Difference between FSK error and EVM (Agilent N9020a)

Using Agilent 89601a software to do the demodulation of FSK and BPSK method, I encounter the error of expression in FSK is "FSK error", while that in BPSK is "EVM". What is the relationship between "FSK error" and EVM, are they the same?

fm modulation is not thought of as being in a "vector space", i.e. 1 dimensional, while BPSK is thought of as being a subset of phase modulation, and thus is portrayed in a vector space.

Thanks biff44. Your answer gives me help. In fact, I try to find out the relationship between EVM and singal to noise ration (SNR). Can we say that in the demodulation for the vector signal, i.e. B/QPSK, QAM, the measured result (vector) can be compared to the reference result(vector)? the error-EVM is related to the SNR as EVM=sqrt(1/SNR) in AWGN channel. For FSK demodulation (i.e. 2FSK), as FM is not related to the phase, and the measured result may be other frequency, the result is not in the two dimensional space formed by the two carriers. So FSK error can not related to EVM or SNR?