Alternative hypothesis

The alternative hypothesis, written $H_a$, is the research claim you suspect is true and hope the data will support. It uses an inequality: two-sided ($\neq$, meaning not equal) when any difference matters, or one-sided ($<$ or $>$) when only one direction does. For example, if a new drug should lower blood pressure, $H_a: \mu < \mu_0$, where $\mu$ is the true mean and $\mu_0$ the old standard. You choose the direction from the question before collecting data, never after seeing which way the result came out.