Table 3.

Uniform contamination model 12

		$\varsigma =0.05$						$\varsigma =0.10$
		${\tau }_t=0.2$		${\tau }_t=0.5$		${\tau }_t=0.8$		${\tau }_t=0.2$		${\tau }_t=0.5$		${\tau }_t=0.8$
Trade configuration	Test	P	FPR	P	FPR	P	FPR	P	FPR	P	FPR	P	FPR
$n_t=50$	$V_{\left\{1\right\}}^{\left(t\right)}$	0.034	0.865	0.196	0.546	0.586	0.302	0.030	0.779	0.200	0.346	0.574	0.178
$m_t=50$	TS	0.002	0.000	0.008	0.000	0.154	0.013	0.000	1	0.019	0.000	0.133	0.007
$n_t=100$	$V_{\left\{1\right\}}^{\left(t\right)}$	0.058	0.788	0.436	0.297	0.938	0.184	0.043	0.705	0.425	0.175	0.924	0.097
$m_t=100$	TS	0.004	0.000	0.070	0.054	0.574	0.003	0.002	0.667	0.063	0.000	0.539	0.002
$n_t=200$	$V_{\left\{1\right\}}^{\left(t\right)}$	0.060	0.778	0.810	0.179	1	0.151	0.097	0.484	0.801	0.109	1	0.097
$m_t=200$	TS	0.006	0.500	0.356	0.000	0.964	0.002	0.005	0.444	0.345	0.003	0.959	0.004
$n_t=500$	$V_{\left\{1\right\}}^{\left(t\right)}$	0.272	0.401	1	0.160	1	0.154	0.281	0.226	1	0.069	1	0.081
$m_t=500$	TS	0.028	0.263	0.932	0.000	1	0.004	0.029	0.065	0.928	0.000	1	0.000

Shown are estimated power (P) and false positive rate (FPR) for the first-digit statistic $V_{\left\{1\right\}}^{\left(t\right)}$, using the asymptotic quantile ${\chi }_{\mathrm{8,0.99}}^2$, and for the TS version of the procedure of Barabesi et al. (6), based on $T^{†}=\mathrm{10,000}$ Monte Carlo replicates for each pair $\left(m_t,n_t\right)$. The nominal test size is $\alpha =0.01$.