Indeed. Though, it's a little more complex than that for spells or attacks that use multiple dice. The interactions aren't linear, and the overall effect when using large dice pools is actually subdued.
Inner Radiance Torrent, at Rank 2, does 4d4 damage, and E[4d4] = 10. But E[4d4c1] = 11.5, rather than 12.5.
Things get even weirder with larger dice, since cascading provides the potential for more rolls, but done at lower dice sizes. E[1d6] = 3.5, and so E[4d6] = 14. But E[1d6c1] works out to be about 15.9, so we're gaining less than 2 damage. A Rank 3 Fireball has E[6d6] = 21, while E[6d6c1] = 23.875, a gain of less than 3.
And this is lower than the naive expectation, because E[1d6e1] = 4.083 and 4*E[1d6e1] = 24.5.
Compare this to increasing the die size. IRT would have an expected roll of E[4d6], which, again, is significantly higher than 11.5, while Fireball at 6d8 has E[6d8] = 27 (vs 23.875). So, this is providing a comparatively small boost.
But, of course, this all assumes fair rolling. When you see what's happening at my table -- which I believe is mostly caused by less than vigorous rolling -- the difference gets even smaller. Assuming a 50% chance of getting a 1 (so weights of [3, 1, 1, 1] on a d4, or [5, 1, 1, 1, 1, 1] on a d6), we end up with E[4d4c1*] = 11, and E[6d6c1*] = 23.25.
The game-impact:enjoyment ratio winds up being quite high. Much higher than I had expected.
