KEY TAKEAWAYS

• Confusing the meaning of ‘significance’ in scientific research can lead to misinterpretation of findings.
• Differentiating between statistical significance and scientific importance is key when communicating scientific research.

Statistical significance and scientific importance are key aspects when communicating scientific research, but are often confused. This can prompt misleading interpretations of findings. An article by Professor Jane E Miller in Open Access Government explores the issue and provides guidance on improving clarity in scientific communication.

Statistical significance versus scientific importance

Prof. Miller suggests that the confusion of these terms is likely due to differences in the meaning of ‘significance’ in statistics versus everyday use, where it can mean ‘big’ or ‘important’. ‘Statistical significance’ means that a statistical test (measured by a P value) found evidence of an effect based on sample data, ie, that the result is unlikely to be due to random chance. Going further, Prof. Miller notes that for a result to have ‘scientific significance’, it must have real-world applications, such as clinical or educational relevance, or be useful for informing a decision, such as the design of an intervention.

How can scientific writers effectively communicate the significance of their results?

Prof. Miller suggests that scientific writers should explain scientific importance before statistical significance, to ensure that readers consider all aspects. Whether the significance being described is scientific or statistical should always be clarified.

“Prof. Miller suggests that scientific writers should explain scientific importance before statistical significance, to ensure that readers consider all aspects.”

Prof. Miller outlines some key dimensions to include when communicating scientific importance:

• the size and direction (eg, “positive” or “inverse”) of the effect
• the study sample (“when, where, who”), and if the results of the study can be generalised to the population
• other factors that may explain the association (ie, confounding)
• whether causality can be inferred from the research
• whether the observed outcomes are sustainable.

Next, communicating statistical significance must start with identifying the target audience. Prof. Miller advises paraphrasing ‘statistical significance’ and explaining the likelihood of the observed results occurring by chance when writing for lay audiences, and considering replacing ‘statistically significant’ with ‘statistically discernible’ for statistically trained audiences.

Prof. Miller concludes that by applying these principles, statistical communication in science can be improved.

————————————————–