Harmonized average of ranks to get the sharpest talent

When making employment selections with multiple assessors with different values It is easier to get sharp and talented by selecting by harmonic mean of rank than by selecting by arithmetic mean of rank. #sharp and talented

• The value rank, which is non-linear, has a significant impact.

  • Because of the high concentration of people around the average, the slightest difference in ability will cause a significant drop in the rankings.

• arithmetic mean

  • The additive mean places an implicit process where the difference between the 1st and 10th ranks is equal to the difference between the 10th and 19th ranks.
  • Assuming there are 100 applicants, person X who assesses that assessor A is ranked #1 and B is ranked #100, and person Y who assesses that both A and B are ranked #50, would judge Y to be superior.
  • (10, 100) and (55, 55) are equivalent.
  • Those who have sharp abilities and are highly valued only by some assessors will fall.
  • It would take a mediocre person that all assessors would consider "better than average".

• harmonic mean

  • In the case of the harmonic mean, the difference between first and second place is greater than the difference between second and last place.
  • (10, 100) and (18, 18) are about equal.
  • This results in a neat evaluation of sharp talent.

It would be interesting to compare which team performs better in Average orientation is the worst environment, when selected by additive average or when selected by harmonic average. Perhaps the team selected by harmonic mean is higher. #Experiments required

2024-05-03

• I noticed Reciprocal Rank Fusion.

This page is auto-translated from /nishio/尖った人材を取るためには順位の調和平均 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.