Any psychiatrist that has half a brain is going to have a strong foundation in evidenced-based science when it comes to psychotropics (or anything in medicine). I get peeved when a psychiatrist says prescribing is an art and they are artist. I almost became an artist, and I have a brother who is an artist, and a sister that is a film-director.
It's not an art. If you prescribe Nexium because you like that it's a purple pill, I will start thinking you don't know WTF you are doing, and unfortunately I have seen doctors practice in this manner. The only time people should be giving stuff out where there's no to little evidence-based data is when everything evidenced base has been tried and ruled out. Then the person should at least go onto the stuff that could work based on theory, and then if that doesn't work, you shoot in the dark (all the while, informing the patient, and getting second opinions). I'd hardly call this art.
CATIE gives a good foundation on the older antipsychotics, and I wish that someone would do a CATIE with the newer meds. I have talked to people working with the NIMH and they told me it'd likely never happen.
When you, however, get the existing data with the newer meds, they only show they could be potentially superior in very specific niche areas. It is in those areas, such as bipolar depression, where I'll consider they could be used on something higher than 4th line, but then you have to factor in the cost, and in the areas I practice now, I don't have samples of it, making it difficult for me to give it out much even if I wanted to do so.