Correlation vs Causation: Definitions, differences and examples Correlation

Correlation vs Causation: Definitions, differences and examples


Table of Contents

What is Correlation?

Correlation is a statistical method of defining the degree of association between two variables. Hence, the correlation between two datasets tells how much they resemble each other. 

When one two variables are being observed at the same time, you are implying a correlation between them. Like when variable 1 is observed, variable 2 is observed too. They either show some change together or show up at the same time. 

There are three main types of correlations that are identified:

  • Positive correlation – when variable 1 increases, variable 2 increases as well and when variable 1 decreases, variable 2 decreases as well. Both variables move in same direction. Example: when you increase your marketing revenue, the sales also goes up (increase).
Correlation vs Causation: Definitions, differences and examples Correlation
  • Negative correlation – when variable 1 increases, variable 2 decreases and vice versa. Example: as the time increases, the distance of a car from its destination decreases. 
Correlation vs Causation: Definitions, differences and examples Correlation
  • No correlation – when two variables are not related or independent of each other and the change in variable 1 does not cause any change in variable 2. Example: An increase in sales of a company does not affect the distance between a random car and its destination.
Correlation vs Causation: Definitions, differences and examples Correlation

Transform your insight generation process

Create an actionable feedback collection process.

online survey

What is Causation?

Causation between two variables defines that variable 1 has a cause-effect relationship with variable 2. Meaning, variable 1 causes variable 2 or vice versa. The variables causing the effect is called the independent variable and the one getting affected is a dependent variable. 

This cause-effect relationship is determined by proper experiments where two groups are made at random from the same. One group is exposed to the treatment and the other isn’t.  Depending on how different are the outcomes of both groups, we can define the causes and effects of the variables.

Why Correlation does not mean Causation

Although correlation means proportional changes in two variables, the change in one variable doesn’t need to be caused by the other. Let’s understand this with an example:

Jack sells woollen clothes. When he collected the data on woollen clothes sales and sales of heaters in his area, he found out that both of their sales are increased and decreased at the same time. 

He then concludes that the sales of woollen clothes and heaters are positively correlated to each other. 

Although, he couldn’t conclude that increasing sales of woollen clothes causes the increase in the sale is heaters as well. Here, a third variable “winter season” is the regulator of both sales of woollen as well heaters in the area. 

Download Market Research Toolkit

Get market research trends guide, Online Surveys guide, Agile Market Research Guide & 5 Market research Template

Making the most of your B2B market research in 2021 PDF 3 s 1.png

Why is it important to know the difference between correlation and causation?

Knowing your research problem and identifying the relationship between the variables is an important task to go ahead with your research. Whether the variables have a cause-effect relationship or their behaviour is just a coincidence, should be cleared out at the beginning. 

Example: you are wondering whether the increase in the visits to your websites is due to better SEO blogs. For this, you need to carry out proper tests to determine if it is just a coincidence or causation.

See Voxco survey software in action with a Free demo.

How to tell if something is a coincidence or causality?

  • Experimental study – randomized experiment is the best way to prove causal relationships. There are two groups: the control group and the treatment group both with similar situations but one gets to experience independent variable treatment and the control group doesn’t. Depending on how different the outcomes of the two groups are, you can tell the causal effect of the independent variable on the dependent variable. 
  • Quasi-experimental study – this works as an experimental study. The only difference is, when there is a need to select the sample before assigning them to the group, you need to conduct a study for that. Meaning, the sample selection is not random. The issue with this is that you cannot tell whether the difference in the outcome is due to the variable itself or due to the absence of randomization. 
  • Correlational study – it is when you determine the correlation between the two variables. That is, when one variable increases, the other increases too and vice versa. Although, not all correlations are causations, as discussed above.
  • Single subject study – related to an individual subject, it is often used in psychology and education. The subject is in charge of its change and the researcher is concerned about changing the subject’s behaviour or thinking. 
  • Stories – these are the stories you hear about your product or service from various sources. Like one customer says they are avoiding buying your products due to poor customer service. One of the customer care employees says customers don’t buy the products due to high prices. This miscommunicates the real motto behind decreasing sales and has no meaning at this point of random stories. 

Explore all the survey question types
possible on Voxco

Read more