From my personal experience with force sensors it is difficult to apply the force directly onto the force sensing area of the sensor. For thin sensors like the flexiforce range, they encourage you to stick a "puck" like object onto the sensing area so as to ensure that the force is applied to the sensor and not the surrounding area. Also with a moving object, it may be difficult to apply the force exactly perpendicular to the sensor, and thus you may introduce a shear stress which is not desirable. I also found that with the flexiforce range the force readings taper off over time.
A crude but possible alternative would be to use a pressure film such as fuji film prescale. It is not ideal for cyclic loading, but it could easily be inserted into a shoe and wouldn't affect the persons gait. Also there would be no circuitry required. It might be worth testing in a lab setting first to see if cyclic loading causes an error in the readings. Also this would give you a profile of the pressures across the whole foot rather than at just one point. You could measure the areas and calculate the forces that way if you still required.
If you want a cheap one, and total reaction only, put a rigid plate on three force transducers (measurement for one foot-ground contact, but it's repeatable)
Check Baropodometrics tools like this http://www.podiatech.com/our-products/analysis-and-measurement-tools-170-2.html I have one of this in my service and It´s easy to manage...Be free to contact me
I'm working with a clever reasearch group at "Roma Foro Italico" University on Motor Science. Prof. Francesco Felici, Prof. Paola Sbriccoli and Prof. Ilenia Bazzucchi (see by means of a simple google browser their address and E-mail) and they have a good expertise on this matter.