Theme 8 - bad science and misuse of data Flashcards
give ways in which data can be misinterpreted unintentionally
-simpson’s paradox
-prosecutor’s fallacy
name ways how the media and advertising can make numbers twisted or confusing
-using misleading graphs
-using a misleading study design
what is simpson’s paradox?
where the sane set of data can appear to show opposite trends depending on how it’s grouped and typically occurs when aggregated data hides a confounding variable (usually one that significantly influences results)
Give an example of simpsons paradox
the low birth weight paradox:
-where underweight babies are SEEMINGLY more likely to survive low birth weight if the mother is a smoker compared to if she is not
-this is inaccurate because although smoking increases the risk of low birth weight there are other causes of low birthweight that are more likely to be fatal
what is prosecutors fallacy?
a statistical fallacy in which the probability of A given B is assumed to be the same as the probability of B given A
How can bar graphs be misleading?
-as when displaying the absolute magnitude of values for the associated categories, they do not always start at 0 and so the size of the bars end up looking misleading
-if the bar graph is being used for comparison where it has two sides, the axes may not be equal
give an example of a misleading study design
-when in 2007, Colgate were fined and prevented from running an advertisement that stated 80% of dentists recommended Colgate
-this is because they allowed dentists to name any number of toothpastes and although 80% did include Colgate, they also included many other brands
what us the ‘publish or perish’ conflict that affects scientists?
a scientific misconduct caused by pressure where scientists twist their findings to be more publishable to further advance their career and scientific success
give three consequences of scientific misconduct
-preventable diseases not getting the correct treatment leading to loss of life
-large number of wasted resources, time and money (bad as funding is often millions from tax payers)
-reputational risk and eroding of the public’s trust in science
give 2 scientific misconduct incidents
Alzheimer’s disease controversy:
-professor Sylvain Lesne published an article in 2006 claiming a specific amyloid B protein assembly in the brain impairs memory
-there was signs of fabrication and tampering to the results
Talking animals
-It was claimed animals can talk when evidence of koko the ape was released in clips of him talking
-instead it was just confirmation bias from scientist hearing what they want to hear