A reliable method of estimating the variance function in immunoassay

  • Authors:

  • W. A. Sadler;M. H. Smith

  • Affiliations:

  • Department of Nuclear Medicine, Christchurch Hospital, Christchurch, New Zealand;Department of Mathematics, University of Canterbury, Christchurch, New Zealand

  • Venue:
  • Computational Statistics & Data Analysis
  • Year:
  • 1985

Quantified Score

Hi-index 0.03

Visualization

Abstract

In immunoassay there is strong heteroscedasticity in the observations and it is necessary to estimate the parameters of the variance function. A method for estimation based on approximate conditional likelihood is given. A numerical algorithm for estimating linear, parabolic and exponential variance functions is also descrimed, and this proved reliable for extensive simulated data and real immunoassay data.