4 ms·
The formal terminology is “selection induced negative correlation”. If a quality score is the sum of two factors, those two factors will tend to be negatively c
by janalsncm 2y ago
The formal terminology is “selection induced negative correlation”. If a quality score is the sum of two factors, those two factors will tend to be negatively correlated.
Mathematically a trivial example is the equation 1=x+y, where 1 represents some cutoff and could be any value. Clearly x and y are inversely correlated.
- fermisea 2y agoAlso a type of collider bias in causal inference, which generates all sorts of Simpsons paradoxes