Regression dilution is the term which is applied when a covariate in the model cannot be measured directly and instead of that a related observed value must be used in analysis. In common, if the model is correctly specified in the terms of the 'true' covariate, then an equivalent form of the model with a easy error structure will not hold for observed values. In such type of cases, ignoring the measured values will lead to the biased estimates of the parameters in the model. It is often also referred to as the errors in variables problem.