Products & safety
Representative sampling: batches, pooled samples, chain of custody and limits of a test
On this page
Sampling selects material to answer a question about a larger target, such as a defined production lot. A representative sample needs a justified selection plan and appropriate handling. A highly capable instrument can describe the submitted material accurately while the selection fails to represent the batch a consumer cares about.[1]
Different sampling choices produce different evidence
| Term | What it establishes | Boundary |
|---|---|---|
| Probability sampling | Selection gives target units known probabilities of inclusion | Supports specified statistical inference when assumptions hold[2] |
| Judgmental sampling | Selection uses knowledge about likely locations or conditions | Useful targeting does not provide the same population inference[2] |
| Composite sampling | Several portions are combined for analysis | Pooling can obscure variation among the original portions[3] |
| Lot acceptance sampling | A defined plan informs acceptance or rejection of a lot | Its main goal is the decision, rather than a full estimate of lot quality[6] |
Define the target before choosing packages
A question about one production lot differs from a question about a brand across multiple factories or years. EPA’s sampling-design guidance distinguishes probabilistic designs from judgment-based targeting and explains how design constrains inference. Random selection, stratification and targeted sampling serve different purposes. A large pile of conveniently purchased products can still omit important parts of the target. Sample size cannot fix a selection process that systematically misses locations, dates or product types relevant to the claim.[2]
Contamination may be unevenly distributed
FDA’s microbiological sampling chapter emphasizes representative sampling when pathogens or toxins are sparsely distributed. Lot composition, food consistency and handling affect the plan. Original unopened containers and suitable clean collection methods help preserve validity. A single scoop from an accessible edge may answer a different question from planned portions across a lot. This microbiological guidance illustrates the principle; it is not a universal chemical-testing recipe or a home instruction for safely collecting hazardous materials.[1]
Evidence & uncertainty
One result cannot automatically describe every package, production date or future batch. Product testing, environmental swabs and emergency investigations have different purposes. Useful interpretation starts with the target, how samples were chosen and what conclusions the design can support.[5]
Selected-source overview · Independent review pendingPublished questions
Can a single tested package stand for the whole brand?
Selected-source overview · Independent review pending. Read the evidence context.
Only to the extent justified by the target and sampling design. A convenience purchase cannot automatically represent different factories, production periods or formulations. Ask how the sampled units were selected and which population the conclusion actually covers.[2]
Evidence behind this answer
- Guidance for choosing a sampling design, QA/G-5S
technical or regulatory guidance · 2002
Selected original official guidance passages inspected; no independent full-document appraisal · Sections 2.3–2.4 and 4: probabilistic versus judgmental inference
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Does a pooled result show every portion was below a limit?
Selected-source overview · Independent review pending. Read the evidence context.
No. A composite describes combined material under its preparation procedure. It can conceal differences among portions. Decisions about individual units or localized high concentrations may need a different design from decisions about an average.[3]
Evidence behind this answer
- Measurement uncertainty arising from sampling, second edition
technical or regulatory guidance · 2019
Selected original PDF sampling-uncertainty and design passages inspected · Table 1; sections 6 and 9; sampling, preparation and duplicate designs
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
What does chain of custody add to a report?
Selected-source overview · Independent review pending. Read the evidence context.
It documents identity, handling and transfers so the examined sample can be connected to the collected material. FDA’s official procedures emphasize identification and sealing as well as records. It strengthens that connection; it does not establish that every unsampled package is identical.[4]
Evidence behind this answer
- 2026 IOM chapter 4: sampling
technical or regulatory guidance · 2026
Selected original PDF official sample-integrity passages inspected · Sample validity; identity, sealing, documentation and chain of custody
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Does a negative sample prove that a batch contains no pathogen?
Selected-source overview · Independent review pending. Read the evidence context.
A result concerns selected material and a defined method. Sparse or uneven distribution can prevent selected portions from capturing contamination. FDA emphasizes using the appropriate sampling plan and preserving sample condition. Negative testing alone should not be expanded into a guarantee of absence.[1]
Evidence behind this answer
- BAM chapter 1: food sampling and preparation
technical or regulatory guidance · 2022
Selected original PDF microbiological sampling and handling passages inspected · Introduction; collection, transport, receipt and sample identity
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Is a positive factory swab the same as a contaminated food serving?
Selected-source overview · Independent review pending. Read the evidence context.
No. Environmental and product samples have different targets. A surface result can identify a concern requiring investigation, but it is not by itself a measurement of a particular serving. Product results and relevant production context are needed for that conclusion.[5]
Evidence behind this answer
- Sampling to protect the food supply
technical or regulatory guidance · Date not stated
Selected original official guidance passages inspected; no independent full-document appraisal · Types of sampling: product, environmental and emergency response
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Does batch passed mean each item was tested?
Selected-source overview · Independent review pending. Read the evidence context.
Usually a lot-acceptance statement refers to a specified sampling plan, not inspection of every unit. Ask for the plan, acceptance criteria and sampled lot. The evidential strength depends on those choices and cannot be inferred from the word passed alone.[6]
Evidence behind this answer
- What is acceptance sampling?
technical or regulatory guidance · Date not stated
Selected original official guidance passages inspected; no independent full-document appraisal · Lot decisions; attribute and variable plans; purpose boundary
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Pooling answers a question about the combined material
Eurachem’s sampling-uncertainty guide treats collection, preparation and analysis as one measurement process. Heterogeneity, sampling strategy, transportation, preservation and homogenization can contribute uncertainty. Combining portions can efficiently estimate a pooled concentration, but the result cannot identify which original portion had the highest value. Duplicate independent sampling can help assess variability; repeatedly measuring the same prepared aliquot mainly examines later analytical variation. The appropriate design depends on whether the decision concerns an average, a hot spot or individual units.[3]
A report should connect the result to the collected material
FDA’s 2026 Investigations Operations Manual requires official samples to be handled, identified, sealed and documented to preserve integrity and demonstrate chain of custody. This is the record connecting collection, transfers and examination. Useful consumer-facing report details include the product and lot identity, collection date, submitted condition and laboratory receipt information. A certificate whose sample cannot be connected to the advertised batch has a different evidential value from a documented result for that batch.[4]
A factory swab is not a measured serving
FDA distinguishes product, environmental and emergency-response sampling. Environmental samples investigate surroundings that may contribute to contamination; product samples examine food or ingredients. Neither a positive surface swab nor a negative selected product sample independently quantifies every consumer’s exposure. Likewise, a targeted investigation should not be interpreted as a simple prevalence survey of all products in a category. Read the sampling purpose before interpreting the proportion of positive tests.[5]
In one contamination study, sampling dominated the variation
A 2006 study examined aflatoxin testing in 20 suspected hazelnut lots using repeated independently drawn samples. Under its specified collection, grinding and analytical workflow, sampling accounted for about 99% of total variance at the contamination level examined. Preparation and instrument analysis contributed much smaller shares. This concrete example shows why a more precise instrument cannot repair an unrepresentative collected sample. The percentage belongs to that product, contamination pattern and workflow; it is not a universal sampling fraction, a current prevalence estimate or a home-testing recipe.[7]
Passing a sampling plan is not inspecting every unit
NIST explains lot acceptance sampling as a decision based on a random sample, with plans using attributes or measured variables. It occupies a practical position between no inspection and inspecting everything. Its main purpose is deciding whether a lot is likely acceptable. Different plans can have different probabilities of accepting problematic lots or rejecting acceptable ones. A claim that a batch passed should identify the applicable plan and requirements rather than imply that every unit was individually tested.[6]
References
- BAM chapter 1: food sampling and preparation
Technical or regulatory guidance
Andrews and Hammack / FDA · Content/publication date: 2022
Accessed: 2026-10-13 UTC · Selected original PDF microbiological sampling and handling passages inspected
Cited sections: Introduction; collection, transport, receipt and sample identity - Guidance for choosing a sampling design, QA/G-5S
Technical or regulatory guidance
EPA · Content/publication date: 2002
Accessed: 2026-10-13 UTC · Selected original official guidance passages inspected; no independent full-document appraisal
Cited sections: Sections 2.3–2.4 and 4: probabilistic versus judgmental inferenceFinal technical guidance, used for statistical principles rather than current contaminant policy
- Measurement uncertainty arising from sampling, second edition
Technical or regulatory guidance
Eurachem / EUROLAB / CITAC / Nordtest / AMC · Content/publication date: 2019
Accessed: 2026-10-13 UTC · Selected original PDF sampling-uncertainty and design passages inspected
Cited sections: Table 1; sections 6 and 9; sampling, preparation and duplicate designs - 2026 IOM chapter 4: sampling
Technical or regulatory guidance
FDA · Content/publication date: 2026
Accessed: 2026-10-13 UTC · Selected original PDF official sample-integrity passages inspected
Cited sections: Sample validity; identity, sealing, documentation and chain of custody - Sampling to protect the food supply
Technical or regulatory guidance
FDA · Content/publication date: not stated
Accessed: 2026-10-13 UTC · Selected original official guidance passages inspected; no independent full-document appraisal
Cited sections: Types of sampling: product, environmental and emergency response - What is acceptance sampling?
Technical or regulatory guidance
NIST / SEMATECH · Content/publication date: not stated
Accessed: 2026-10-13 UTC · Selected original official guidance passages inspected; no independent full-document appraisal
Cited sections: Lot decisions; attribute and variable plans; purpose boundary - Sampling hazelnuts for aflatoxin: uncertainty associated with sampling, sample preparation, and analysis
Original sampling variance study
Özay and colleagues · Content/publication date: 2006
Accessed: 2026-10-20 UTC · Full original indexed abstract and bibliography read; publisher full text unavailable
Cited sections: 20 suspected lots, repeated independent samples and partitioned sampling/preparation/analytical varianceHistorical product-specific variance estimate; not a universal percentage, current contamination rate or consumer sampling protocol.
Scope & review
- Population
- General readers comparing product claims and analytical reports; not an individual exposure assessment
- Exposure or intervention
- Representative sampling
- Comparison
- Source-specific alternative products and matched/control conditions
- Outcomes
- Meaning of testing, interpretation of claims and limits on generalization
- Timeframe
- Report-specific measurements and dated standards; no guarantee for untested batches or future products
- Jurisdiction
- Selected US agency sources, dated professional guidance and original international research; jurisdiction and eligibility are source-specific
Selected-source educational explanation. Official guidance, original study findings and editorial interpretation are distinguished. No exhaustive search, independent systematic synthesis or qualified medical review has been completed.
Sources checked on their individual access dates. Selected sources were checked; ongoing surveillance and qualified review have separate statuses on the coverage page.
Revision history
- 2026-10-13 UTC: Added a selectively sourced product-testing explanation; qualified review remains pending.
- 2026-10-20 UTC: Added bounded selected-source explanations and practical distinctions after full-body overlap audit; preserved previous content and citations. Qualified review remains pending.
What this article still leaves open
- Does the report identify the precise lot and selection process needed for its advertised conclusion?
- How much of the uncertainty comes from sampling rather than analysis for the relevant product?
Suggest a correction
Submit a content correction to the editorial queue, or download a local note. Leave out personal medical details. Submitted text is stored privately with a browser receipt session and is visible to this browser and the editor. Clearing cookies removes access to your submission history. This is not an account or a medical service.
Your submitted corrections for this article
Explore the topic map
Browse the health domains connected to this article. The map also identifies topics whose articles remain unwritten.