It depends on mode. Here are the basics:
In manual mode, narrowing the aperture will absolutely reduce the amount of light hitting the sensor as long as you are at sync speed or below. Naturally, the lower the speed, the more the ambient light factors into the equation, which clouds the issue a bit.
In TTL mode, narrowing the aperture will cause the flash to power up in an attempt to get a correct exposure. You can exceed the capability of the bounced flash -- 430EX flashes aren't the most powerful kids on the block. Anyhow, the more the flash powers up, the longer the duration and the more the sync speed matters. Again, ambient light will cloud the issue.
The best way to test it is this: Set up a simple target, relatively close to the camera. A still life or test card will do. Put the flash on camera and try the experiment of varying the aperture in a dim room to remove ambient light from the equation.
Next, add the bounce into the mix. Try it on manual and then auto. You should get the results you expect.
Finally try on-camera and bounce with some ambient light mixed in.
With these results and the camera in the same position looking at the same subject, you should be able to see what is causing your unusual results. But be aware, stopping the lens down is not exactly the same as powering down the flash -- just mostly the same. Flash vs. ambient is the largest differentiating factor; "smarts" built into the camera and flash are another factor.
It would be great to hear what your results are.