Agreement statistics are a fleet of tests/measurements which provide an answer to the question: “how similar are my two data sets/groups of data?”. I like to compare this question to the one answered whilst performing a standard two sample t-test. In a t-test we usually want to know: “how different are my two data sets/groups of data?”. Of course, it’s important to remember that in statistics these two questions are very different and you can’t simply use a t-test to answer the former question.
The question of “how similar are my two data sets?” is a question often asked when you are wanting to compare the performance of a drug or medical device to a predicate or industry standard. You want to show that your new drug or medical device performs just as well as the predicate or industry standard.
There are many ways of answering the question of similarity which also depend on the type of data you are comparing.
Below lists the statistics which can be used to statistically measure agreement. Click on the links (if applicable) to learn more.