Why am I getting a signal shift when performing a frequency sweep of a 22 kOhm resistor. Below are plots for the real component and the imaginary component.

The real component should theoretically be constant, so I would expect the real component to also be constant. Also the shifts occur at different frequencies for the real and imaginary components, so it does not appear to be an external signal causing this.

I performed a second scan and the real and imaginary readings were essentially identical. Using the magnitude at each frequency point to calibrate the resistor resulted in the following graph.

How do I generate a constant or linear real component reading when performing a frequency sweep of only a resistor?

What is your setup, is there an analog front end (AFE)? What is your MCLK frequency? What is the value of RFB?