No label defined (Q3703159)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 3930197
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 3930197 |
Statements
1985
0 references
allometry
0 references
maximum likelihood estimation
0 references
allometric relationship
0 references
log- normally distributed variates
0 references
symmetrical models
0 references
conditional models
0 references
allometric constants
0 references
bivariate log-normal sample
0 references
multinomial
0 references
group index of variability method
0 references
likelihood ratio tests
0 references
standardization quotient
0 references
likelihood ratio confidence regions
0 references
0 references
The theory of maximum likelihood estimation of parameters in an allometric relationship between two log-normally distributed variates is extended to the case where the position of the origin is not known. A distinction is made between symmetrical models where the two variates are assumed to have equal status and conditional models where one variate is assumed to be conditioned by the other. A prior decision has to be taken which situation is considered to apply to the data, for three different sets of estimates can be obtained from them but only one will be appropriate. The nature of the allometric constants is discussed and a bivariate log- normal sample with y conditional on x and size \(N=320\) is generated. Estimates of parameters are obtained by conditional continuous models with 2,3,4, or 5 parameters, by two conditional models, multinomial in x, one approximates the other more exact, and by inappropriate symmetrical models, one continuous the other multinomial in x. Estimates for variate y are compared with those obtained by the group index of variability method, log-normal model, and by the three-parameter log-normal model for the total sample of y data. Results are tested at 95\% confidence by likelihood ratio tests and by comparison with the standardization quotient, assumed to be normally distributed. One-, two-, and three-dimensional 95\% likelihood ratio confidence regions are figured and it is shown that some of them are highly asymmetric with respect to the estimates on which they are based.
0 references