"In the 1970s, the University of California, Berkeley, faced allegations of discrimination against women based on admission rates, which showed a 44% acceptance for men and 35% for women. However, when analyzed by individual subjects, it became clear that more women were admitted in four of the six largest departments, suggesting a preference for female students."
"Simpson's paradox illustrates how trends can differ when data is divided into subgroups. This was first described by Karl Pearson in 1899 and later rediscovered by George Udny Yule. Edward Simpson's 1951 publication brought renewed attention to this counterintuitive statistical phenomenon."
"In 2021, data indicated that COVID-19 was nearly twice as deadly in Italy compared to China, despite every Italian age group having a higher chance of survival. This exemplifies how aggregate data can mislead interpretations when subgroup dynamics are not considered."
Statistics can yield surprising results, as demonstrated by the University of California, Berkeley's admission rates in the 1970s. Initially, data suggested discrimination against women, with a 44% admission rate for men compared to 35% for women. However, a deeper analysis revealed that in several departments, more women were admitted than men, indicating a preference for female students. This phenomenon is known as Simpson's paradox, where trends can change when data is segmented into subgroups. Similar patterns were observed in COVID-19 mortality rates between Italy and China.
Read at www.scientificamerican.com
Unable to calculate read time
Collection
[
|
...
]