03 · What You Need to Know
How to Plan for Unexpected Problems Without Compromising Research Integrity
What Is a Research Contingency Plan?
A research contingency plan describes how the research team intends to respond when specified problems interfere with planned activities.
It may identify alternative procedures, decision criteria, responsible personnel, and actions necessary to protect participants or preserve data quality.
For example, a protocol may specify how an interrupted electronic assessment will be handled, including when an assessment may be resumed, repeated, or recorded as incomplete.
Contingency planning differs from merely listing possible risks. A risk register identifies what could go wrong, whereas a contingency plan establishes what should happen if a relevant problem occurs.
Both can contribute to research quality management, but their usefulness depends on whether the proposed responses are scientifically appropriate and operationally feasible.
Why Should Contingencies Be Considered Before Data Collection?
Advance planning can reduce the likelihood that consequential decisions will be made inconsistently or under pressure.
For example, if a primary assessment is interrupted, different research assistants might independently extend the time limit, restart the assessment, or exclude the participant.
These responses could produce systematically different measurement conditions.
Establishing decision rules before collection can improve consistency and help distinguish planned responses from later methodological departures.
However, advance planning does not guarantee that every response will be appropriate. The research team must still evaluate actual circumstances and applicable governance requirements.
Which Unexpected Problems Should Receive Priority?
Priority should be given to events whose consequences could materially affect the study, even when their probability is relatively low.
| Potential problem |
Principal risk |
Possible contingency |
| Recruitment shortfall |
Insufficient sample size or altered participant composition. |
Approved recruitment extensions or alternative channels. |
| Equipment failure |
Invalid or missing measurements. |
Validated backup equipment or temporary suspension. |
| Digital platform outage |
Interrupted assessments or incomplete records. |
Defined interruption, rescheduling, and recovery procedures. |
| Research personnel unavailable |
Delayed or inconsistent implementation. |
Qualified backup personnel and documented delegation. |
| Participant withdrawal |
Incomplete follow-up and potential selection effects. |
Approved withdrawal and data-handling procedures. |
| Data security incident |
Unauthorized access or disclosure. |
Containment, escalation, and required notification. |
| Intervention disruption |
Reduced fidelity or altered exposure. |
Permitted rescheduling or controlled alternative delivery. |
| Site closure or access loss |
Interrupted recruitment or collection. |
Suspension, approved relocation, or reassessment of feasibility. |
These examples are not universally required contingencies. Researchers should prioritize problems relevant to the actual study and its methodological vulnerabilities.
How Should Contingency Risks Be Assessed?
A practical assessment considers the likelihood of an event, its potential consequences, the ability to detect it, and the availability of effective responses.
For example, temporary internet outages may be relatively common in a distributed digital study, but their consequences depend on whether responses are saved locally and whether assessments can resume without changing their meaning.
A rare failure involving loss of identifiable participant data may warrant strong preventive and response arrangements because its consequences could be serious.
Researchers should avoid treating an arbitrary numerical risk score as an objectively established probability.
Qualitative risk classifications can be useful when supported by clear reasoning and reviewed by personnel familiar with the procedures.
What Should a Contingency Plan Specify?
A useful plan identifies the event, triggering conditions, permitted response, responsible authority, documentation, and criteria for returning to normal procedures.
| Planning element |
Question to answer |
| Trigger |
What event or condition activates the contingency? |
| Immediate action |
What must happen to protect participants or preserve critical information? |
| Permitted alternative |
Which response is already authorized? |
| Decision authority |
Who determines whether the alternative can be used? |
| Documentation |
What information must be recorded? |
| Escalation |
When is scientific, ethics, or institutional review necessary? |
| Recovery |
What conditions must be met before normal activities resume? |
| Impact assessment |
How will affected participants, measurements, or procedures be evaluated? |
The plan should be specific enough to guide decisions without creating unnecessary instructions for inconsequential events.
Should Recruitment Shortfalls Have a Contingency Plan?
Recruitment shortfalls deserve advance consideration when they could prevent the study from obtaining sufficient evidence.
Possible responses include extending the recruitment period, using additional approved recruitment channels, or reassessing feasibility.
However, researchers should not automatically relax eligibility criteria or change the target population merely to increase enrollment.
Such changes may alter the scientific meaning of the sample and require additional approval.
In confirmatory studies, sample-size adjustments and stopping decisions may also affect statistical inference.
Recruitment contingencies should therefore be consistent with the approved participant recruitment plan and relevant sample-size commitments.
How Should Equipment Failure Be Managed?
Equipment contingencies should identify whether a measurement can safely continue, whether an approved replacement is available, and how measurement comparability will be evaluated.
For example, replacing a malfunctioning temperature sensor may be appropriate when the alternative instrument meets the required specifications and calibration conditions.
However, replacing an achievement assessment with a different instrument is not automatically equivalent merely because both produce numerical scores.
Where equivalence cannot be established, suspending collection may be preferable to obtaining measurements of uncertain validity.
The contingency should identify who authorizes equipment substitution and how affected records will be flagged or reviewed.
What Should Happen During a Digital Data Collection Failure?
Digital studies may experience platform outages, connectivity interruptions, software errors, or incomplete synchronization.
A contingency plan should distinguish temporary technical interruption from actual data loss.
For example, an assessment platform may preserve responses locally during an outage, allowing collection to resume without changing the test conditions.
Another system may lose unsaved responses, making a repeat assessment necessary or potentially inappropriate.
The research team should test recovery procedures before deployment where feasible.
Technical convenience should not determine whether a repeat measurement is scientifically acceptable.
Should Participant Withdrawal Be Treated as an Unexpected Problem?
Withdrawal is a foreseeable aspect of participant-based research and should be addressed through appropriate consent and data management procedures.
Participants should not be pressured to remain in a study merely because withdrawal would complicate analysis.
The protocol should explain relevant withdrawal arrangements and how previously collected information will be handled, consistent with applicable consent terms and governance requirements.
Missing outcomes resulting from withdrawal may require appropriate analytical treatment.
However, withdrawal should not be framed as participant noncompliance with an obligation to complete the study.
How Should Unexpected Participant Protection Concerns Be Handled?
Participant protection takes priority when an unexpected event creates a serious or immediate risk.
The protocol should identify who is responsible for responding, how relevant activities may be suspended, and which authorities must be notified.
For example, a confidentiality breach may require immediate containment and assessment under institutional information security procedures.
Urgent protective action may be necessary before the ordinary amendment process is completed, subject to applicable requirements.
Researchers should not assume that every operational problem qualifies as an emergency permitting unrestricted procedural changes.
What If a Research Assistant Becomes Unavailable?
Personnel contingencies should address continuity of critical tasks and the qualifications necessary for replacement staff.
For example, an assessment administrator should not be replaced by an untrained assistant simply because the original collector is unavailable.
The team may identify qualified backup personnel, maintain current training records, and establish procedures for transferring responsibilities.
Where responsibilities involve consent, confidential information, or specialized equipment, authorization requirements should be verified before reassignment.
The objective is continuity without weakening the safeguards governing the activity.
How Should Intervention Disruptions Be Addressed?
Intervention studies should consider disruptions that could change the intended treatment exposure.
For example, a university closure may prevent scheduled instructional sessions, or a digital platform failure may interrupt AI-supported feedback delivery.
Permitted alternatives may include rescheduling within an approved window or using a validated backup delivery arrangement.
However, removing essential intervention components or changing the comparison condition may constitute a substantive methodological change.
The plan should identify essential intervention functions and how deviations will be recorded.
Implementation records may also support later evaluation of intervention fidelity.
Should Contingency Plans Include Missing Data Procedures?
Yes, when missing data are foreseeable and potentially consequential.
The protocol should distinguish preventing avoidable missingness from analyzing data that remain missing despite appropriate collection procedures.
For example, the team may establish reminders for scheduled assessments, procedures for authorized rescheduling, and checks for incomplete records.
The statistical analysis plan should separately address how missing outcomes will be handled under defensible assumptions.
Researchers should not automatically replace missing values with zero or exclude incomplete cases without considering the consequences.
Missing-data decisions should remain consistent with the study design and intended estimand.
Can a Contingency Plan Include Alternative Research Instruments?
Potentially, but instrument substitution requires careful justification.
Different instruments may not measure the same construct or produce directly comparable scores.
For example, replacing a validated achievement assessment with an untested questionnaire may fundamentally alter the primary outcome.
If alternatives are anticipated, the protocol should specify the conditions for their use and the evidence required to establish suitability.
Where an alternative changes the approved measurement strategy, additional scientific or ethics review may be necessary.
It may be more defensible to suspend a measurement than to collect data whose meaning cannot be established.
How Do Contingency Plans Relate to SOPs?
The protocol should establish consequential contingency decisions, while SOPs may describe how approved responses are implemented.
For example, the protocol may permit rescheduling a primary assessment following a qualifying technical interruption.
The corresponding SOP may describe how the interruption is recorded, who authorizes rescheduling, and how the participant's assessment record is updated.
This division avoids unnecessary duplication while preserving the scientific commitments of the study.
Researchers should ensure that supporting instructions remain consistent with the approved protocol.
What Is the Difference Between a Contingency and a Protocol Amendment?
A contingency is a planned response to a specified condition. A protocol amendment is an authorized change to the approved study plan.
Implementing an alternative already permitted by the protocol may not require amendment.
However, a response outside the approved boundaries may change the design or procedures and require additional review.
For example, rescheduling an assessment within an approved window differs from permanently replacing the primary outcome instrument.
The principles for prespecifying operational decisions can help distinguish planned alternatives from consequential unplanned changes.
Can Contingency Planning Affect Confirmatory Inference?
Yes. Some contingencies involve decisions that influence statistical properties or the interpretation of results.
For example, extending recruitment after examining treatment effects or changing the primary outcome after preliminary analysis can compromise the credibility of confirmatory claims.
Prospective decision rules may reduce these risks, but they must be statistically and methodologically appropriate.
Researchers should distinguish adaptations driven by operational information from decisions influenced by emerging outcome patterns.
Where a change is unavoidable, its timing, justification, and consequences should be disclosed.
How Do Contingency Plans Work in Qualitative Research?
Qualitative studies may require flexible responses to changing field access, participant circumstances, or emerging lines of inquiry.
A contingency plan can establish ethical boundaries and decision principles without predetermining every methodological action.
For example, an interview study may explain how researchers will respond if participants become distressed, decline recording, or request a different interview arrangement.
Iterative sampling and responsive questioning may already be part of the approved methodology rather than unexpected deviations.
The relevant distinction is between justified methodological flexibility and changes that exceed the approved research purpose or participant protections.
Should Contingencies Be Tested Before Collection?
Testing can be valuable when a contingency involves complex or critical procedures.
For example, researchers may simulate a digital platform outage, verify backup procedures, or rehearse the transfer of responsibilities to replacement personnel.
Such exercises can reveal ambiguities that are difficult to identify through document review alone.
However, not every low-risk contingency requires a formal simulation.
Testing effort should reflect the likelihood and consequences of the relevant failure.
How Should Contingency Decisions Be Documented?
Documentation should establish what occurred, which contingency was activated, who authorized the response, and whether affected data require additional assessment.
For example, an interrupted assessment record may include the event time, technical problem, approved response, affected participants, and resulting measurement status.
Where the response falls outside approved procedures, a deviation or amendment record may be necessary.
Documentation should be proportionate to the event's significance and protect confidential participant information.
The principles for documenting research SOP deviations are relevant when the response involves a procedural departure.
How Much Contingency Planning Is Enough?
There is no universal number of contingency scenarios required for a research protocol.
The appropriate amount depends on the complexity of the study, foreseeable vulnerabilities, participant risks, and the consequences of interrupted procedures.
A simple anonymous survey may need relatively limited contingency arrangements. A multisite intervention involving critical measurements and sensitive data may require more extensive planning.
Researchers should prioritize scenarios that could materially compromise the study rather than attempt to anticipate every minor inconvenience.
The objective is resilience with methodological accountability, not exhaustive prediction.
Watch Out
A contingency plan should not function as unrestricted permission to change the study when implementation becomes difficult. Alternatives affecting eligibility, primary outcomes, intervention components, or confirmatory analyses may require additional authorization and disclosure, even when the underlying disruption was unexpected.
