ALT For a disease present in 2% of the population using a screening test with 95% accuracy. Of 1,000 people, 20 have the disease (of which 19 test positive) and 1 tests negative (false negative). Of 980 people without the disease, 49 test positive (false positive) and 931 test negative. The total number of 'positive' results is 68, but only 19 of these are correct, so the chance you have the disease if you have received a positive result is actually only 28%. A 28% chance is higher than the 2% chance you had before taking the test, but for most diseases this isn't enough to start treatment and further investigation would be needed.
ALT How was the data collected? Experimental data: researchers collect data under controlled conditions to answer a specific question - experiment can be designed to compensate for potential biases recognised by the researchers Observational: researchers use pre-existing data which was collected for another reason - biases cannot be easily controlled for, and should be acknowledged
ALT Tile reads: How to answer the right question when people ask about certainty. 'are you certain its safe' When someone asks if scientists are certain that vaccines are safe, the emphasis is “safe” not “certain”. In this instance, talking about risks and benefits here is more useful than explaining how scientific certainty is rarely achieved or that nothing is 100% safe. Few things are 100% certain, but that's often not the real issue. Sense about Science logo. #EvidenceKnowHow
ALT Tile reads: risk of what? we will have better discussions about risk if we are specific about the outcomes. for example, discussions of 'risk of climate change' on X have referred to the following: 1. risk of sea levels rising due to climate change, 2. risk of biodiversity loss due to climate change, 3. risk of intense heat waves due to climate change
ALT Tile reads: These concepts of uncertainty are like the difference between a Victorian map with Central Africa marked "unexplored" (a ‘known unknown’ to European explorers), and medieval maps made before Europeans knew that the New World existed which show sea where the land should be (an 'unknown unknown' to European explorers). Kown knowns = things we know we know, known unknowns = things we know we dont know, unknown unknowns = things we dont know we dont know
ALT Photo of David Spiegelhalter, British statistician and a Fellow of Churchill College. Text reads: "Statisticians use standard levels of 'unlikely'. Commonly they use significant at the 5% level (sometimes written as p=0.05). In this case, a difference is said to be 'significant' because it has a less than 1 in 20 probability of occurring if all that is going on is chance".
ALT Image shows two graphs with the caption 'same average, different picture'. The mean is 2 on both pictures but the way the images are presented makes one seem more evenly distributed than the other
ALT Alt: Blue block with text which reads: "Vitamin A deficiency (VAD) causes blindness in around half a million people, mainly children, every year. Half of them die within 12 months of going blind and others die of diseases like malaria because VAD severely affects the immune system. It is prevalent in poor communities that rely on rice as their major food because their diets don’t contain the beta-carotene that the body needs to convert to vitamin A."
