Variance homogeneity, the most influential assumption in regression analysis

In regression analysis, a special and most attention must be paid on assumption validating the model obtained from a sample drawn from a population. The ultimate objective of any regression model is in fact, making some inference on the population. To use the model effectively and discover (make inference) the unknown population, the model should have passed through many tests (verifications) to make sure that it is the best model for the set of data representing. These tests are in fact the assumptions on regression model or assumptions in regression analysis. Among assumptions, the variance of the response should be constant or what we call homogeneity of variance, and if this assumption is not met then the model will suffer from what we call heterogeneity or the variance of the response is not constant.[5] The paper discussing the conceptual approaches to detect the violation of assumptions in general, and steps that to be taken to make the model acceptable and reliable to make inferences on the population. There is no certain application or any type of calculations. The aim of this paper is to the importance of these assumptions and enhance the sense of good understanding of things will lead to better utility and rich information