Error in result of numerical trapezoidal integration


I wondered if there was anybody out there who could tell me how to work out the error in my numerical integration.

I have the falling side of a quasi-maxwell-boltzmann distribution and have 12 experimental data points and I only want to calculate the area between my first and last data point. I used the trapz function on MATLAB which calculates the composite trapezoidal rule for me and I've looked at the different results when using 2-6 and 12 data points and am happy that it has reasonably converged toward the answer.

Can I use the difference between using 6 and 12 data points and use this as a maximum error (as it should reduce further than this on every iteration) or is there a better way to treat this?