For simplicity. That means that the MM experiment doesn't prove that aether theory is wrong, just that it needs to be modified. I've read that a viable aether theory exists that explains all experiments up to around 1920, but that it was very complicated, and not worth using since relativity was much simpler in comparison. Since 1920 we've had lots of other experiments that can be explained using relativity. Eg. the QED calculation of the electron moment, which relies on special relativity, matches the experimental result to 1 part in 109 (or whatever the limit of experimental precision is). It suspect aether theory probably could not match that no matter how it's modified.
There's a second reason. The Galilean transform was already known from experiments to be applicable to Newton's laws. Assuming the aether to have a unidirectional constant velocity was the simplest way of using the Galilean transform to analyse the Michelson-Morley experiment.
The aether notion was rooted in classical physics [circa 19th century]. Scientists noted that sound required a media through which to propogate. They attempted to apply this logic to EM wave propogation. They deduced the speed of light, as is the case with sound, would be affected by motion through the media. The MM experiment assumed the speed of light would be seasonally affected if it propogated through a media [due to earth's motion through the 'aether']. No such affect was observed. See http://en.wikipedia.org/wiki/Michelson-Morley_experiment for further details.