Few phrases in the National Statement on Ethical Conduct in Human Research 2025 (a.k.a. the National Statement) generate as much confusion as this one: participants are “entitled to withdraw from the research at any stage.” It appears in paragraph 2.2.20, and on its face it sounds straightforward. But for evaluators and researchers working in the real world of surveys, focus groups, and program evaluations, it raises a genuinely difficult question: what happens when a participant wants to withdraw after their data has already been analysed, or after findings have been published?
This post unpacks the practical meaning of paragraph 2.2.20, explains where the National Statement draws its lines, and sets out what we advise evaluators and researchers to include in their consent forms to handle this issue properly.
What paragraph 2.2.20 actually says
The paragraph is short:
“Participants are entitled to withdraw from the research at any stage. Before consenting to involvement in the research, participants should be informed about any consequences of such withdrawal.”
The key word here is “consequences.” The National Statement does not say that withdrawal is consequence-free. It says that participants must be told what those consequences are before they consent. This is a disclosure obligation, not a guarantee that data can always be retrieved or removed.
Paragraph 2.2.6(g) reinforces this. It identifies as one of the matters that should be communicated to participants:
“...the participant’s right to withdraw from further participation at any stage, along with any implications of withdrawal and whether it will be possible to withdraw data.”
Note that second clause: “whether it will be possible to withdraw data.” The National Statement is explicitly acknowledging that data withdrawal may not always be possible. The obligation is to tell participants this upfront, not to make it possible.
The confusion in practice
The phrase “at any stage” is the source of most of the confusion we see. Evaluators and researchers sometimes interpret this as meaning a participant can demand that their data be removed from a published report, or that their responses be extracted from an already-aggregated dataset. Ethics committee members occasionally face applications that have not grappled with this tension at all, either glossing over it in the consent form or making promises that cannot be kept.
Consider a few common scenarios:
- A community survey for a local government program evaluation collects 400 responses. Data are aggregated and analysed; a report is provided to the commissioning agency. Three months after the report is finalised, one participant contacts the evaluation team and requests that their responses be withdrawn. By this point, their individual responses cannot be meaningfully separated from the aggregate analysis. Withdrawal from “the research” is possible in the sense that the participant will not be approached again for follow-up. But withdrawal of data is not.
- A focus group is conducted for a market research study exploring attitudes toward a new product concept. Verbatim quotes, attributed only to a participant type (e.g., “female, 35–44, metropolitan”), appear in the final report. A participant later recognises a quote as their own and wants it removed. If the report has been provided to the client and is in active use, removal may be impractical or impossible.
In both cases, the obligation under paragraph 2.2.20 is not to make withdrawal retroactively possible. It is to have told participants, before they consented, that this was the situation they were agreeing to.
What this means for your consent forms
Our advice to evaluators and researchers is to address this directly in consent documentation rather than leaving it vague or, worse, overpromising.
The most effective approach is to amend your participant information sheet and consent form to explain both rights clearly and separately: the right to withdraw from the activity itself, and the more limited right to request withdrawal of data, along with the point at which the latter becomes impracticable.
A well-drafted clause might read something like this:
“You may withdraw from this [research/evaluation] at any time without penalty and without needing to give a reason. If you wish to withdraw your data, you may do so up until [date / commencement of data analysis / submission of the final report]. After this point, your data may have been de-identified and incorporated into aggregated findings in a way that makes individual responses impossible to separate or remove.”
Where a fixed date is known (for instance, data collection closes on a specified date and analysis begins the following week), use it. Specificity is better for participants because it gives them a concrete window within which to act. It is also better for evaluators and researchers because it reduces ambiguity about what commitments have been made.
Where a fixed date is not known at the time of consent, for example in longer-running projects or evaluations with iterative data collection, it is acceptable to refer to a stage of the project instead. “Commencement of data analysis for the phase in which you participated” is a reasonable reference point and is consistent with the National Statement’s expectations.
What ethics committees look for
Understanding what ethics committees look for when reviewing consent documentation can help evaluators and researchers get this right from the outset. It also makes the review process more transparent. When an application comes before a committee, reviewers will typically examine how withdrawal is addressed in the participant information sheet. A consent form that simply restates 2.2.20 without contextualising it for the specific project is not sufficient. The questions a committee will ask include:
• Does the form distinguish between withdrawal from participation and withdrawal of data? These are different things, and participants deserve to understand the difference.
• Is the cut-off point for data withdrawal identified? Either a date or a project stage should be specified. If neither is provided, the committee will ask the applicant to explain why and whether a stage reference is feasible.
• Are the consequences of data aggregation or de-identification explained? Where data will be combined with others’ responses or stripped of identifiers, participants should be told that this affects the practical possibility of data removal.
• Is the language realistic? A consent form that promises data can be removed at any time, without qualification, is likely to create problems down the track and may not reflect what is actually feasible for that project.
A note on informed consent
It is worth stepping back to notice that this is fundamentally about informed consent, which is the bedrock of Chapter 2.2 of the National Statement. Participants who are given accurate, specific information about what withdrawal does and does not mean are in a stronger position to make a genuinely voluntary decision about participation. Glossing over the limits of data withdrawal does not protect participants; it simply defers a problem that may be harder to resolve later.
The National Statement’s framing is honest: consequences of withdrawal must be disclosed, and the form should reflect the practical realities of the specific project. That is not a weakening of participant rights. It is what genuine respect for those rights actually looks like.
Summary
Paragraph 2.2.20 of the National Statement entitles participants to withdraw at any stage and requires that they be informed of the consequences of doing so before they consent. It does not guarantee that data can always be removed. Evaluators and researchers should address this explicitly in consent documentation, distinguishing between withdrawal from participation and withdrawal of data, and specifying the point after which data withdrawal becomes impracticable. Ethics committees look for this distinction when reviewing applications and will raise it where it is absent.
If you have questions about how to handle withdrawal of consent in your consent documentation, or if you are unsure whether your current forms adequately address this issue, get in touch with the team at Iris Ethics.
AI Disclosure: Initial drafts of the content for this article were prepared using Large Language Models with input from Iris Ethics staff who guided the scope and design. Subsequent revisions and final versions were developed and approved by Iris Ethics staff.