When One Result Leads to Conflicting Judgments

A set of ELISA results used to assess biological activity fell outside the test specification. QA considered the product unsuitable for release. Manufacturing believed the process had not gone wrong and questioned whether the specification was too strict. R&D attributed the results to the technique of the QC personnel. I was the QC manager, and I questioned whether the method-validation sample size was sufficient to support confidence in consistent method performance.

QA had stated a position on release. The other assessments proposed explanations or questioned the basis for interpreting the results. Knowing that the results fell outside a specification did not establish why it had happened. Another round of departmental positions could not determine what to investigate unless we examined the reasoning behind them.

One Result, Several Assessments

This happened at a GMP pharmaceutical manufacturer, where I worked with QA, manufacturing, and R&D on deviations, deficiencies, and corrective and preventive measures in antibody manufacturing.

The disagreement continued for a long time. Our boss became involved, but because the medicine was intended for clinical trials, he also remained unwilling to decide. Escalation had occurred without supplying an explanation for the results.

I recognized that assessments arising from different professional perspectives could be interpreted as passing responsibility between departments. Treating the exchange only as a dispute over responsibility would leave the competing explanations unexamined.

I brought the participants together for a calm discussion, and we developed a detailed research plan. Over two months, we progressively investigated and excluded suspected factors. The conclusion, as I recall it, was that the results reflected measurement error inherent in the method rather than a departmental lapse. That finding did not prove my concern about validation sample size, establish the acceptability of the out-of-specification results or product, or settle release.

Reading these assessments now, treating them as blame-passing would confuse their different purposes. A position against release does not identify the cause of a result. Confidence in the process does not, by itself, establish that the specification is too strict. And attributing a result to technique requires support that a concern about validation cannot supply. These are connections to examine, not interchangeable answers to the same question. Calling an assessment defensive would add a judgment about its source without answering its substance.

The practical lesson is in the work we arranged. The detailed plan gave the suspected factors a way to be examined. More research was useful here because it addressed the competing explanations.

Find the Premise Behind the Recommendation

Begin by writing the decision the review is meant to support and identifying the information participants are using. If contributors are deciding different things or relying on different records, clarify that first. Once the scope is shared, ask each contributor to connect the proposed action to the claim that makes it follow.

In the ELISA incident, my concern about validation sample size could justify examining the basis for confidence in the method. It was my explanation to put under examination alongside the others. Convening the investigation gave me no reason to exempt it.

For the person convening the review, that means making their own reasoning available for examination. I would want the group to see what connects my sample-size concern to confidence in the method, and where that connection needs support. The same investigation should be capable of weakening my explanation as well as strengthening it. If it does weaken it, the next work should follow the better-supported explanation, even when someone else proposed it.

Have the contributor check the written premise. The chair should not translate an expert’s reasoning into a convenient claim the expert never made. A possible cause must remain a possible cause. If its connection to an action is missing, ask for that connection rather than supplying it.

A recommendation may depend on several premises. Start with one whose resolution could change the recommendation or determine what review is needed next. Keep other material issues visible, but do not give every issue equal agenda time simply because each department raised one.

Ask how the proposed work would differ if a disputed claim were accepted or set aside. If both contributors still want the same records examined next, their disagreement about the eventual explanation can remain open during that work. Give earlier attention to a claim that changes which records are relevant, whose expertise is needed, or whether the proposed examination can address the concern. That distinction keeps the agenda focused without treating the deferred disagreement as settled.

Ask each contributor what finding, clarification, or changed condition would lead them to reconsider the recommendation, and why. Let them supply the connection. If they cannot name one yet, explore whether they need a better understanding of the issue or are objecting to a consequence that analysis cannot settle. A concern that cannot currently be tested may still restrict the action; record what it affects.

Choose Work That Can Address the Dispute

Four lenses can help choose the work: observation, interpretation, execution constraint, and trade-off. They overlap and do not cover every disagreement. This is a retrospective framework, not a procedure we used in the ELISA investigation.

Observation: examine whether the measurement is dependable. Identify the relevant records and ask the appropriate experts to define an informative investigation. My concern about method validation belongs here: whether the evidence adequately supported confidence in consistent method performance. Naming that question gives the technical review a subject.

Interpretation: compare what contributors think the finding explains or supports. Ask what supports each explanation and what finding could distinguish them. Expert examination may reveal that the available material cannot yet do so and that further evidence is needed.

Manufacturing’s challenge to the specification shows why these distinctions need care. Its scientific and applicable procedural basis would need examination. Treating that challenge immediately as a negotiation over acceptable risk would skip the question of whether the specification had an adequate basis.

Execution constraint: substantiate a claim about timing, capacity, or dependencies. If a proposed review cannot be completed within the decision window, identify the basis for that timing claim and whether a change in resources or dependencies could alter it. Establishing feasibility narrows the available actions.

Trade-off: make the consequences of the feasible options explicit for the person or body entitled to choose among them. When the relevant findings and practical conditions are accepted, the remaining disagreement may concern which consequence the organization is willing to accept. State which consequences the choice would accept.

Several kinds of work can belong to one issue. An expert review might clarify scientific support while leaving timing unresolved. Confirming the timing might then expose a choice about delay. Record that sequence so completion of the first task does not quietly stand in for resolution of the whole disagreement.

An analysis assignment should explain why its material bears on the question. In the ELISA case, a concern about method reliability needed work capable of examining that reliability; repeating the reported result could not explain it. The move to investigating suspected factors matters because an explanation could then be challenged through the work. The practical demand is to connect a proposed examination to the reason for doubting the result, and let technical experts assess whether the examination can make that connection.

Timing questions need the same discipline. Compare an estimate with the commitments on which it depends, and identify whether those records are specific enough to test the claim. A broad availability estimate may simply reproduce the uncertainty; a further comparison then needs a clearer purpose.

At review, separate delivery of the analysis from its answer: it may support a conclusion, undermine it, or leave the original question unresolved. In that last case, say why the evidence remains inconclusive and decide whether another examination could address it. A completed report alone is no reason to close the issue.

When Further Analysis Cannot Settle the Choice

The ELISA dispute needed a technical explanation. Consider a different, hypothetical program review: a scientific lead and a program lead accept the time needed for an additional review of resource assumptions. They disagree about whether an internal planning decision should wait for it. The scientific lead recommends waiting so the assumptions can be examined before the plan is committed. The program lead recommends retaining the date to avoid postponing planning.

Both leads accept that waiting could clarify the assumptions and that committing now could require later changes. They give those consequences different weight. The scientific lead puts more weight on examining the assumptions first; the program lead puts more weight on giving dependent work a plan now. Agreement on the time required does not settle which consequence to accept.

At the next review, the scientific lead should explain what waiting could clarify. The program lead should set out what it postpones and what committing the plan now could make harder to change. The sponsor then chooses whether to move or retain the planning date, within the sponsor’s authority over that date. Keep a material objection and its basis in the record alongside the reason for the choice. That discretion does not waive technical requirements.

An action can be selected while an expert continues to object to its consequences. Selecting an action does not require rewriting the expert’s assessment as agreement.

Before commissioning another analysis, specify what answer it could provide and whether that answer could address the objection. If the disagreement would remain even with the proposed answer accepted, identify the remaining choice directly.

Build the Next Agenda Around Unresolved Premises

Connect each recommendation in the review materials to the premise supporting it, the next work, a responsible role and review point, and any remaining dissent.

For the hypothetical timing example, one agenda item is enough: decide whether the planning date should move, using the two leads’ comparison of waiting and committing now. The program lead assembles that comparison for the next review.

Bring in the contributors and technical detail needed to examine each issue, including shared discussion where the reasoning crosses functions. Departmental updates can still provide context, but their order should not determine how the issue is handled.

The record should also help a later reviewer recognize when changed conditions undermine the basis for the choice. A continuing difference in preference, on its own, adds no new evidence.

Before sending the next agenda, check which unresolved premise each item will examine and whether the assigned work addresses it.

Check whether the materials separate supported conclusions from contested ones, in wording the contributors recognize as their own reasoning.

Then check that consequential dissent and its reasoning remain in the record. The next reviewer may have only that record, without having heard the discussion.