More info: Look at the first sim of the passive network. Note R6. When creating the schematic with the LTSpice graphics editor, R6 can be installed either way around, meaning that this defines which way current flows through it when data is saved when the simulation runs. This applies to all symmetric two terminal components, like resistors, capacitors, inductors, etc.
It is simple to tell which way around any given component is instantiated, but only after the Sim has been run. Moving the cursor over the component shows a miniature "Amprobe", where the current direction is shown with an arrow.
I plot Zin = V(in)/I(R6) in the top plot pane. With R6 oriented so that current flows through it from left to right, then everything makes sense, including the Real and Imaginary plots. If R6 was flipped left to right, then Zin = V(in)/-I(R6), and all the plots are effected appropriately. The blue text is from the LTSpice help file, and defines what waveform arithmetic functions are specific to complex waveforms.
Now back to the OP's Amp. I had to turn R6 around, and widen out the frequency range to get these plots. Note that I(R6) reverses and V(in) is higher than the input source for a range of frequencies. This shows that the amplifier is feeding back into its input, exhibiting some signs of instability. It shouldn't be surprising then that the input impedance goes negative...
Remember the "Q-multiplier"?