Variance function estimation for immunoassays

✍ Scribed by Gillian M Raab; Robert Thompson; Ian McKenzie

• Publisher: Elsevier Science
• Year: 1980
• Weight: 400 KB
• Volume: 12
• Category: Article
• ISSN: 0010-468X
• DOI: 10.1016/0010-468x(80)90057-4

✦ Synopsis

A computer program is described which implements a recently described , modified likelihood method of determining an appropriate weighting function to use when fitting immunoassay dose-response curves. The relationship between the variance of the response and its mean value is assumed to have an exponential form, and the best fit to this model is determined from the within-set variability of many smaU sets of repeated measurements.

The program estimates the parameter of the exponential function with its estimated standard error, and tests the fit of the experimental data to the proposed model. Output options include a list of the actual and fitted standard deviation of the set of responses, a plot of actual and fitted standard deviation against the mean response, and an ordered fist of the 10 sets of data with the largests ratios of actual to fitted standard deviation.

The program has been designed for a laboratory user without computing or statistical expertise. The test-of-fit has proved valuable for identifying outlying responses, which may be excluded from further analysis by being set to negative values in the input file.