The normal distribution is a standard reference for statistical probability because it closely represents many natural phenomena. When a set of variables follows normal distribution, the resulting graph is bell-shaped and symmetrical, with most data in the middle of the curve. The mean is the highest point on the graph, with 95% of data falling within two standard deviation of this point. However, while many things are normally distributed (e.g. average height within a human population,) most business processes do not. This is because results are tightly controlled by management and engineering. Descriptive statistics summarize and describe data, making inferences about data samples (like the average age of mothers to whom a birth certificate is issued in a given state.) Because data does not always follow normal distribution, it is necessary to determine the shape of the graphed data before performing analysis and summary involved in descriptive statistics. In statistical samples where results are highly controlled (or in the case of physical processes that allow for a narrow range of outcomes due to natural laws,) descriptive statistics cannot be computed without first determining the shape of data distribution.
William. (2004, September 1). Normal Distribution.