zbMATH — the first resource for mathematics

Asymptotically optimal rank tests for the two-sample problem with randomly censored data. (English) Zbl 0658.62063
The paper is devoted to the asymptotic study of the two-sample testing problem under random censorship of the data. For m,n\(\geq 1\), let \(X_{j1},...,X_{jm},X_{jm+1},...,X_{jN}\), \(j=1,2\), be independent r.v.’s. For \(j=1\) the \(X_{ji}'s\) are distributed according to a continuous d.f. \(F(x-c_{Ni}\Theta)\) for some unknown \(\Theta \in R^ 1\), \(c_{Ni}\) being the scores. For \(j=2\) the r.v.’s \(X_{ji}'s\) are not observable, unlike the r.v.’s \(X_ i=\min (X_{1i},X_{2i})\) and the censoring r.v.’s \(\Delta_ i=I(X_{1i}\leq X_{2i})\), the indicator functions of the events \(\{X_{1i}\leq X_{2i}\}\), \(1\leq i\leq N\). The d.f. H of \(X_ i\) for \(\Theta =0\) is \(H=1-(1-F)(1-F_ 2),\) and, for \(\Theta =0\), \(\Delta_ i\) has binomial (1,\(\pi)\) distribution with \[ \pi =\int (1-F_ 2)dF=\int F dF_ 2. \] Asymptotically optimal rank tests for testing the null-hypothesis \(H_ 0:\Theta =0\) versus the alternative \(H_ 1:\Theta >0\) on the basis of the observable observations \(Z_ i=(X_ i,\Delta_ i)\), \(1\leq i\leq N\), are constructed and their properties studied. The proposed tests are distribution free under \(H_ 0\), i.e., independent of F and of the censoring d.f. \(F_ 2\), and are asymptotically optimal for certain fixed \(F,F_ 2\). The form of the test statistics \(S_ N\) is as simple as in the uncensored case, namely \[ S_ N=\sum^{n}_{i=1}C_{Ni}\{a^{(1)}_{NR_ i}\Delta_ i+a^{(2)}_{NR_ i}(1-\Delta_ i)\}, \] where \(a_{Ni}^{(j)}\), \(1\leq i\leq N\), \(j=1,2\), are two sets of scores.
Reviewer: J.Antoch

62G10 Nonparametric hypothesis testing
Full Text: DOI
[1] DOI: 10.2307/2289478 · Zbl 0652.62037
[2] DOI: 10.2307/1402489 · Zbl 0504.62038
[3] DOI: 10.1214/aos/1176346164
[4] Hajek, J. and Sidak, Z. 1967. ”Theory of Rank Tests”. New York: Academic Press.
[5] Hewitt, E. and Stromberg, K. 1965. ”Real and Abstract Analysis”. Springer: Springer. · Zbl 0137.03202
[6] Janssen A., Local asymptotic normality for randomly censored models with applications to rank tests. (1987)
[7] DOI: 10.1515/9783110850826 · Zbl 0594.62017
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.