The US Army has a
little table of recommended separation distances from certain objects. Unfortunately, it doesn't include GPSs or anything like that.
In World War II, the steel helmet weighed over 4 pounds, and is said to be the cause of some pretty major navigational errors.
Here's what I do. I get the compass out in the open, far from metal stuff, and then I move the object I'm testing close to the compass. If you see the needle deflect, it's affecting the compass. Pick some starting orientation of the object and the compass, and then be sure to try turning the object 90 degrees.
Just sitting here at the computer, I can hold the compass in one hand and move the GPS around it with the other. I'm getting maybe 5 degrees deflection when they're almost touching. Moving them apart far enough to simulate the distance from my wrist to my hand, there is no noticeable effect.