We use Cohen's kappa to measure the agreement between two observers who assess disease severity using a categorical scale of measurement.
The closer Cohen's kappa is to one, the poorer the agreement between two observers.
Cohen's kappa used to measure the agreement between two observers will increase in value if the scale of measurement (eg of disease severity) has more categories.
Cohen's kappa is calculated by comparing the frequencies with which two observers agree with one another with the frequencies of disagreement between the observers.
A weighted kappa is used to measure the agreement between two measurements of a numerical variable on a set of individuals.