1. Distance to the building in the control footage.
2. Lack of comparable compression in the control footage.
3. Measurements dependent on subjective assessments (of "best fit" frame positions of airplane and building edges) performed by un-blinded assessor.
These are three substantial problems with your control case, and hence with your study's conclusions. Note that they would be considered problems regardless of what those conclusions were.
You asked for responses to your study taken at face value, without regard to external issues such as what live witnesses saw or whether the live-composite conjecture is a priori plausible. So, there it is. Do you have a response?
Ideally, I'd like to see you continue your study and correct these problems with your control. You've been honest about presenting your methods so far, which is what makes this criticism possible at all, so perhaps you might be willing to carry it through to a conclusive result, regardless of what that result might be.
You cannot correct every possible source of error -- for instance, the speed of the planes will inevitably be different, and there's little recourse.* That's one reason why no single scientific study is ever completely conclusive. But you can and should at least address those factors for which plausible explanations of how they could contribute to error in the findings have already been presented, such as I did in my previous post.
Each of the three issues listed above is potentially fixable, at least in large part. For #1, you're clearly not going to find a flight path that coincides with a building at exactly the same distance (that is, that crashes into the building), but many airports have commercial strips of hotels and convention centers that are within a mile or two of a runway, where you should be able to find a viewpoint where the building is at least 2/3 the distance to the plane instead of prominently in the near foreground. Even better if you can use a service building, enclosed tower, etc. that's on the airfield itself.* For #2, just compress the control video using the same algorithm as the 9/11 footage you're using, or get someone else to do it for you. For #3, get some volunteers and have them perform the procedure of locating the plane's outline in the frame, on individual randomized frames in which the identifying features of the 9/11 versus the control footage have been cropped off or blanked out. The logistical details of this might be tricky, but it can be done. As a bonus, you'll have data that shows the variation in assessments by different people. You can even give the same frame to each assessor more than once, so you'll have data on how consistent those assessments are.
*With jetliner/airport footage you'll never be able to get the pixel size and pixel speed to both be comparable to the 9/11 shot, because jetliners are not flown that speed near the ground normally. But here's a suggestion for how to model those same pixel sizes and speeds using a different scale: consider using footage of planes taking off or landing on aircraft carriers. If you can find footage from the right distance, you should be able to get the pixel size of the plane in the image and the pixel speed of the plane across the frame to both be comparable with the 9/11 footage, even though the plane's actual size and speed will be smaller and lower. Then you can use the carrier's island superstructure, which is necessarily very close to the flight path, as your "building" position.
Respectfully,
Myriad