Consider just a resistor at DC. The power is the voltage at node a times the current *minus* the voltage at node b times the current. Or V(Va, Vb) * I. 
Naming the nodes may make things a bit easier to see. 

If you sum the absolute value of the two you'll get some weird value that depends on the voltage at the resistor wrt ground. 

Speaking of which- I don't see a ground in your circuit. SPICE wants to see a ground. 

Also, using my model for the IRLZ44N  I actually get about 600mW for the power dissipation. 

[![enter image description here][1]][1]


  [1]: https://i.sstatic.net/TpXBk79J.png