The GRIM test was developed by James Heathers and Nicholas Brown. An excellent overview of the test was written by James HERE and their test has been preprinted HERE. They have an excel calculator for their test HERE. The GRIM test should be part of the standard screening process for new articles, and to simplify its use and help people understand the test I have implemented the GRIM test below.
When calculating the average of whole numbers with small sample sizes only certain fractions are possible. As a result, you can check for an error in a publication simply by using the reported mean of the data. This error may be an honest mistake, or could indicate fraud. The test below will check if a reported mean to two decimal places is mathematically possible for sample sizes less than 100. A more general GRIM test is available casino games.
See this help page for rounding issues.