Here are some LTspice simulations utilizing a generic LM358 model. The 1st plot shows the result above with 10K and 0.1uF components (compare with #28), the 2nd shows the added "shunt" 1K bias resistor (compare with #27).
Note how the simulations agree well with the actual circuit in the pass band, but begin to deviate as you move into the stop band.
This is a common occurrence with generic SPICE models, where the model doesn't represent the physical component well in certain areas, especially extreme cases.
The LM358 and most GP op-amps, have a very complex output impedance and rely on negative feedback to force the impedance to a low level. As the op-amp open loop gain deteriorates with increased frequency, the output impedance rises in an inductive fashion and can create issues with some circuits as shown here.
As most less seasoned tend to run to the simulations without totally understating the circuit under inspection, caution is advised, especially when "pushing" the limits for critical components.
In the case here the LM358 model is totally inadequate in the higher stop band region, and should be used with caution.
Anyway, more fun things for folks where you can compare physical and simulated circuits and begin the understand the limitations of both
Best,