The power of a significance test is technically defined to be "the probability that it will reject a false null hypothesis" (or, in other words, the probability of not making a type II error).

But it is usually difficult to compute that probability, because to do so requires knowing things that we have no way of knowing. So it is probably not useful to think of the power of a test as a number, but rather as a good quality of the test that we might want to enhance. But how can we increase the power of a significance test? Here are two ways:

• Pick a larger "alpha-level" (instead of the usual 5%), so that more often an experimental result will be considered significant evidence against the null hypothesis. But of course this would make a type I error (rejecting the null hypothesis when it is true) more likely.
• Use a larger sample for the experiment. For example, in a z-test, that would make the SE smaller -- because the denominator in the formula SD/(sample size) would be larger -- so the z-value would be larger -- because the denominator in the formula (x - EV)/SE would be smaller -- so the p-value would be smaller, i.e., more likely to be significant. Of course, this might be giving a "spurious power" to your test;the apparently convincing p-value may discourage careful review of your experimental method for error or bias.