home  /  evidence & admissibility  /  sampling, QA/QC & data usability
evidence & admissibility · toxic tort

Sampling, QA/QC and data usability.

A laboratory result is only as good as the sample behind it. Most challenges to environmental data are not about the analysis — they are about everything that happened before it.

get started

What was the exposure?

Start a conversation with our AI Research Concierge, already scoped to sampling, QA/QC & data usability. Pick a starting point, or describe your situation directly.

AI Research Conciergesampling, QA/QC & data usability · triage, not a substitute for an expert
Tell me what data you have or what sampling is planned, and what question it needs to answer. I'll help scope the usability review or the design.

Environmental data arrives looking authoritative: a number, a method reference, a laboratory letterhead. Whether it can support the conclusion drawn from it depends on a chain of decisions upstream of the instrument — where samples were taken and why, whether the container and preservative were right, whether holding times were met, whether blanks demonstrate that nothing was introduced during collection, and whether detection limits were low enough that a non-detect means anything. A dataset assembled for regulatory compliance often fails to answer litigation questions not because it is unreliable but because it was designed to characterise a site rather than to establish whether contamination reached a particular receptor.

mechanisms

Where data usability fails.

Each of these has been the basis for excluding or discounting an entire sampling programme.

Chain of custody

Documented control from collection through analysis. Gaps undermine every result they touch.

Holding times & preservation

Analyte-specific limits after which results no longer reliably reflect the sample as collected.

Blank contamination

Field, trip and equipment blanks showing whether the sampling itself introduced the analyte. Critical for PFAS.

Detection & reporting limits

Whether limits were low enough that a non-detect excludes the concentration of interest.

Spatial & temporal design

Whether locations and timing were chosen to answer the question being asked of the data.

Method appropriateness

Whether the analytical method suits the matrix and the analyte — compliance methods are not always fit for forensic use.

methodology

What the evidence shows — and what we examine.

How data is assessed and how programmes are designed.

Data usability reviewSystematic assessment against project objectives — what the dataset can and cannot support.
Objective-driven designDefining the question first, then designing sampling to answer it, rather than reusing a compliance programme.
QA/QC protocolBlanks, duplicates, spikes and splits sufficient to demonstrate the data is what it claims to be.
Split samplingParallel samples to the opposing party or an independent laboratory, which forecloses a whole category of dispute.
what's at stake

What data quality decides.

Data defects are unglamorous and frequently fatal — they can remove the factual basis for every expert opinion at once.

exclusion of an entire dataset the factual basis for every opinion plume delineation and class boundaries the cost of resampling, if still possible credibility of the technical case whether resampling is possible at all

PFAS made blanks decisive.

PFAS occur in ordinary field materials — some waterproof clothing, tubing, containers and personal care products — at levels that matter when the concentrations of interest are extremely low. Field and equipment blanks are not a formality here; a programme without them has no answer to the argument that the samplers brought the contamination with them.

common questions

Data usability — practical questions.

Can regulatory compliance data be used in litigation?

Often yes, with limitations that should be identified early rather than discovered at deposition. Compliance data is generally well documented and was collected under obligation, which helps. But it was designed to answer a regulator's question — is this facility within its limits, is this site characterised — not a litigant's. Monitoring locations may not sit on the pathway to the claimed receptor, sampling frequency may miss episodic releases, the analyte list may exclude the compound at issue, and detection limits may be above the level that matters for exposure.

What does a non-detect actually establish?

Only that the analyte was not present above the reporting limit, which is a much weaker statement than "not present" — and the difference is frequently decisive. If the reporting limit sits above the concentration relevant to exposure or to a screening value, the non-detect excludes nothing that matters. This is a recurring problem where older data is offered to show contamination was absent: analytical capability has improved substantially, and a historical non-detect at a high reporting limit is consistent with concentrations that would be actionable today.

How important is split sampling?

Disproportionately, relative to its cost. Providing parallel samples to the opposing party or an independent laboratory removes an entire category of dispute — the other side cannot argue the data is unreliable when they hold a split showing the same result, and where splits diverge that is itself important information discovered early rather than at trial. It also signals confidence. Where sampling is happening at a site both parties are aware of, splits are usually worth offering even when not required.

What if sampling was already done badly?

Establish exactly what the defects are and what the data can still support, rather than either defending or abandoning it wholesale. Some defects are fatal to particular uses and irrelevant to others — an exceeded holding time for a volatile compound may invalidate that result while leaving metals data unaffected. Where the site still exists, resampling is usually better than litigating over a flawed dataset. Where it does not, the analysis has to be candid about the limitations, and conclusions drawn have to be ones the compromised data genuinely supports.

related

Related specialization areas & resources.

Will the data hold up?

Describe the dataset or the sampling planned. We will scope it and connect you with the right expert.

contamination assistanttriage · not a causation opinion
Tell me what data you have or what sampling is planned, and what question it needs to answer. I'll help scope the usability review or the design.