A new type of test called B-test has been developed to compare two sets of data. B-tests use subsets of the data to calculate a statistic that shows if the two sets are different. By choosing the size of these subsets, researchers can balance the test's accuracy and speed. B-tests are more powerful than some existing tests and are easier to compute than others. They also have a normal distribution under certain conditions, making them easier to interpret. Choosing the right type of calculation can improve the test's accuracy.