Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It only takes a minute to sign up.
Sign up to join this community
This is in the following context:
$$ q(t;\theta) = P(T=t;\theta) = \sum_{\{\textbf{x}:T(\textbf{x})=t\}} f(\textbf{x}; \theta) $$
Where $T=T(\textbf{X})$ is a statistic, $q(t; \theta)$ is the pmf of $T$, $X_1, X_2, ..., X_n$ is a random sample and $f(\textbf{x}; \theta)$ is the pmf of $\textbf{X}$.
This is simply summing the probabilities of all combinations of $X$ that lead to $T(X)=t$, since it asks for $P(T(X)=t)$. For example, $T=X_1+X_2$: a statistic with respect to the outcome of the two dice, $X_1,X_2$ are face values, and $t$, e.g. $10$ is the sum of them. You'll have the following sum, $\sum_{X_1+X_2=t}P(X_1,X_2)$ that resembles your general sum.
This simply means sum up the values of the function $ f(\textbf{x}; \theta)$ over all values of $\textbf{x}$ where the test statistic $\textbf{T(x)}=t$
