MarkT:
I doubt the sensor itself has a linear relation between force and resistance (or conductance)
in the first place.
Your doubt is based on what exactly? The design of this sensor seems to suggest linearity with conductance. Not perfect but much more linearity than resistance. The mfg curves more or less back that up. Inverse relationships are killers for ADC resolution with software post-processing invert. One end of the scale has much higher resolution than the other making one of the sensor range unusable. It is much better to do the invert in the analog domain (circuit) if possible. It gets more use out of the sensor.