Interval estimation of the critical value in a general linear model (Q1092553)

This paper concerns interval estimation of the critical value \(\theta\) which satisfies \(\mu (\theta)=\sup_{x\in {\mathcal X}}\mu (x)\) under the general linear model \(Y_ i=\mu (x_ i)+\epsilon_ i\) \((i=1,2,...)\), where \(\mu (x)=\sum^{p}_{j=1}\beta_ jf_ j(x)\) for \(x\in {\mathcal X}\) and the functional forms of \(f_ j's\) are known. From an asymptotic expansion it is shown that, under reasonable conditions, the limiting distribution of \(\sqrt{n}({\hat \theta}_ n-\theta)\) is normal. Thus in the large-sample case a confidence interval for \(\theta\) can be obtained. Such a result is useful when one is interested in carrying out a retrospective analysis rather than designing the experiment (as in the Kiefer-Wolfowitz procedure). In Section 3 a sequential procedure is considered for confidence intervals with fixed width 2d. It is shown that, for a given stopping variable N, \(\sqrt{N}({\hat \theta}_ N-\theta)\) is also asymptotically normal as \(d\to 0\). Thus the coverage probability converges to 1-\(\alpha\) (preassigned) as \(d\to 0\). An example of application in estimating the phase parameter in circadian rhythms is given for the purpose of illustration.