Even if a treatment has no effect, it is still possible to obtain an extreme sample mean that is very

Even if a treatment has no effect, it is still possible to obtain an extreme sample mean that is very different from the population mean. What outcome is likely if this happens