Statistics

Risk, Relative Risk and Odds

Understanding relative risk and odds ratio

Sergen Cansiz
Towards Data Science
5 min readJan 28, 2021

--

According to the definition of risk in Wikipedia; the risk is the possibility of something bad happening. Therefore, when the term of risk is used, the number of negative cases should be considered. In other words, the risk depends on how much negative cases are more than positive ones. Then, what is the relative risk and odds ratio? In this article you will have the answers to the following questions;

  • What is risk and relative risk?
  • How to find the relative risk?
  • What are the differences between risk, relative risk, and odds ratio?
  • How to decide which risk measurement we should use?
Photo by Loic Leray on Unsplash

Risk and Relative Risk

It’s better to go through an example to understand the risk and relative risk the easy way. Suppose there is an e-commerce site which people can buy watches. Let’s assume, this e-commerce site has decided to move the products carousel at the top of the page. They are thinking about relocating this carousel at the right of the page instead of locating it at the top. In that kind of scenario, the risk is that clients (users) don’t find this relocation useable or they can’t get used to it. Considering the site traffic with an average of 10 thousand users per day, making such UI changes can negatively affect the number of purchases. If users are used to the old layout, getting them used to the new one is a huge challenge. Therefore, the e-commerce site has decided to examine only two different groups (samples) before making this change. The first group will see the e-commerce site as a new layout with a product carousel in the right sidebar. The second group will see it as the old layout with a product carousel located at the top. So, the first group will be the experimental group and the second group will be the control group.

They have decided to take 1000 users for each group and they don’t want users who rarely logins to the site. These users’ data might misdirect the results. Therefore, they chose the users who logins to web site at least 2 times a week as average. These users’ actions on product carousels will be tracked. If a user clicks on any item on the product…

--

--