A one-dimensional particle-in-cell simulation incorporating Monte Carlo collisions (PIC-MCC) has been utilized to investigate a model Ar/O2 discharge. This benchmarking study is unique in many respects, but most notably in the size of the parameter space it encompasses. In total, data from more than fifty self-consistent kinetic simulations, covering a wide range of conditions in terms of collisionality, electronegativity, and negative ion destruction mechanism, has been compiled. This data conveys a unique perspective of the complex charge species dynamics associated with electron attaching discharges. Under certain discharge conditions quasi-neutrality violating double layer structures are observed. A largely unappreciated negative ion heating mechanism is identified, and negative ion temperatures greatly exceeding those of the surrounding gas are observed. A generic global, or volume-averaged, plasma chemistry model has also been developed. We utilize the benchmark simulations to critically evaluate the performance of such models over a wide range of parameters. It is found, as expected, that the most significant limitation appears to be the oft-used assumption of a Maxwellian electron energy distribution. Accounting for this deficiency is shown to improve model-simulation efficacy considerably. Although many works have been published which endeavor to incorporate the effects of self-consistent electronegative plasma segregation into models exploring the primary discharge parameters and scaling laws, it is found that the conventional global model formulation is quite robust to the occurrence of such complex state-variable profiles.