I have never known what the actual rule is for this. It doesn't really matter unless you actually want to tell someone what it is.

The way I've been figuring it is to find the AR for each wing and average them together.

The other way I've done in the past is to take the average span and divide by the distance from the forward most leading edge to rearward most trailing edge. I'm fairly certain that's wrong though.

What's the proper way?

Here's the formula to calculate aspect ratio. Don't know about multi-winged aircraft, though.

http://www.grc.nasa.gov/WWW/Wright/airplane/geom.html

The aspect ratio of multi-wing airplanes is normally calculated as the average aspect ratio of the individual wings, weighting the average in relation to the area of each wing. In reality, the effective aspect ratio of a multi-wing airplane will normally be much lower than the geometric aspect ratio, due to mutual wing interference. At a normal gap/chord ratio, for example, a bipe with a geometric aspect ratio of 6 will exhibit induced drag typical of a monoplane with an aspect ratio of about 3.5. Increasing the gap will help raise the effective aspect ratio. This is often called the Monoplane Equivalent Aspect Ratio.

I like to use higher than average aspect ratios on my aerobatic bipes, in order to reduce induced drag. Normally use about 8.0, and have gone to 10.0, with good results - nice flat glide, much less speed loss in tight turns.

I feel with bipes there's two distinct wings.
The interference between them and the extra wing tips preclude assigning a sensible aspect ratio to the combination.
An equivalent area monoplane performs much better than a bipe in terms of aerodynamic efficiency... flies faster on the same power, glides better.