1: \begin{abstract}

2: The multiplicative censoring model introduced in Vardi

3: [\textit{Biomet\-rika} \textbf{76} (1989) 751--761] is an incomplete data

4: problem whereby two independent samples from the lifetime distribution

5: $G$, $\mathcal{X}_m=(X_1,\ldots,X_m)$ and

6: $\mathcal{Z}_n=(Z_1,\ldots,Z_n)$, are observed subject to a form of

7: coarsening. Specifically, sample $\mathcal{X}_m$ is fully observed

8: while $\mathcal{Y}_n=(Y_1,\ldots,Y_n)$ is observed instead of

9: $\mathcal{Z}_n$, where $Y_i=U_iZ_i$ and $(U_1,\ldots,U_n)$ is an

10: independent sample from the standard uniform distribution. Vardi

11: [\textit{Biometrika} \textbf{76} (1989) 751--761] showed that this

12: model unifies several important statistical problems, such as the

13: deconvolution of an exponential random variable, estimation under a

14: decreasing density constraint and an estimation problem in renewal

15: processes. In this paper, we establish the large-sample properties of

16: kernel density estimators under the multiplicative censoring model. We

17: first construct a~strong approximation for the process

18: $\sqrt{k}(\hat{G}-G)$, where $\hat{G}$ is a solution of the

19: nonparametric score equation based on $(\mathcal{X}_m,\mathcal{Y}_n)$,

20: and $k=m+n$ is the total sample size. Using this strong approximation

21: and a result on the global modulus of continuity, we establish

22: conditions for the strong uniform consistency of kernel density

23: estimators. We also make use of this strong approximation to study the

24: weak convergence and integrated squared error properties of these

25: estimators. We conclude by extending our results to the setting of

26: length-biased sampling.

27: \end{abstract}