Regarding the green curve - you don't really need A/B test to see that it can sound better - having the 320-950Hz range playing higher than 100-320 Hz cannot possibly sound good.
A/B listening is important to me because what looks initially unacceptable in the presented steady state curve may not sound as "bad" as it looks. From time to time, what looks worse in a graph, may actually sound good to one's ears -- or, at least perceptually "neutral", or not even necessarily "worse" compared to the original (no EQ or "treatment") response.
I don’t like to EQ the response above 100Hz anyway.
If the default natural voicing of your speakers is good enough for your ears, then there's probably no further need for that... leaving it be should be fine.
But, with adequate data -- i.e. multiple measurements, filtering and/or viewing methods -- one should still be able to safely apply more detailed full-range room and speaker EQ with little cause for concern.
@abdo123 A fun function of the newer REW versions is the RT60 decay mode where you can choose any frequency and get the decay for that frequency alone. Would be nice if you could pick a few random frequencies and show before and after just to see how the relative decay changes. Say you pick frequencies where the frequency response is most similar between with and without curtains.
Easier to visualize than the whole waterfall
Besides the somewhat underutilized RT60 decay view mode in REW, wavelet spectrograms and "spectral" decay graphs can be quite revealing as well. The difficult part is connecting the dots or correlating what is seen in the graphs vs what is actually heard/perceived.
---
The ff. are measurements from a badly positioned (corner-backwall-placed) left rear surround speaker pointing directly at the opposing wall/right side surround speaker instead of my actual Desk MLP which is closer to the front half of the room. Do note, however, the room is partially already dampened from construction by default; and, I have, through the years, acoustically "treated" the space -- though, haphazardly! BTW, this is because I'm no acoustician and only half of the time know what I'm doing. It's got everything: highly reflective and heavy sound absorptive curtains, foam acoustic wedges, a ceiling cloud, and various other DIY panels.
The speaker above is a modified (sealed and internally overstuffed) LSR305. Directivity of this small wave-guided speaker is good despite some non-linearities seen in the steady state response.
The graphs should also illustrate why, IMO, single-point measurements at the main listening position are actually not "useless" or wholly uninformative -- in fact, a lot can be garnered from that method alone. But, on the other hand, if the task also included equalizing above 500 Hz or so, I wouldn't truly be so confident of not messing things up without the additional use of other reference curves. And because this is only one speaker in a larger 7.1 MCH system, I find A/B listening between the rest of the channels (having different speaker brands/types in the mix) also
very useful very important.