6.3. TECHNIQUES FOR MEASURING CORRELATION

Three important tools used to study correlation are scatter diagrams, Karl Pearson’s coefficient of correlation, and Spearman’s rank correlation.

In this section, we introduce the three main tools for measuring correlation. These tools help us understand how two variables relate to each other. The scatter diagram visually represents the relationship by plotting points on a graph. Karl Pearson’s coefficient provides a numerical value indicating the strength and direction of the linear relationship. Spearman’s rank correlation is useful when the data does not fit a linear model or when the variables can only be ranked.

Imagine you're trying to see how studying affects test scores. You plot the hours spent studying against the test scores on a scatter diagram. Each point on the graph represents a student. If most points trend upward, it suggests that more study hours might lead to higher scores, showing a correlation.

A scatter diagram visually presents the nature of association without giving any specific numerical value. In this technique, the values of the two variables are plotted as points on graph paper.

Scatter diagrams allow us to visually inspect the relationship between two variables. By plotting the data points, we can easily see trends, clusters, or whether there is any apparent correlation. If the points are tightly clustered around a line, it indicates a strong correlation, while widely scattered points suggest a weak correlation.

Think of a scatter diagram like a family photo. If everyone is standing next to each other, it suggests a close relationship. However, if some people are scattered far apart, it indicates loose connections. Just like in the photo, tight clusters on a scatter diagram show strong relationships.

This is also known as product moment correlation coefficient or simple correlation coefficient. It gives a precise numerical value of the degree of linear relationship between two variables.

Karl Pearson’s coefficient quantifies the correlation between two variables, producing a value between -1 and 1. A value closer to 1 indicates a strong positive correlation (as one variable increases, so does the other), while a value closer to -1 indicates a strong negative correlation (as one increases, the other decreases). A value of 0 suggests no linear correlation. This coefficient is useful for predicting one variable based on another if their relationship is linear.

If you think about a rubber band, stretching it relates to its length. If the correlation coefficient were 1, it means every time the band stretches, it increases in length proportionally. If it were -1, the more you stretch it, the less it shows its original shape, indicating an inverse relationship.

Spearman’s rank correlation was developed by the British psychologist C.E. Spearman. It is used in situations where we cannot measure variables precisely.

Spearman’s rank correlation is a method for assessing the association between two variables that do not require precise measurements. Instead of using actual values, we rank the data. This method is especially useful when dealing with ordinal data or when the relationship appears to be non-linear. It focuses on the ranks rather than the data values themselves, making it more robust against outliers.

Imagine you're judging a baking competition, and you rank contestants based on their cakes. You may not accurately measure taste or texture, but you can rank the best to worst. Spearman’s method allows us to dig deeper into the rankings to understand how one judge's opinion correlates with another's, even if they're looking at different aspects.

Like the Pearsonian Coefficient of correlation, it lies between 1 and –1. However, generally, it is not as accurate as the ordinary method.

The properties of the correlation coefficient highlight that it is a pure number, without units, and represents the degree of association between two variables. It behaves similarly to Pearson's coefficient but can provide less accuracy because it only considers ordinal rankings. Furthermore, Spearman’s correlation is not affected by extreme values, making it useful in real-world scenarios where data might be skewed.

Imagine a group of athletes competing in various sports. Even if one runner is exceptionally fast (an outlier), it won't greatly affect the rankings when considering all athletes together. Thus, when using Spearman's ranking system, we can still determine who generally performs well across sports without the distraction of outliers.

--