Table 2.

Best model fits ${\mathrm{G}}^2$ for the P-model and the A-model along with the estimated model parameters $\widehat{\pi }$ (doping prevalence), $\widehat{\mathrm{n}}$ (non-compliance), and $\widehat{\mathrm{p}}$ (inclusion probability)

	P-model			A-model
	$G^2$	$\widehat{\pi }$	$\widehat{n}$	$G^2$	$\widehat{\pi }$	$\widehat{p}$
WCA doping	20.6	21.2%	31.9%	35.0	36.7%	73.4%
PAG doping	15.5	10.6%	9.9%	6.3	39.9%	79.7%
PAG supplements	18.7	8.6%	11.4%	10.6	55.3%	70.5%
Sum	54.9	–	–	51.9