The McNemar Index can be used to identify bias in the disagreements between 2 observers making binary observations. This is a simple index based on the premise that random disagreements should have an equal occurrence.
Observer 2
Observer 1
Yes
No
Yes
agree 1
disagree 1
No
disagree 2
agree 2
McNemar index =
= ((disagree 1) - (disagree 2)) / ((disagree 1) + (disagree 2))
Interpretation:
• minimum index: -1
• maximum index: 1
• The closer the index is to 0 the less bias there is between the observers.
• The higher the absolute value for the index the greater the differences between the observers and the greater the bias.
• Bias in observer 1 results in a positive index while bias in observer 2 results in a negative index.
Limitations:
• The number of observations that agree are not considered.
• The index can reach a maximal value with a disagreement pair of 1 and 0.
Continue Reading