[Statistical approaches to evaluate agreement]

Reliability and agreement of measurement methods is a fundamental issue in health sciences which is not usually borne in mind. In this document the connotations of using measurement methods with error and the switchability among measurements from methods which disagree are highlighted. These implications are illustrated through examples showing up the confounding effect that measurement error can produce. Throughout the document several procedures to assess agreement and to identify the error sources are suggested. These procedures are classified according to the sort of data, quantitative or qualitative data, as well as the way of agreement is assessed, in an aggregate way by means a values or in a disaggregate way analysing separately the error sources. By means of these procedures is showed that frequently used approaches to assess agreement as the averages comparison, the correlation coefficient or the regression model appear as insufficient or inadequate approaches.