A bias claim about drinking water data is a claim about the measurement system before it is a claim about the water. Circle of Blue has published a piece, dated October 23, 2026, reporting on a new study that suggests PFAS contamination in drinking water may be underestimated. The item originally appeared on Inside Climate News and is listed under the headline PFAS Contamination in Drinking Water May Be Underestimated, New Study Suggests. The summary opens with the question of what happens when concentrations "fall below laboratory reporting limits," which is the point where a reported low value and a true low value can separate.
What the summary can and cannot support
The text available here is a title and an opening paragraph. It does not give the study design, the sample count, the analytical methods, or the size of any adjustment, and nothing in this article should be read as a figure from the study. What a reader can do is check the claim against the records that would show bias if it exists. A stated direction without a size is still a claim, but it is not yet a number that belongs in a file.
Three records that show whether a result can carry a bias
Field blanks are clean water carried through the sampling process, opened and handled like a real sample. A blank that returns a detection points to contamination from containers, sampling equipment, or handling rather than from the tap. A clean blank narrows that explanation but does not rule out losses during storage or transport.
Spike recovery adds a known amount of the target compounds to a sample before extraction. The percentage recovered shows how much of each compound the method actually captures in that matrix. Low recovery points to losses during preparation or analysis, which would push reported values down. High recovery in clean water does not guarantee recovery in a dirty one, so matrix-matched spikes matter.
Duplicate results show precision. Two bottles drawn from the same point should agree within limits set by the method. Wide disagreement at low concentrations usually means the reported value sits close to the method's noise, which is where reporting limits do their work.
Reporting limits are a data decision, not only a measurement
A reporting limit is the lowest concentration a laboratory will report as a quantified value, and it sits above the method detection limit for that reason. Below it, results are typically flagged, reported as estimates, or reported as nondetect. How those flags feed into averages, trends, or compliance comparisons is a choice made in the data file. If a study argues that low-end values are underestimated, the useful question is how its authors treated flagged results, and whether the same treatment appears in the data your own laboratory sends you.
Reading a bias claim against your own file
Before a PFAS bias claim enters a utility, consultant, or dealer file, log three things: the source and its date, the quality control records the claim relies on or a note that those records were not available, and the specific compounds and matrices the claim addresses. A general statement that results may be low does not identify which results are low. Your own laboratory package should show blank, spike, and duplicate data for the batch that produced each result. If it does not, that gap is more useful to flag than the headline.
What to do with the claim now
Do not change a compliance record, a treatment decision, or a customer notice on the strength of a headline. Record the claim as an open question, request the study methods from the authors or the publisher, and compare them with your own quality control data. Homeowners with a home test should read the report in the same way, checking the reporting limit printed on the result and the list of compounds tested. Questions about personal exposure or health belong with a physician or a state health agency, not with a sample sheet.
Where this leaves the record
The study may show that low values are biased in a particular way, or it may show a narrower effect than the headline suggests. The available source supports neither conclusion on its own. Keep the claim in the file with its date and its open questions, and keep the laboratory's blanks, spikes, and duplicates beside it so that any later correction can be traced to a record rather than to a memory of the headline.