Conditions to be met for Linear Regression Analysis

The mathematical and statistical assumptions on which the standard regression model is 

Because we do not have, at the moment, replicate measurements of y at x, we are not able to check conditions (2) to (4). With the above calculated values of y) we obtain the residual e,-  

In contrast with usual calibration (case B), modeling in trace analysis, and also of environmental relationships (case A), will probably fail to fulfil condition (2) because x is usually also subject to errors. Alternative linear models must then be considered. 

---