With RAID 0, there is a process called "striping". That means that data is written across two disks.
With RAID 5 for example, you need at least 3 disks, two for striping data and one for parity checking the data on the other two drives.
So with RAID 0, if the data is lost on one, or it crashes, you can't get it back from the other since it is used in tandem with the one that crashed.
With RAID 1 (mirroring) at least you would have an exact duplicate on the second disk, except the writing to disk is much slower in relation to RAID 0.