Download now Free registration required
This paper concerned with developing a new technique for parameter estimation in the linear functional relationship model when replications are not available. The authors propose a grouping algorithm aided by a graphical technique to obtain pseudo-replicates from unreplicated data. Critically, this allows them to obtain an estimate of the ratio of error variances. Other parameters can then be estimated by maximum likelihood estimation based on this estimate. This algorithm could be regarded as an extreme generalization of earlier ad hoc methods but has the advantage of producing estimates of the ratio of error variances which they cannot get by any other available method. A simulation study shows that this technique compares favorably with other proposed methods for estimating the slope parameter.
- Format: PDF
- Size: 94.02 KB