1: \begin{abstract}

2: In this paper, we study the nonparametric maximum likelihood estimator

3: for an event time distribution function at a point in the current

4: status model with observation times supported on a grid of potentially

5: unknown sparsity and with multiple subjects sharing the same

6: observation time. This is of interest since observation time ties occur

7: frequently with current status data. The grid resolution is specified

8: as $c n^{-\gamma}$ with $c > 0$ being a scaling constant and $\gamma>0$

9: regulating the sparsity of the grid relative to $n$, the number of

10: subjects. The asymptotic behavior falls into three cases depending on

11: $\gamma$: regular Gaussian-type asymptotics obtain for $\gamma< 1/3$,

12: nonstandard cube-root asymptotics prevail when $\gamma> 1/3$ and

13: $\gamma=1/3$ serves as a~boundary at which the transition happens. The

14: limit distribution at the boundary is different from either of the

15: previous cases and converges weakly to those obtained with

16: $\gamma\in(0,1/3)$ and $\gamma\in(1/3,\infty)$ as~$c$ goes to

17: $\infty$

18: and $0$, respectively. This weak convergence allows us to develop an

19: adaptive procedure to construct confidence intervals for the value of

20: the event time distribution at a point of interest \textit{without

21: needing to know or estimate $\gamma$}, which is of enormous advantage

22: from the perspective of inference. A simulation study of the adaptive

23: procedure is presented.

24: \end{abstract}