DQ 2 : What are misrepresentations that can occur while reading other’s work and using its data?

The misrepresentations worth writing about are the ones committed by careful people. Reading only the abstract, which is written by authors who want the study read and routinely states the result more confidently than the discussion does. Citing a secondary source's summary of a study rather than the study, which propagates whatever the summariser got wrong. Quoting a finding from a population unlike yours as though it applied directly. Presenting an association as a cause because the sentence reads better that way. Reporting significance without magnitude. Ignoring the limitations section, where the authors have already told you what their study cannot support. None of these is dishonest and all of them produce a claim the original data do not carry, which is what misrepresentation means in practice. Each produces a citation that is technically accurate and materially wrong, and each is common enough that you have probably done one this month.

The middle question — when is data a useful source of evidence — is the defence, so answer it as criteria rather than as a feeling. Data is useful when the population is close enough to yours that transfer is arguable, when the outcome is one you can also measure, when the methods are reported fully enough to appraise, when the effect is reported with its uncertainty, and when you have read past the abstract to the limitations. Then the third question: yes, statistics can mislead without a single false number. Relative risk without absolute risk turns a change from two in ten thousand to three in ten thousand into a fifty per cent increase. A truncated y-axis makes a trivial difference look dramatic. Subgroup analyses run until one is significant produce findings that will not replicate. Composite outcomes let a hospitalisation effect be reported as if it touched mortality. Name two and say what you would report instead.