LogicalMANIPULATIVE

False Precision

What it is

Reporting a figure to more digits than the method could justify — a death toll to the last person, a share of jobs to the percentage point, a phone “at” a ballot box — so that specificity is read as accuracy and the uncertainty disappears.

How it works

Precision is the number of digits; accuracy is closeness to the truth; they are independent, and audiences treat the first as evidence of the second. Jerez-Fernandez, Angulo and Oppenheimer showed that people read a precise estimate as a sign that its source is confident and competent, and Janiszewski and Uy found that precise numbers anchor more firmly than round ones. A poll of 400 people has a margin of about five points, so reporting 52.3 percent is reporting noise to the decimal. Worked: a survey estimate of 654,965 deaths with an interval from 392,979 to 942,636 is honestly “somewhere around 400,000 to 900,000”; quoting the central figure to the last digit converts a wide interval into a fact. The device also converts a method's limits into certainty: a phone-location record with an error radius of tens of meters cannot distinguish a person at a ballot drop box from one on the sidewalk beside it, yet the record can be presented as a visit. Huff's example was an average of 7.831 hours of sleep from a self-reported survey. The honest practice is to round to the digits the method supports and print the interval; the manipulation is to print the digits because the audience will mistake them for knowledge.

Real-world examples

  • •The 2006 Lancet cluster survey of mortality after the invasion of Iraq estimated 654,965 excess deaths, with a 95 percent interval from 392,979 to 942,636; opponents of the war cited the central figure to the last digit while the interval spanned more than half a million.
  • •The 2022 film 2,000 Mules claimed that commercial cellphone geolocation data identified individuals repeatedly visiting ballot drop boxes in 2020; such data cannot place a phone within the few meters needed to distinguish a drop box from the street, and in 2024 the distributor, Salem Media, withdrew the film and apologized to a Georgia voter it had depicted.
  • •Frey and Osborne's widely quoted estimate that 47 percent of US employment was at high risk of automation rested on subjective classifications of seventy occupations extrapolated by a model; the two digits travelled far further than the method behind them.
  • •Election-night and campaign coverage reports poll movements of a fraction of a point between surveys whose margins of error are several points, presenting sampling noise as a trend.
  • •Consumer devices report calories burned and sleep stages to the unit while validation studies show errors of tens of percent; the digits are design choices, not measurements.

Ethical guidelines

  • ●Round to the digits your method supports and print the interval beside any estimate; a figure without an uncertainty is a claim to certainty you do not have.
  • ●Do not report a central estimate to more digits than its interval is wide — 654,965 with a range of half a million is “roughly 650,000, and anywhere from 400,000 to 900,000”.
  • ●State what a measurement can and cannot resolve (a location within thirty meters, a poll within four points) before presenting a conclusion that requires finer resolution.
  • ●Do not convert units or scale figures in ways that create digits that were never measured.

How to defend against it

  • ►Ask for the margin of error or the interval, then decide how many digits are real; if the interval is not given, assume the last one or two digits are decoration.
  • ►Ask how the number was produced — a survey, a model, a sensor — and what the smallest difference that method can actually detect is; a claim finer than that is not from the method.
  • ►Be more, not less, skeptical of oddly exact figures from sources with an interest in the conclusion; the exactness is a persuasion cue and it is known to work.
  • ►Round the number yourself before reasoning with it — “about 650,000”, “roughly half” — which strips the false confidence the digits carry.
  • ►For location, tracking and matching claims, ask about resolution: within how many meters, with what error rate, and how many innocent matches that error rate would produce at scale.

From the Defense Playbook

Calibration Practicetakes practice

Regularly attach a probability to your predictions, record them, and score them against what happened, so that "I am sure" comes to mean something and you can recognize false certainty in others.

Ask for Absolute Numbersseconds

When someone quotes a relative change ("cuts your risk by 50 percent", "doubles the danger"), ask what the risk was before and after in plain counts out of the same number of people, because the same fact sounds enormous as a ratio and modest as a difference.

The Base-Rate Checkminutes

Before accepting what a test result, a profile, or a striking detail seems to prove, ask how common the thing is in the first place, because evidence that sounds accurate can still be wrong most of the time when the underlying condition is rare.

The Toulmin Checkminutes

Test any single argument by asking Toulmin's six questions of it: what is the claim, what are the grounds, what warrant connects them, what backs the warrant, how strongly is the claim qualified, and under what conditions would it fail?

Decide Metrics Before Testingminutes

Before an A/B test or pilot starts, write down the primary metric, the harm metrics that must not get worse, the sample size or stopping rule, and the planned analysis, so that the result can tell you something you did not already want to hear.

Every playbook entry states how strong its evidence is and when not to use it. Browse the full playbook.

References

  1. Jerez-Fernandez, A., Angulo, A. N., & Oppenheimer, D. M. (2014). Show me the numbers: Precision as a cue to others' confidence. Psychological Science, 25(2), 633-635
    Precise estimates are read as signals of the source's confidence and competence.
  2. Janiszewski, C., & Uy, D. (2008). Precision of the anchor influences the amount of adjustment. Psychological Science, 19(2), 121-127
    Precise numbers anchor judgments more strongly than round ones.
  3. Burnham, G., Lafta, R., Doocy, S., & Roberts, L. (2006). Mortality after the 2003 invasion of Iraq: A cross-sectional cluster sample survey. The Lancet, 368(9545), 1421-1428
    The 654,965 central estimate and its 392,979-942,636 interval.
  4. Frey, C. B., & Osborne, M. A. (2017). The future of employment: How susceptible are jobs to computerisation?. Technological Forecasting and Social Change, 114, 254-280
    The 47 percent estimate and the subjective classification method behind it.
  5. Huff, D. (1954). How to Lie with Statistics. W. W. Norton
    The 7.831-hours example and the general warning that decimals are not evidence.
Last reviewed
Suggest a correction

Detect False Precision in any text

Paste any message, email, or article into our free Manipulation Detector to see if False Precision or other techniques are being used on you.

Try the Detector FreeCreate Free Account