Thankfully, you don’t have to care very often: because the rounding errors are small, the only practical situation that I’ve seen this issue arise for is when you want to test whether an arithmetic fa...Thankfully, you don’t have to care very often: because the rounding errors are small, the only practical situation that I’ve seen this issue arise for is when you want to test whether an arithmetic fact holds exactly numbers are identical (e.g., is someone’s response time equal to exactly 2×0.33 seconds?) This is pretty rare in real world data analysis, but just in case it does occur, it’s better to use a test that allows for a small tolerance.
