<i>χ</i>² of distribution collapse begins to differentiate models.

Goodness-of-fit (χ2) as a function of scaling exponent γ. The value of γ that best collapses each data set minimizes the χ2 goodness of fit between the P(ρ(t)) calculated at each timescale. Experimental data all collapse at a value of 0.5 γ < 1, consistent with a superdiffusive MSD.