Discrepancy

The kappa is 0.455, indicating a moderate agreement.

You may say, "That's not acceptable. We should aim at something around 0.8 or above."

There are, however, two layers of consideration that may skip your eyes.

Let me bring you through the superficial one first.

***************************

The reality is simple: Not all examiners are the same.

If we take this as a de facto assumption, the inevitable consequence is we should deliberately increase the discrepancy of scoring for each and every pair of examiner. (And, I can assure you, it is one of the factors that determine how examiners pair up with each other.)

I can hear you gasp.

But true. It is logical to give an examiner who keeps giving a low score a partner from the other side of the spectrum.

Or, to put it simply, you may prefer to see two doves in the examination, but you have to consider other classmates who would then have to see a pair of hawks.