Those look more typical - keep in mind, though, that unless you've found a CAL file for the ECM-8000, there will be some non-linearity due to frequency response of the mic not being flat; however, not a big concern because any untreated room will show at least that much variation in response for the LF tests, and that wild a comb filtering (full freq response) until you have absorption at least where first reflection points are located.
If you open the help file and follow the first time user tutorial stuff (don't recall the exact name, sorry but my laptop isn't at hand ATM) it will guide you through the steps for setup (part of which you seem to have already found and done)
Basically, you first set up BOTH channels you're using so they are connected input to output (these MUST be LINE level inputs, not mic level) - once those tests have been done, ETF knows what your soundcard is like and you know how flat to expect your responses to be.
Then you set up as you mentioned, keeping the left channel looped so ETF can "subtract" your system variations and get closer to "real" measurements.
I'm just getting into this software myself, so I'm bound to miss a TON of things to check - I've found that it's very difficult (so far, at least) to interpret what I'm seeing, as to its cause. I bought the entire package from ETF; all extra SW modules and their "calibrated" omni mic and preamp, so I could keep things "mobile" using my laptop, since I intend to add this to my consulting "bag of tricks" once I get half a clue what I'm seeing with it - I've found that moving the mic a small amount can sometimes make a HUGE difference in LF response, but was surprised (in my crappy sounding living room) to see almost NO DIFFERENCE when I raised and lowered the mic, may have to redo those tests after I get more familiar.
I've also noticed that, when ETF is set up to "near real-time" mode, just walking around the room causes VERY noticeable differences in LF response; but when I stand still and let ETF do several "shots" in a row, they change negligibly. So, the conclusion is - if I can get 3 more people to stand in the corners, I'll save big bux on rockwool
BTW, once you get close to minimising the peak-to-trough response with mic at listening position for each speaker, you CAN use both speakers - feed both with a "Y" connector, and set each one's level (with the other one OFF) identical with steady state tones at the mix position, then turn both speakers on - this will give you an idea of any other interferences that may not show up with just one speaker on at a time. (This part is still "gut feel", as I mentioned I'm just getting into this - I've done it, but saw little diff in results (LF tests only) between two or one speaker)
Calibration - if you don't "calibrate" etf (button on one of the test screens) then ETF references the highest SPL encountered to be "100 dB" - this is OK for seeing the total variation in response as a comparison for moving different acoustic materials around, but will NOT tell you actual levels -it may also not produce the same scale for comparison, not sure yet) - to CALIBRATE to nearly accurate levels, you need a SLM - look in your help files under "calibrate", the method of calibrating to REAL SPL's should show up.
General advisory - nearly every program you install under Windows has a help file; virtually all have a "search" function. The first time you open this "search" function, it gives you three choices - one is "maximize search capability" - unless your hard drive is ready to explode, I ALWAYS choose this option - that way, if you're looking for one specific example of a specific word, if it's ANYWHERE in that help file it will sooner or later show up in your searches...
Beyond that, good luck; as I said, I'm still "idiot qualified" on this myself... Steve
