Stakes alone has little to do with it for most people. It's more about your possible winrate at each stake. If you can beat any given stake at one table, it's generally easier and more profitable to learn to add tables to increase your hourly rate and maximize volume incentives like RB, chase promotions, bonus clearing, and/or VIP rewards, than it is to move up to a stake you can't beat or are not rolled for.
Let's say you beat 5NL for 5bb/100 but can only beat 25NL for 2bb/100 (and this is probably skewed because 25NL is way harder to beat than 5NL). Multitabling cuts into your winrate, so let's say your winrate drops to 2.5bb/100 at 10x5NL and 1.5bb/hr at 2x25NL. You play 100 hands/hour at any single table. Your hourly at 2x25NL would be 3bb/hr or .75/hr while at 10x5NL it would be 25bb/hr or 1.25/hr. And that's not even counting the volume incentives you can earn on top. And if you can beat 25NL at a similar winrate to 5NL then you obviously add more 25NL tables until you can beat 50NL at enough of a winrate to move up and repeat the cycle.
This is why people massively multitable. It also compresses your variance such that swings last for a shorter duration.
(I typed all that on my iPad
and just off the top of my head, hopefully I didn't butcher the math too bad)
However your example is a bit aberrant in that the micros often suffer from brutal rake that seriously erodes your profit potential. So for the specific stakes you mentioned, it might actually be better to play 2x25NL if you can beat it at a decent winrate. I never played that low and can't recall the specifics of just how much the higher rake takes out of your winrate but I'm sure someone will chime in on that.