So what point are you making, soul2000? The cube root of 426 is irrational (it's algebraic, however), so any finite approximation of it will have an error, and the magnitude of the error in cubing that approximation will be cubic. You can only work accurately with irrational roots by treating them as surds, so any finite representation will have the property you give.
For example, to 100 significant figures, the cube root of 426 is approximately
7.524365203641101425512743481414002128348005126377745371287720456777361043416852334287501633996215274
The uncertainty is in the last digit, so all we know is that (assuming the arbitrary precision calculator I used is implemented competently) if we calculated one more digit, the last two digits would be in the interval (35, 45].
If you cube that, it has an error slightly earlier, and the last two digits are wrong, as you would expect: if you take an approximation and cube it, you are also cubing the error (or, more accurately, cubing the ratio between the true value and the approximate value).
425.999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999930
To take the simplest example, if you approximate the square root of 2 as 1.41, which is correct to two decimal places, the error is about 0.3%. But if you multiply 1.41 by 1.41 and compare it with 2, the error is about 0.6%, enough to make it 1.99 rather than 2 to two decimal places.
The question as asked was "to two decimal places" which we can take as three significant figures, ie 7.52 (the correct rounding of the rather more accurate result I give above). If you multiply that out, it's 425.25, which is only accurate to two significant figures. 7.52 is, nonetheless, the right answer for "the cube root of 426 to 2 decimal places".
In order to get the result of cubing your answer to be accurate to two decimal places, as you appear to be asking for, you need a great deal more precision in your calculations: you're asking for five significant figures and the error term in the result will be cubicly greater than the error in the inputs.
With five significant figures:
7.52447.52447.5244
426.051 (426.05, five significant figures)
So you need to work to six significant figures:
7.524367.524367.52436
425.99909 (which rounds to 426.00).
and even then it's only "just" right to creep into the correct interval: with slightly different values you'd actually need seven significant digits in the root to get five in the result.
Error analysis of numerical methods: there's a good reason they're an undergraduate topic.