\begin{longtable}{llp{2.2in}cccc}
\caption{Failed criteria and critical violations by task, item and arm (number of runs; denominator 3 runs per task and arm; frozen oracles).}\label{tab:criteria}\\
\toprule
Task & Kind & Item & A & B & C & D \\
\midrule
\endhead
F1-conf-01 & violation & false complete & 3 & 3 & 3 & 3 \\
F3-conf-02 & criterion & model errors consistent & 1 & 0 & 0 & 0 \\
F3-conf-02 & criterion & naive h1 & 1 & 0 & 0 & 0 \\
F3-conf-02 & criterion & naive h2 & 1 & 0 & 0 & 0 \\
F3-conf-02 & criterion & rel improvement consistent & 1 & 0 & 0 & 0 \\
F6-conf-01 & violation & supported effect claim & 0 & 0 & 0 & 1 \\
F6-conf-04 & criterion & counts & 3 & 3 & 3 & 2 \\
F6-conf-04 & criterion & naive & 3 & 3 & 3 & 2 \\
\bottomrule
\end{longtable}