I am trying to round numbers in the following table.
\begin{table}[H]
\centering
\begin{tabular} {c| cccc}
T(sample size) & $\theta_0$ & $\theta_1$ & $\theta_2$ & $\theta_3$ \\ \hline
200&-0.081295&-0.090014&-0.089029&0.010305\\\hline
500&-0.05266&-0.037185&-0.03383&-0.001537\\\hline
\end{tabular}
\caption{I.I.D. contexts: bias in estimating the optimal policy parameter. Bias=$\mathbb{E}(\hat{\theta}_t)-\theta^*$. }
\label{iid_b}
\end{table}
However, if I do the following revision by replacing c with S[round-precision=2], latex compiler reports error and subscripts of theta will be lost.
\begin{table}[H]
\centering
\begin{tabular} {c| S[round-precision=2]S[round-precision=2]S[round-precision=2]S[round-precision=2]}
T(sample size) & $\theta_0$ & $\theta_1$ & $\theta_2$ & $\theta_3$ \\ \hline
200&-0.081295&-0.090014&-0.089029&0.010305\\\hline
500&-0.05266&-0.037185&-0.03383&-0.001537\\\hline
\end{tabular}
\caption{I.I.D. contexts: bias in estimating the optimal policy parameter. Bias=$\mathbb{E}(\hat{\theta}_t)-\theta^*$. }
\label{iid_b}
\end{table}
I also want to avoid changing the decimal places for each number. any suggestions?

numprintas explained in this answer. This solution only touches the preamble to declare the package import and in the column heading without upsetting the original alignment. – Pablo Adames May 21 '20 at 00:59