Correlation between Ti-O bondlength and Ti-3+ Concentration

I see a linear correlation between the amplitude of peak 2 and the length of the ti-o bond reported from Ti-K EXAFS! As the Ti-3+ concentration increases, so does the ti-o average bondlength. No perceviable change in the Ti-O PDF width was noticeable (I subtracted a fitted Debye model + static offset to check), as all deviations from the debye model were within error bars (except the low-T point). I believe that this change in bondlength is consistent with reports of polarons involving Ti-3+ ions in the Ti-O octahedra.