\begin{table}[htbp]
\centering\footnotesize
\caption{Sensitivity of the provincial time-dummy path to the sample screens (log ×100)}\label{tab:trimming}
\begin{threeparttable}

\begin{tabular}{lrrrrrr}
\toprule
Class & Sample & Sales & Total change (Huber) & Total change (OLS) & RMSE monthly vs headline & Max level gap \\
\midrule
house & Headline sample & 422{,}658 & 56.6 & 58.4 & 0.00 & 0.0 \\
house & Assessment-ratio outliers kept & 437{,}144 & 57.2 & 59.6 & 0.27 & 0.9 \\
house & Type conflicts dropped & 420{,}995 & 56.5 & 58.5 & 0.04 & 0.1 \\
house & New construction dropped & 409{,}697 & 56.0 & 56.8 & 0.17 & 0.9 \\
house & Monthly 1–99\\% price trim & 413{,}888 & 56.6 & 58.4 & 0.08 & 0.2 \\
house & Roll-classified sales only & 419{,}217 & 56.5 & 58.2 & 0.07 & 0.2 \\
condo & Headline sample & 122{,}989 & 43.9 & 44.1 & 0.00 & 0.0 \\
condo & Assessment-ratio outliers kept & 124{,}678 & 43.6 & 43.7 & 0.12 & 0.6 \\
condo & Type conflicts dropped & 122{,}037 & 43.4 & 43.7 & 0.05 & 0.5 \\
condo & New construction dropped & 122{,}450 & 43.8 & 43.8 & 0.09 & 0.3 \\
condo & Monthly 1–99\\% price trim & 120{,}357 & 43.9 & 43.6 & 0.09 & 0.2 \\
condo & Roll-classified sales only & 122{,}982 & 43.8 & 44.0 & 0.01 & 0.1 \\
\bottomrule
\end{tabular}

\end{threeparttable}
\end{table}