What Is a Linear Association? How It Works and Why It Matters in Data Analysis

Key takeaways

A linear association describes a relationship between two variables that forms a straight-line pattern when graphed [1]. 

• Pearson correlation coefficient (r) measures the strength and direction of a linear association.

• A positive linear association occurs when both variables increase together, while a negative association occurs when one decreases as the other increases.

• You can go beyond linear association to use linear regression to make predictions on how changes to your independent variable(s) will affect your dependent variable.

Explore the types of linear associations, examples of each, and how you can build related skills to enter an exciting career in this field. Or, start learning now with the Meta Data Analyst Professional Certificate. In as little as five months, you can learn how to use statistical analysis, including hypothesis testing, regression analysis, and more, to make data-driven decisions. By the end, you’ll have a shareable certificate to add to your professional profile.

Types of linear association

A linear association, sometimes called a “linear relationship,” is a statistical relationship between two variables that you can represent with a straight line [1]. When you have two variables that are linearly associated, changes in one will correspond proportionally to changes in the other. 

In some cases, you might quantify this relationship using the Pearson correlation coefficient (r), which provides a value between negative one and positive one (-1 to 1). If you have values closer to positive or negative one, this represents a stronger association. A value of positive one indicates a perfect positive linear association, while a value of negative one indicates a perfect negative linear association. If you have a correlation coefficient closer to zero, this suggests little to no linear relationship. In real-life examples, a perfect association can be rare, and there is frequently a degree of randomness, which is why scatter plots are a common visualization [2]. 

Positive linear association

A positive linear association is when both of your variables increase or decrease together. If you had your data points on a scatter plot, the data points would cluster around an upward-sloping line from left to right. For example, you might have a positive linear association between hours studied and your exam score, number of advertisements and product views, or height with shoe size.

Negative linear association

A negative linear association is when one variable increases while the other decreases. On a scatter plot, your points would cluster around a downward-sloping line from left to right. You might have a negative linear association between hours exercised and body fat percentage, product price and quantity demand, or household income with stress scores. 

No linear association

If your variables have a correlation around zero, this indicates that you can’t use changes in one variable to consistently predict changes in the other. On a scatter plot, your data points may appear randomly scattered without a straight-line pattern. In finance, you might look at how stock prices correlate with one another to see whether they have a linear association. If the value is close to zero, then you may not be able to use the movement of one to predict the other.

What is linear association used for?

When you have a linear association between variables, you can use changes in one to reliably predict changes in the other. Understanding whether a linear association exists between your variables and how strong it is can be a great first step for exploring your data and deciding on your next modeling steps. In general, you can use a scatter plot to understand the strength (is the association strong or weak?), direction (is it positive or negative?), and form (is it linear or nonlinear?) of your association. 

In everyday life, you see this in simple formulas like the conversion from Celsius to Fahrenheit and calculating distance as a product of rate and time. In data analysis, the same idea helps you to interpret real-world patterns and make realistic predictions. For example, you might predict a home price based on the square footage or the revenue of a business based on its marketing spend. 

In some cases, this type of association has “independent” and “dependent” variables. The independent variable is a variable that you can manipulate directly, while the dependent variable changes in response. In business, your advertising budget might be your independent variable, and your product sales might be your dependent variable. If you knew that the amount of money you spent on advertising was linearly related to your yearly profit, you could use your target annual profit to determine how much to budget for advertisements that year.

Read more: Correlation vs. Causation: What’s the Difference? 

What is linear regression?

Once you've identified a linear association, linear regression allows you to go a step further to model the relationship mathematically and make specific predictions. While linear association tells you that your variables relate, linear regression gives you the equation describing exactly how. For example, you might know that blood pressure and age (independent variables) were linearly related to the risk of stroke. Based on this, you could input several patients' values into the equation to calculate each person's individual risk. This could help you identify high-risk patients and initiate necessary interventions earlier.

What is the difference between linear and nonlinear association? 

If your data has a linear relationship, your variables will change predictably with one another in a straight line. If your data has a nonlinear relationship, your variables do not move linearly with one another. For example, a drug may have no effect on fever temperatures at a low dosage, be very effective in a certain range, then become harmful and ineffective at a very high dosage. If you were to plot the relationship between the medication dosage and efficacy, you would see a curved (nonlinear) pattern instead of a straight line. 

Understanding whether your data has a nonlinear or linear association is important, as the assumptions and constraints of your model will vary depending on the analysis you choose. While linear models are typically simpler than non-linear models, they require stronger assumptions for a 'good' fit, often related to variable behavior.

Who uses linear association?

Professionals across many different quantitative fields use linear associations to explore their data and make predictions. A few common careers you might see utilizing this method include:

• Financial analysts and economists: Financial analysts and economists build prediction models and analyze investment performance to explore questions like how interest rate changes relate to stock prices or how income predicts consumption. 

• Health care researchers: Health researchers, such as epidemiologists, use linear association and regression models to explore exposure-outcome relationships, like examining how treatment protocols relate to recovery rates or whether health care access links to disease prevention.

• Marketing analysts: Marketing analysts use linear association to understand customer behavior and optimize campaigns by analyzing the relationship between spend and revenue, customer engagement and conversion rates, and more.

Pros and cons of using linear association

Linear association is typically considered a straightforward technique to understand and interpret, thanks to measures like the Pearson r. This can offer advantages when you want to communicate findings to non-technical stakeholders, or you’re just getting started with your exploratory analysis. Linear models are also commonly used for hypothesis testing, making them a popular statistical tool across many fields.

While linear association is a powerful tool, it relies on certain assumptions within your data, especially if you use regression techniques. For example, the core premise is that your variables have a linear relationship. If you're unsure how your variables relate to one another, you may need to run diagnostics (e.g., a scatter plot) to determine whether a linear model is truly the best option. In some cases, you might opt for a Spearman rank correlation metric, which is similar to the Pearson correlation but designed for non-linear associations in which x increases or decreases as y increases or decreases, not necessarily at a constant rate. Other assumptions, like your variables being independent, are also typically important when you're building a linear model, and something you should consider during your early, exploratory stages [3].

What is an example of a linear association?

A common example of a linear association is the relationship between height and weight in adults. Taller individuals typically weigh more, resulting in a positive linear relationship when plotted. While this isn't true for every individual (real-world data naturally varies), the overall trend follows a reasonably straight line in practice. When using linear association in real-world contexts, it's important to understand both the strengths and limitations. Because height and weight are linearly related, you can use the pattern to describe general trends and make rough predictions, but you can't predict the exact values for every individual. For example, two people with the same height can have very different weights due to differences in muscle mass, body composition, or other factors.

How to start building linear association skills

As you start building your analytical or statistical foundation, linear association is a great early skill because it leads into more complex analyses, such as regression models, which professionals use across fields [4]. If you want to build these skills through a formal educational pathway, earning a degree in statistics, data science, computer science, or business analytics can be a strong choice. However, if you’re not quite ready to sign on for a full degree program, you can complete online courses or bootcamps at your own pace to get a feel for the subject matter and build skills to help you enter an entry-level position.

Once you master the basics, you can gain experience in the workforce by entering into an entry-level career like a data analyst, marketing analyst, operations analyst, or research analyst. Looking for one of these roles, or something similar, provides an opportunity to build relevant skills like data visualizations, data exploration, and basic data analytic skills that will complement and refine your linear association knowledge.

Explore more data analytics concepts with our free resources

Explore career paths, assess your skills, and connect with resume guidance while browsing our Career Resources Hub. Or if you want to learn more about data analysis, check out these free resources:

• Explore careers: 8 Questions with an Expert: Google Financial Data Analyst

• Learn from peers: Meet the Data Analyst Using His Creativity to Tell Visual Stories

• Build your foundation: Data Analysis Terminology

Whether you want to develop a new skill, get comfortable with an in-demand technology, or advance your abilities, keep growing with a Coursera Plus subscription. You’ll get access to over 10,000 flexible courses.

Build job-ready skills with Coursera Plus

Start 7-day free trial

• 
• 
• 
• 

Start 7-day free trial