‘Significant’

Photo: Florian Fisch

An energy specialist recently warned in the Swiss newspaper Tages-Anzeiger that the AI boom is going to cause a “significant spike in electricity consumption”. What he meant was: these computing centres are going to need a lot more energy. This is how many people use the word ‘significant’ in their everyday lives. But scientific studies treat the word differently. For example: “Fluvoxamine showed a significant reduction in fatigue”. This means that a statistical analysis was carried out, proving that the drug in question almost certainly makes a difference.

Whether that difference is going to help sick people in their everyday lives, or whether they can even tell the difference (which would make it ‘significant’ in colloquial terms) is irrelevant to its statistical significance. The article in question states the following later on about Fluvoxamine’s impact on the Fatigue Severity Scale: “The estimated mean differences in FSS score were clinically meaningful in magnitude”. This ‘meaningfulness’ is something that we have to bear in mind, as a leading doctor at the Hôpital du Jura told the newspaper Le Temps anonymously. This is because, “as a general rule, when study results are trumpeted out, a significant helping of caution is appropriate”.