Simple agreement or chance-corrected statistics can be selected according to the label type and study design. Low agreement often indicates ambiguous instructions, difficult cases, or subjective criteria rather than merely careless raters. Reporting the statistic and protocol helps readers assess label reliability.