Which is why we use a proprietary compensation curve, so our measurement technique replicates the sound as it is heard by real people. We spent many months blind and sighted testing our curve and comparing it to flat-corrected speakers. The aim was to ensure perfect translation of headphone mixes to speaker systems. Hence basically we needed to achieve very little tonal change when switching between corrected headphones and corrected speakers.
We sometimes use sine waves to check for technical problems like resonances, but for drawing response curves it wasn't too useful. The problem is that with sine you'd be chasing dips and peaks which might change with headphone placement. We try to combat that by doing many measurements with slightly changed headphone positioning and then averaging the results.
Also listening to sine will tire your ears very rapidly. I'd like to be able to enjoy my correction after I've done it, fatigue keeps me from doing that with sine.