Instead of delving into generalities, correlation analysis helps provide the statistician with a certain precise measures that help account for the relationship between the variables. There are many ways to measure this. One of the most popular is known as the scatter diagram, which is also the most simplistic in nature in view of its representation. This basically is the first step towards the determination of the relationship between two variables, which is done through the plotting of data on a diagram.
However, this best fit line would not show the exact nature of the relationship, by either under estimating or over estimating the relationship between two variables. Which makes it not a suitable method as far as relying over the results are concerned.
To then calculate the nature of the variance or dependence between the variables, the coefficient of correlation is then applied. This measures the strength of the relationship between the two variables. However, there is a flaw in this design. To make sure the result that comes out is a correct representation of the data one needs to assure that the coefficient is not affected by the units of the variables used. Secondly, this method can only be applicable in a scenario where only two variables are being tested against each other. Also, the end result does not show that whether a change in the amount of one can cause a change in the amount of the other variable, that is, one cannot precisely determine the cause and effect. The only thing that one can determine is that there is a relationship or not existent between two variables.