I don't think that's actually true in my experience. I mean, we all have our own experiences, right, but in D&D, there's a key and vital difference between Rogue skill checks and spells, and that's that spells always succeed.
So when the chips are down, and you really need to succeed, you use a spell. You don't rely on a roll. If you're just chancing it, fine, but it's unlikely a spell would have been expended on that anyway. Spells also frequently enable vastly simplified plans. A Rogue-based plan might have a half-dozen steps, all theoretically requiring separate checks (which with D&D's binary-pass-fail and zero ability to mitigate failed checks, is a lot of risk), whereas a magic-based plan may well require zero checks or far fewer. Some of this is DM-dependent too - if you have the kind of DM who doesn't go in for kinda-unnecessary or repeated checks, any kind of Skill check-based approach is vastly more likely to succeed than otherwise, but my experience is a lot of DMs will demand a Skill check for literally every possible point of failure (or close to it), which makes some kind of serious problem very likely in a binary pass/fail-based system.
(The solution is pretty simple - make it so spells can fail and so the effects they provide are flawless, but 5E hasn't done that. I imagine 6E may well though.)
Rogues (and no other classes, except I think one Bard subclass maybe) can mitigate this issue a bit with Reliable Talent, and it helps them that it's been lowered from 11th in 2014 to 7th in 2024, but I feel like they probably need to lower it even further, at least for say the skills the Rogue has Expertise in.