Perhaps someone who is designing integrated circuits can give a better answer, but this is what I think: You can't make a current mirror with two discrete mosfets because the threshold voltage has to be identical for both. That is easy to do in an IC but you would have to measure a lot of mosfets to find two identical. The parameters you listed are for the integrated circuit mosfet, L = length, W = width, etc.
that's because µCoxW/L is known as Kn (for an NMOS that is... for a PMOS it's called Kp). if you look at the saturated mosfet equation, it is Id=(Kn/2)*(Vgs-Vt)^2*(1-λVds)
and Kn/2 is the value they are giving in the datasheet.
λ=1/Va (Va is the "early voltage")
the value µCox is also known as Kn' (prime)
therefore, Kn = Kn' * W/L
the (1-λVds) term is often left out for estimations.
in a current mirror, the Vgs values are the same (and typically you assume the Vt and Kn' values are the same also) so the current ratios depend only on W/L. if you include the last term, then the current ratios are dependent on W/L and the drain-source voltage.
83 CENTS IS TOO EXPENSIVE? How about BC847BV dual bipolar NPN for 44 cents? Mouser (www.mouser.com) has the 2N7002 dual N-type FET for 0.83US and the BC847BV for 0.44US.