**IMPORTANT: The methods covered in this section on correlation are only applicable for LINEAR relationships.**

**CO-4:**Distinguish among different measure up scales, pick the ideal descriptive and inferential statistical methods based upon these distinctions, and also interpret the results.

You are watching: The strength of the relationship between two quantitative variables can be measured by

**LO 4.21:**For a data analysis situation including two variables, recognize the ideal graphical display(s) and/or number measures(s) that have to be provided to summary the data.

## Introduction

So far we have visualized relationships between two quantitative variables making use of scatterplots, and described the overall pattern of a connection by considering the direction, form, and also strength. We noted that assessing the strength of a relationship simply by looking in ~ the scatterplot is quite difficult, and therefore we should supplement the scatterplot v some sort of number measure that will help us assess the strength.

In this part, we will certainly restrict our fist to the **special case of relationships that have a linear form**, since they room quite common and reasonably simple to detect. Much more importantly, over there exists a number measure the assesses the stamin of the **linear** relationship in between two quantitative variables v which we can supplement the scatterplot. We will present this numerical measure here and also discuss that in detail.

Even though from this point on we are going to focus only on **linear** relationships, the is essential to psychic that **not every relationship between two quantitative variables has actually a straight form.** We have actually seen several examples of relationship that room not linear. The statistical devices that will certainly be introduced right here are **appropriate only for assessing linear relationships,** and together we will certainly see, once they are provided in nonlinear situations, these tools have the right to lead come errors in reasoning.

Let’s start with a inspiring example. Think about the complying with two scatterplots.

We can see the in both cases, the direction the the relationship is **positive** and the form of the relationship is **linear**. What about the strength? Recall that the toughness of a connection is the extent to i beg your pardon the data monitor its form.

**Learn by Doing:**toughness of Correlation

The objective of this example was to illustrate just how assessing the toughness of the **linear** relationship from a scatterplot alone is problematic, since our judgment might be affected by the scale on i m sorry the values are plotted. This example, therefore, offers a motivation for the **need **to complement the scatterplot through a **numerical measure** that will **measure the strength** of the **linear** relationship between two quantitative variables.

## The Correlation Coefficient — *r*

**LO 4.26:**Explain the limitations of Pearson’s correlation coefficient (

*r*) together a measure of the association between two quantitative variables.

**LO 4.27:**In the special case of a straight relationship, interpret Pearson’s correlation coefficient (

*r*) in context.

The number measure that assesses the toughness of a **linear** partnership is referred to as the **correlation coefficient**, and also is denoted by r. We will:

**Correlation Coefficient:**The correlation coefficient (r) is a numerical measure up that actions the

**strength**and

**direction**of a

**linear**relationship in between two quantitative variables.

**Calculation: **r is calculated using the complying with formula:

However, the calculate of the correlation (r) is no the ebrickandmortarphilly.comasis of this course. Us will usage a statistics package to calculate r for us, and the **ebrickandmortarphilly.comasis **of this course will certainly be on the **interpretation** of that is value.

## Interpretation

Once we obtain the worth of r, that interpretation v respect to the strength of **linear** relationships is rather simple, together these photos illustrate:

In bespeak to obtain a better sense for how the worth of *r* relates to the toughness of the **linear** relationship, take a look at the complying with applets.

**Interactive Applets:**Correlation

If you will be utilizing correlation frequently in your research, I highly urge girlfriend to review the following more detailed discussion of correlation.

**(Optional) exterior Reading:**Correlation Coefficients (≈ 2700 words)

Now that we recognize the use of *r* as a numerical measure up for assessing the direction and strength of **linear** relationships in between quantitative variables, we will look in ~ a few examples.

## EXAMPLE: Highway authorize Visibility

Earlier, we supplied the scatterplot below to discover a **negative linear** relationship in between the age of a driver and also the maximum distance at i m sorry a highway sign was legible. What around the toughness of the relationship? It turns out the the correlation in between the two variables is r = -0.793.

Since r Note that in both instances we supplemented the scatterplot through the correlation (r). Currently that we have actually the correlation (r), why carry out we still must look at a scatterplot when analyzing the relationship in between two quantitative variables?The **correlation** coefficient can **only** be taken as the **measure the the toughness of a straight relationship**, therefore we need the scatterplot come verify that the relationship indeed looks **linear**. This allude and its prestige will be clearer after we examine a couple of properties of r.

## Properties of *r*

We will now discuss and illustrate several essential properties that the correlation coefficient as a numerical measure of the toughness of a **linear** relationship.

**change the systems of measurement**of the explanatory change and/or the response variable, this has

**no impact on the correlation (r)**.

To show this, listed below are 2 versions of the scatterplot of the relationship in between sign legibility distance and also driver’s age:

The optimal scatterplot screens the initial data where the maximum ranges are measured **in feet**. The bottom scatterplot display screens the very same relationship, yet with maximum distances readjusted to **meters**. An alert that the Y-values have changed, yet the correlations are the same. This is an example of how transforming the devices of measure up of the an answer variable has actually no effect on r, yet as we indicated above, the same is true for changing the systems of the explanatory variable, or of both variables.

This can be a good place to comment the the correlation (r) is **“unitless”**. The is just a number.

**only measures the stamin of a linear relationship**between 2 variables.

**It ignores any kind of other type of relationship, no issue how strong it is.**For example, take into consideration the relationship in between the average fuel intake of steering a addressed distance in a car, and also the speed at i beg your pardon the automobile drives:

Our data define a fairly simple non-linear (sometimes referred to as curvilinear) relationship: the lot of fuel consumed decreases swiftly to a minimum for a vehicle driving 60 kilometers per hour, and also then increases slowly for speeds exceeding 60 kilometers per hour. The partnership is an extremely strong, together the monitorings seem to perfect fit the curve.

Although the connection is strong, the correlation r = -0.172 suggests a weak **linear** relationship. This provides sense considering that the data stops working to adhere carefully to a direct form:

**not**enough to recognize whether or not a relationship is linear. To view this, let’s think about the examine that check the effect of monetary incentives on the return price of questionnaires. Listed below is the scatterplot relating the percentage of participants that completed a survey to the monetary motivation that researchers promised to participants, in which we uncover a

**strong non-linear (sometimes called curvilinear) relationship:**

The partnership is non-linear (sometimes dubbed curvilinear), yet the correlation r = 0.876 is rather close to 1.

In the last two examples we have seen 2 very strong non-linear (sometimes dubbed curvilinear) relationships, one through a correlation close come 0, and also one through a correlation close to 1. Therefore, the correlation alone does not suggest whether a partnership is **linear** or not. The necessary principle here is:

**Always look at the data!**

**linear**relationship.

**Interactive Applet:**Correlation and Outliers

Hopefully, you’ve i found it the correlation decreasing as soon as you developed this kind of outlier, which **is no consistent **with the pattern of the relationship.

The next task will present you how an outlier that **is consistent** with the direction that the direct relationship in reality strengthens it.

See more: Does Miley Cyrus Fake Teeth ? Veneers? Teeth Surgery? Does Miley Cyrus Has Fake Teeth

**Learn by Doing:**Correlation and Outliers (Software)

In the ahead activity, us saw an example where there was a positive **linear** relationship in between the two variables, and also including the outlier just “strengthened” it. Think about the hypothetical data shown by the following scatterplot:

In this case, the low outlier gives an “illusion” that a positive **linear** relationship, vice versa, in reality, over there is no **linear** relationship between X and also Y.

Tagged as: situation QQ, CO-4, Correlation, Direction, Exploratory Data Analysis, straight Relationship, LO 4.21, LO 4.26, LO 4.27, numerical Measures, Pearson"s Correlation Coefficient, Strength

Previous

This material was adapted from the Carnegie Mellon university open discovering statistics course accessible at http://oli.cmu.edu and is license is granted under a creative Commons License. Other products used in this job are referenced when they appear.

**If girlfriend have uncovered these products helpful, DONATE by clicking the "MAKE A GIFT" link listed below or at the optimal of the page! The room of Biostatistics will use funds created by this Educational enhancement Fund particularly towards biostatistics education. **