ABSTRACT
To the extent that this assumption is violated, it is said that the model suffers from heteroscedasticity; that is, the variations in the error terms around the regression line are not all the same for all values of X. This problem is most likely to arise in the use of cross-sectional data as opposed, for example, to time series data. The reason for this phenomenon is that heteroscedasticity is likely to be caused by disparity in sizes. With time series data pertaining to, for example, annual sales for a single enterprise, the data will not exhibit wide fluctuations from year to year. However, if we are using cross-sectional data we would likely find that sales varied considerably from smaller to larger firms in our data set.
