationale, which is what an ethics committee and a sceptical client both want. Precedent ("we paid this last time") is only a valid input where burden, population, source and period all match, which is rarer than it seems. Then check the level against both failure modes explicitly, because they are opposite and both are real.

- **Under-incentivised.** Participation concentrates in people with their own reason to take part. On a satisfaction study, that is disproportionately the delighted and the aggrieved, and the resulting distribution is more polarised than the population. On a category study, it is the highly involved, and category engagement, usage frequency and stated interest are all inflated. It also raises drop-off, which removes people non-randomly and shortens open ends.
- **Over-incentivised.** The payment becomes the reason to participate. Misqualification rises, because there is now a reason to be the person the screener wants. Effort falls relative to speed, because throughput becomes the objective. And in a population where the amount is large relative to means, the ethical question in Step 8 becomes live.

A correct result records the chosen level, the burden it compensates, the rate implied, and a written statement of which failure mode is the greater risk in this study and what will be watched in field to detect it.

**Step 4. Choose the structure, understanding what each one does.** The structure is a separate decision from the level and it has distinct effects.

- **Fixed guaranteed payment.** Every participant receives a stated amount. It is the most straightforward, the easiest to justify ethically, and the most predictable in its effect on participation, because the participant can evaluate the offer against the burden. It is also the most expensive per complete.
- **Prize draw.** A small chance of a large prize. Cheaper per complete, and its effect on participation is less predictable, because it depends on how individuals evaluate a small probability, which varies widely and systematically. It tends to appeal differentially to people who find the prospect attractive, which is not a random subset, and it provides no compensation to the large majority who receive nothing, which is difficult to defend as compensation for time. Prize draws also carry legal requirements in many jurisdictions (rules, eligibility, publication of winners) that must be met.
- **Tiered by burden.** Different amounts for different levels of participation within one study, for example a screener payment plus a session payment. Fairer where burden varies, and it must be structured so the tier depends on the burden undertaken and never on the answers given.
- **Milestone or completion-weighted, in multi-stage studies.** A larger payment at the final stage improves retention in a diary or longitudinal design. It has a known cost: it makes dropping out expensive, which is a mild form of pressure to continue, and it should be balanced with a guaranteed payment for the stages actually completed.
- **Charitable donation.** Sometimes the only permissible option for professionals, and it works well where the population's motivation is professional or civic rather than financial. It compensates the participant less directly, so it should not be assumed equivalent in pulling power.
- **In-kind, credit or product.** Convenient and often lower cost, and it narrows the population to people who value that particular thing, which is a composition effect that is easy to overlook when the item is the client's own product.

The honest general statement: **guaranteed payment is easier to justify and more predictable in its effect; prize draws are cheaper and less predictable, and are harder to defend as compensation.** Which is appropriate depends on the population, the burden, the budget and the governance regime.

**Step 5. Handle professional, clinical and B2B populations as a different problem.** Three things change. First, the time is genuinely more expensive, so the level must be higher or the pull must come from elsewhere. Second, payment is frequently restricted: many healthcare systems, public bodies and large corporations restrict or prohibit acceptance of payment, and transfer-of-value rules may require disclosure. Third, money is often not the strongest motivator: access to findings, peer benchmarking, professional relevance, and the chance to influence something in their field commonly outperform a modest cash payment among senior and professional respondents. Design accordingly, offer a charitable alternative where personal payment is restricted, and never place a participant in the position of having to decline or conceal a payment their employer prohibits. A correct result names the applicable restrictions, or states explicitly that they were checked and none applies.

**Step 6. Design non-monetary value deliberately, not as a consolation.** Results-sharing is the most commonly offered and most commonly botched non-monetary incentive. Offered as "we will share the findings" and then not delivered, it damages willingness to participate again more than offering nothing would have. Offered properly, with a stated form and date and then delivered, it is genuinely valuable in professional, community and organisational populations, and it can strengthen the relationship the next study will depend on. Other forms include early access, recognition where the participant wants it, influence on a decision, and professional development value. Three rules: only offer what will actually be delivered; be specific about what and when; and do not present a non-monetary offer as equivalent to compensation for a burden that genuinely costs someone money or a day's work.

**Step 7. Design the payment experience, because reliability affects participation more than level does at the margin.** Specify: when payment is made, by what method, what happens on partial completion, what happens if someone is screened out after investing time, what happens if the payment fails, and who the participant contacts. Prompt, reliable payment is one of the strongest determinants of whether a person takes part again, which matters directly for panel-based and longitudinal work and matters indirectly everywhere, because the population's cumulative experience of research is what determines who is willing to answer next year. Screen-outs deserve particular attention: someone who spent five minutes answering screening questions and was excluded has given real time, and a small screening payment or entry to a draw is both fairer and better for the long-term health of the source. A correct result is a payment specification a participant could read and understand in full.

**Step 8. Apply the ethical test, and route it where it belongs.** The relevant standard is undue inducement: an incentive that is large enough, relative to the participant's circumstances, to distort their judgement about risks or burdens they would otherwise decline. It is a function of the participant's circumstances, not of the absolute amount, so the same offer can be entirely appropriate in one population and coercive in another. Three situations need particular care and all three route to **13.05 Research Ethics and Consent Design**: participants in financial hardship, for whom a modest payment may not be refusable; participants in a dependent relationship with the sponsor (patients, employees, service users, students), where the payment compounds a pre-existing pressure to comply; and studies with genuine burden or risk, where a high incentive may be doing the work that a clear explanation of the risk should be doing. Two further rules apply in all cases: the incentive is disclosed in the recruitment material, before consent, not revealed afterwards; and the right to withdraw must not cost the participant the compensation for what they have already done, because a withdrawal that forfeits earned payment is not a free withdrawal.

**Step 9. Test the level at soft launch and record what any change does.** Completion rate, drop-off point and the pace of quota fill are the observable evidence that the offer matches the burden. If participation is well below expectation and the instrument has been cleared as sound (**03.04**), the incentive is a candidate explanation. If it changes, treat it as a partition of the sample: respondents recruited at the old and new offers differ in motivation by construction, capture the offer as a variable, and check the key measures across the two groups before pooling them. Where the level does not change, record that the check was made, since a documented decision to hold is as useful later as a documented change.

**Step 10. Write the disclosure.** The methodology states what participants received, in what form, when, and on what condition, and where relevant the justification for the level. This is not administrative detail. Incentive structure bears directly on the interpretation of participation-sensitive measures, and a reader assessing whether a satisfaction score or a stated intention figure is inflated needs to know whether respondents were paid, how much, and whether payment depended on completion. For studies going to ethics review or publication, this is usually required; for commercial work it is good practice and it costs two sentences.

## 8. Analytical framework

The chain runs from what the study asks to what it can claim:

    Burden → Offer (level × structure × timing) → Who agrees to take part → Effort given → What can be claimed

Two properties make it usable.

**Composition is the mechanism, not response rate.** The temptation is to evaluate an incentive by whether the sample filled. That measures only the fourth arrow's throughput. What matters is the third arrow: which people the offer selected. An offer that fills the sample by attracting the highly interested has not solved the problem, it has moved it into the findings. So the diagnostic question is always **"who is disproportionately likely to accept this offer, and how do they differ on the measures this study exists to produce?"**

**The two failure modes are opposite and both are composition failures**, which is why the level cannot be optimised in one direction. Too low selects for intrinsic motivation (interest, strong opinion, spare time). Too high selects for extrinsic motivation (payment, throughput, willingness to qualify). The design target is the range in which neither dominates, and the honest position is that the range is estimated rather than known, so the level is a hypothesis to be checked at soft launch rather than a setting to be got right in advance.

Read backwards from a result, the chain is a diagnostic. An unexpectedly positive satisfaction distribution, an unusually high stated purchase intent, or a category involvement level that does not match behavioural data are all candidate incentive-composition effects, and checking them requires knowing what the offer was, which is why Step 10 exists.

## 9. Output format

**1. Burden statement.** Every component of participation, with time and effort type, totalled, in language a participant would recognise.

**2. Incentive specification.**

| Element | Decision | Rationale |
|---|---|---|
| Level | | Burden compensated and the rate implied |
| Structure | | Fixed, draw, tiered, milestone, charitable, in-kind, and why |
| Timing and method | | When paid, how, and what the participant is told |
| Partial completion policy | | What a participant receives who starts and does not finish |
| Screen-out policy | | What someone receives who gives time and does not qualify |
| Withdrawal policy | | Confirmation that withdrawal does not forfeit earned compensation |

**3. Composition risk assessment.** Which failure mode is the greater risk here, why, on which measures it would show, and what will be watched in field to detect it.

**4. Restriction check.** Professional, legal, employer and jurisdictional restrictions applicable to this population, with the result of checking each, or an explicit statement that a restriction may exist and has not been confirmed.

**5. Ethics assessment.** The undue inducement test applied to this population, and the routing decision to **13.05** where any of the three Step 8 situations applies. Marked as a review point.

**6. Recruitment material wording** covering the incentive, as the participant will see it before consenting.

**7. Non-monetary commitments**, with the specific form and date of anything promised, particularly results-sharing.

**8. Methodology disclosure paragraph**, in final wording.

**9. Review points** per **K5 §3**.

**When the inputs are thin**, a level that cannot be justified from burden and population data is recorded as `[level not justified]` with the missing input named, rather than presented as a recommendation with a confident number attached. Where restrictions have not been checked, the output says they have not been checked rather than assuming none apply (**K4 §6.1**).

## 10. Quality checks

Run before recruitment material is issued. These sit on top of **K4 §8**.

1. The burden inventory includes screening, set-up, travel and between-session tasks, not only the interview length.
2. The level is derived from burden and population, and the derivation is written down, not asserted.
3. No payment is conditional on the content of any answer, at any stage, including eligibility for a paid follow-up.
4. A partial-completion policy and a screen-out policy both exist and are stated.
5. Withdrawal does not forfeit compensation already earned.
6. Both composition failure modes have been assessed for this study, and the greater risk named.
7. Professional, employer, legal and jurisdictional restrictions have been checked for this population, with the result recorded.
8. The undue inducement test has been applied against the participants' circumstances, not against the absolute amount.
9. Where the population is vulnerable, dependent on the sponsor, or in financial hardship, the ethics route has been taken before the level was fixed.
10. Anything promised that is not money, particularly results-sharing, has a stated form, a stated date, and someone accountable for delivering it.
11. The incentive is disclosed in the recruitment material before consent, in the same terms as in the methodology.
12. In multi-market studies, the level is set per market against local circumstances rather than converted from one market's figure.
13. In a tracker, any change from the previous wave's incentive is recorded as a candidate explanation for movement in the results.
14. The methodology disclosure paragraph exists and states level, structure, timing and condition.

## 11. Common failure modes

| Failure | How to recognise it | How to prevent it |
|---|---|---|
| **Incentive by precedent** | "We paid this last time" with no reference to burden or population | Derive from the burden inventory; precedent is only valid where burden, population, source and period all match |
| **Burden understated in the invitation** | Good response, heavy drop-off in the first minutes | Describe the real burden honestly; the drop-off is participants discovering the truth |
| **Completion-only payment on a long instrument** | High break-off with no compensation for those who tried | A stated partial-completion policy, decided in advance |
| **The interested-only sample** | Category involvement and stated intent far above behavioural reality | Check the level against the under-incentivised failure mode; look for polarised distributions |
| **The professional respondent sample** | High qualification rates, fast completion, short open ends | Level check against the over-incentivised mode, non-obvious screening, misqualification detection (**03.04**) |
| **Prize draw as compensation** | A long, burdensome study offering a chance of a prize | Reserve draws for short, low-burden participation, and never describe a draw as compensation for time |
| **Undeliverable results promise** | "We'll share the findings" with no form, date or owner | Only promise what will be delivered; name the form and the date in writing |
| **Prohibited payment offered** | A clinician or public official asked to accept a fee their employer forbids | Check restrictions before the offer is made; provide a charitable alternative |
| **Mid-field incentive change, unrecorded** | Field pace improves and nobody records why | Treat as a logged amendment that partitions the sample (**03.04**) |
| **Coercive amount in a hardship population** | An amount that a participant in difficulty could not reasonably refuse | Undue inducement test against circumstances; route to **13.05** |
| **AI: invented going rates** | Confident statements about what respondents are typically paid | **K4 §2.4**. Derive from burden and stated local data, or state that no benchmark is available |
| **AI: level without justification** | A specific number produced with no derivation | Output the derivation or output `[level not justified]` with the missing input named |
| **AI: assuming no restrictions** | A payment recommended for clinicians or officials with no restriction check | Assume restrictions may exist in these populations and say they must be confirmed |
| **Silent incentive in the methodology** | The report describes the sample and never says what participants received | Two sentences in the methodology; the reader needs it to interpret participation-sensitive measures |

## 12. AI guardrails

Skill-specific only. **K4** applies in full and is not repeated here.

1. **Never state a going rate, typical incentive or market norm as a fact.** Derive the level from burden and population, or state that no benchmark is available (**K4 §2.4**).
2. **Never produce an incentive level without its derivation.** A number with no burden calculation behind it will be adopted and defended by people who do not know where it came from.
3. **Never design a payment conditional on the content of an answer**, or on qualifying for a later stage in a way the respondent could influence by answering differently.
4. **Never assume payment is permitted** for clinicians, public officials, employees of large or regulated organisations, or any professional population, without a stated check.
5. **Never assess undue inducement against the absolute amount.** It is a function of the participant's circumstances, and the assessment must say whose circumstances were considered.
6. **Never propose an incentive for a vulnerable, dependent or hardship population without routing to 13.05** and marking it as requiring ethical review before the level is fixed (**K5 §2.4**).
7. **Never promise results-sharing, feedback or any other non-monetary return without naming the form, the date and an owner.** An undelivered promise damages the population's willingness to participate in future research.
8. **Never change an incentive mid-field without treating it as an amendment** that partitions the sample and requires the offer to be captured as a variable.
9. **Never omit the incentive from the methodology.** Level, structure, timing and condition bear on interpretation and are part of the method, not part of the administration.

## 13. Best-practice principles

1. **The incentive pays for time and inconvenience, never for an answer.** Every rule in this skill is a consequence of that sentence, and every serious incentive failure is a violation of it.
2. **An incentive that is too low does not produce a smaller sample, it produces a different one.** This is the least intuitive point in the area and the most important. The people who take part without adequate compensation are systematically unlike the ones who do not.
3. **Optimise for who accepts, not for whether the sample fills.** A study that fills quickly at a low incentive has usually recruited the enthusiastic, and the enthusiasm will show up in the findings.
4. **Effort is not linear in time.** Recalling something painful, keeping a diary for two weeks, or letting a researcher into your home are not the same as fifteen minutes of rating adverts, and the compensation should not pretend otherwise.
5. **Pay screen-outs something.** They gave real time and received nothing, and they are part of the population you will need next year. It is fair, and it protects the source.
6. **Reliability beats generosity at the margin.** A modest payment made promptly and without error does more for future participation than a larger one that arrives late or fails.
7. **In professional populations, relevance often outperforms money.** Access to findings, peer comparison and the chance to influence something in their field will move a senior audience that a modest fee will not.
8. **Never promise findings you will not send.** It is the most frequently broken promise in research and it degrades the willingness of a whole population, not just the individual.
9. **Withdrawal must be free.** If leaving costs the participant money they have already earned, the right to withdraw is nominal.
10. **The same amount is generous, fair or coercive depending on who is being asked.** Set the level per population and per market, and never convert one market's figure into another's currency and call it equivalent.
11. **Incentive changes are design changes.** In a tracker, a changed incentive is a candidate explanation for every movement in the results, and it belongs in the wave notes.
12. **Disclose it.** Two sentences in the methodology let a reader judge how much of a favourable distribution is the market and how much is the offer.

## 14. Worked example

**INPUT**

A fictional university social policy department, at Northgate University, is running a study on how low-income households manage energy costs. Design: a 25-minute in-home interview, plus a two-week photo and note diary of heating decisions, plus a 20-minute follow-up call. Target 60 households, recruited through community organisations in three neighbourhoods. Ethics committee approval required. The department's default incentive is a single payment on completion, at a level set five years ago for 30-minute campus interviews.

**PROCESS**

*Step 2, burden.* The default level was being applied to a burden it did not describe. Inventory: 10 minutes screening, 25-minute interview, roughly 5 to 10 minutes per day for 14 days on the diary (a total of 70 to 140 minutes), a 20-minute follow-up, plus being observed at home and discussing a subject many participants find shaming. Total time equivalent of roughly 2.5 to 3.5 hours, with high intrusion and high emotional load. The default incentive compensated about one sixth of it.

*Step 3, level.* Constructed as compensation for time at the local minimum hourly rate, uplifted for intrusion and for the fact that participants would incur their own data and electricity costs to submit diary entries. This produced a number with a derivation, which the ethics committee could evaluate, rather than a round figure.

*The judgement call.* The derived figure was large relative to a weekly income in the target population, and the study population was by definition in financial hardship. This is exactly the undue inducement situation, and it was routed to **13.05** rather than resolved here. The tension was real in both directions: paying too little would exploit people for whom the burden was heaviest, and paying an amount they could not refuse would compromise the voluntariness of consent.

The resolution agreed with the committee had four parts. The total was **split across stages** (screening, interview, diary midpoint, follow-up) so that no single payment was large enough to override judgement, while the total still compensated the whole burden. Payment was **guaranteed for each stage completed**, so withdrawal at any point kept everything earned, and this was stated in the recruitment material in plain language. The **recruitment material described the full burden first and the payment second**, so that the decision was made about the study rather than about the money. And the diary payment was made **at the midpoint rather than the end**, removing the retention pressure that a completion-weighted structure would have created in a population who might have good reasons to stop.

*Step 4, structure.* A prize draw was considered as a way to reduce cost and rejected in one line: a draw provides no compensation to most participants and could not be defended as compensation for a three-hour burden in a hardship population.

*Step 6, non-monetary.* Results-sharing was offered, and made specific: a two-page plain-language summary posted to each household within eight weeks of the study closing, with a named person responsible. The community organisations that hosted recruitment asked for a briefing session, which was agreed and scheduled, since the relationship would be needed for the follow-up study.

*Step 7, payment experience.* Cash and voucher options both offered, because some participants did not have a bank account they wanted the payment to reach and requiring one would have excluded them, which is a coverage effect as well as a fairness one. Payments made at the visit for the first two stages.

*Step 9, soft launch.* First eight recruits reached target quickly and skewed toward one neighbourhood, where the partner organisation was most active. Recorded as a sourcing signal rather than an incentive signal, and passed to **03.01** and **03.04** rather than being addressed by changing the offer.

**OUTPUT**

A staged incentive with a written derivation from burden and local wage data, a withdrawal-safe structure agreed with the ethics committee, a recruitment script that presents burden before payment, a payment method that does not exclude the unbanked, a specific and owned results-sharing commitment, and two review points: whether the midpoint diary payment is sufficient to sustain two-week completion without creating pressure to continue, and whether the neighbourhood skew requires a sourcing intervention before the sample is half filled.

## 15. Advanced usage

**Testing the level.** Where the study is large and the level genuinely uncertain, a split test at two levels, randomly assigned, converts the argument into a measurement. Compare not only completion rate but achieved composition, completion time, open-end length and the key substantive measures. This costs comparability within the study unless the level is captured and controlled for, and it produces evidence usable for years afterwards. It should not be run in populations where the ethics of differential payment for identical burden would be objectionable, which is most populations, so it is usually restricted to low-burden studies where the difference is small.

**Multi-stage and longitudinal designs.** Retention, not recruitment, is the binding constraint. Structure the incentive so that each stage is compensated as it happens and there is a modest additional value for completing the series, then balance that against the pressure it creates to continue. Where attrition is differential (the busiest or the most disadvantaged leaving first), the incentive structure is one of the few levers available, and its effect on who stays should be monitored per wave rather than assumed.

**Multi-market work.** Set the level per market against local circumstances, never by currency conversion. A converted figure is generous in one market and derisory in another, and the resulting differences in who takes part will appear in the data as market differences. State per-market levels in the methodology so cross-market comparison can be read with that in view.

**Incentives in organisational and employee research.** Where the sponsor is the employer, payment adds a layer of pressure to an existing one, and the more effective levers are usually protected time to participate, credible confidentiality, and visible action on previous results. Where an incentive is used, its effect on candour should be assumed non-neutral and the reporting should say so.

**When the standard approach does not fit.** Where money cannot be offered and non-monetary value is thin, the honest options are to reduce the burden until the study is reasonable to ask for, or to accept a smaller and more self-selected sample and report it as such. Neither is a failure. Recruiting a full sample by understating the burden is.

## 16. Skill chain

**Recommended previous skills**
- **03.01 Recruitment and Sample Sourcing.** Establishes the source and the population, which determine what incentive is appropriate and what misqualification risk it carries.
- **02.01 Survey Questionnaire Design** and **02.02 Discussion Guide Design.** Supply the instrument whose length and difficulty are the largest components of the burden inventory.
- **01.05 Research Plan Development.** Supplies the number of contacts, session structure and fieldwork shape that the burden statement is built from.

**Recommended next skills**
- **03.04 Fieldwork Monitoring and Response Quality.** Tests in field whether the offer matched the burden, and handles any mid-field change as a logged amendment.
- **12.03 Research Report Compilation.** Publishes the incentive disclosure drafted here.
- **04.01 Data Validation.** Receives the offer variable where a split test or a mid-field change created more than one incentive condition.

**Runs well alongside**
- **13.05 Research Ethics and Consent Design**, which owns undue inducement and takes over wherever the population is vulnerable, dependent or in financial hardship.
- **02.06 Screener and Quota Design**, since incentive value and screener transparency together determine misqualification risk.
- **13.06 AI Research Governance**, where participation terms and disclosures are part of the governance record.

---
A Yazi Supplied Skill and resource.