Example: 53 examined values were obtained with a reference examination procedure, 106 were obtained by Examination procedure 1 and with Examination procedure 2. Examination procedure 1 resulted in examined values with the same proportion distribution as for the Reference examination procedure, and the examined values have thus no examination bias. The 106 examined values with Examination procedure 2 have a different distribution compared to that of the reference nominal property values, and the examined values thus have an examination bias.

Possible examined values			Examined values
	Reference examination procedure	Examination procedure 1	Examination procedure 2
A	21	42	2
B			40
C	18	36	2
D			36
E	12	24	10
F			5
G	2	4	11
Total number of examined values	53	106	106

One estimate of examination bias for nominal properties is the Bhattacharyya distance, which measures the similarity of two nominal data sets. It equals zero when two proportion distributions are identical as is the case when comparing Examination procedure 1 and Reference examination procedure. There is an examination bias between Examination procedure 2 and Reference examination procedure, because the proportion distributions are different.

Notes:

1. Examination bias is a quantity.
2. If the proportion distributions of the reference nominal property values and the examined values are the same, there is no examination bias for the examining system. If the distributions differ there is an examination bias.
3. In some publications on analytical chemistry, "examination bias" is termed "lack of reliability", but this is not recommended here.