Correlation analysis tries to establish whether there is any relationship between two values in one sample or between two different samples. If a connection is found, then it is necessary to find out whether it is accompanied by an increase in any one indicator by an increase or decrease in another.
Instructions
Step 1
Decide between which indicators you need to carry out a correlation analysis. However, keep in mind that it will help you determine whether it is possible to predict certain values of one value, knowing the magnitude of another. For this purpose, you can use 2 different methods: the parametric method of calculating the coefficient r (Brave-Pearson) and the determination of the correlation coefficient rs (Spearman's ranks), which is applied to ordinal data and is nonparametric.
Step 2
Determine the correlation coefficient - a value that can range from one to -1. Moreover, in the case of a positive correlation, this coefficient will be equal to plus one, and in the case of a negative correlation, it will be minus one. You can plot the correspondence of the values you want to analyze. On it you will get a certain straight line passing through the points of intersection of the indicators of each pair of these values. In turn, if these points (reflecting values) do not line up in a straight line and form a "cloud", then the correlation coefficient in absolute value will be less than one, and as this cloud is rounded, it will approach zero. If the correlation coefficient is equal to 0, it means that both variables are completely independent of each other.
Step 3
Draw conclusions about the relationship between the variables. At the same time, pay great attention to the sample size: the larger it is, the more reliable the value of the obtained correlation analysis coefficient will be. There are special tables that contain the critical values of the correlation coefficient according to Brave-Pearson and Spearman. These indicators can be used to determine a different number of degrees of freedom (it is equal to the number of pairs minus two). Only in the case when the correlation coefficients are greater than these critical values, they will be considered reliable.
