Linear portion of P vs disp curve

I have come to the realization that the slope of the load vs. displacement curve is very sensitive to which "linear" portion you use. For example, on the 1/4" fracture test for lab section 4, the value of the slope varies from -9000 to larger than -11000 depending on which points on the linear portion you use. The problem is that with the large slopes, the PQ intersect becomes significantly less than Pmax, which would make our tests invalid. I will use roughly the first half of the linear portion for this report, but the ASTM guideline recommends basing the slope off the initial linear portion, which is significantly steeper than the middle of the "linear portion". I assume this will be no problem?

This sounds fairly reasonable...just make sure to include an example plot with what you did and also explain in textual form where you took the linear portion of the plot to be located.