I would expect the EM simulated results are different from schematic simulated results in some degrees.
But one thing I found is that, EM simulation shows stable circuit (k-factor >1) while k-factor from schismatic simulation is less than 1. What might be the reasons?
Which result I should rely on?
I presume you mean ADS with "EM simulation". It offers a combination of analog circuit and EM simulation. ADS can be expected to give more precise results in large signal analysis of amplifiers compared to small signal AC analysis of a SPICE tool. But stability results should at least roughly agree. Otherwise one of both analysis is probably using incorrect models or simulations circuits.
Need to see circuit and simulation result details.