I ran some more tests tonight. I thought the sound modes might be affecting the results. I ran two tests with my receiver in Direct mode (i.e. no processing). I expected that changing the crossover would not change the output since Direct mode even disables crossover processing. That is exactly what the measurements showed. The graphs with an 80 Hz and 250 Hz crossovers were identical
I then set the sound mode to Stereo and re-ran the tests (output from REW going to only the right speaker). This time, the responses looked the same, but the 250 Hz crossover curve was 15 dB below the 80 Hz curve, just like before. With an 80 Hz crossover, the output is 95 dB until around 60 Hz, when it starts rolling off.
To isolate my subs, I unplugged my right speaker and ran another sweep. Something unexpected happened. My sub swept up to around 100 Hz, and then my receiver shut off. When I restarted it, it told me that it shutdown because it detected that the right speaker was unplugged. I thought it was interesting that it didn't shutdown until 100 Hz. The output shows that my subs roll off very quickly after reaching the crossover point.
When I superimposed the sub response with the right speaker response, I think I might see what the receiver is trying to do. At 80 Hz, the subs are outputting about 88 dB. At the low and high ends, the subs and mains are outputting about 95 dB. The receiver seems to be keeping the response high all the way through the crossover range and not dropping until it is in the region where the sub is providing strong output. So this might be the receiver's way of trying to ensure a smooth transition during the crossover. Does this seem like a reasonable interpretation of the data? It would explain why the mains are reaching so far into the low frequencies. What do you think?