Scaling, as Larry points out is not only not linear, but is affected by thrust, area, volume, weight, speed, etc... not simple. So experience with engines, areas, and weights of actual similarly-sized stunters is probably more valuable.
I posted a lot on SSW about scaling a few years ago. A search might still find that material. It included things like the volume of a model varying as the cube of the linear dimension and the area as the square. Since weight is determined by volume of parts, rather than just plane volume of the model, even this is not as helpful as might otherwise be. That's why, unless you are really into computational stuff, what I'm going to attach is only useful in predicting trends as you scale a model up or down. It's probably just an FWIW here. Although my total file size for attachments is only 851 Kb, I can't post them all because a message of "too large" comes up. So although I don't even know whether the text is legible, I'll go ahead and make two posts. Edit: It seems to me that I disagreed with one of these authors on something, but I don't have the time to see where that might have been now. 'sorry!
For Larry: Here on Robert's forum, there actually are subscripts and superscripts available. So "A squared" is "A2". Otherwise, I have always used the upward-pointing "^" for exponents/powers. So "A cubed" would be typed as "A^3," for instance. The square root of A could be typed "A^.5" (A to the halfth power). That allows you to use "*" or "x" for multiplication.
SK