|
D. A. Ratkowsky (1983; Nonlinear regression modeling. Marcel Dekker, Inc.)
provides a good review in chapter 7 of his book ("Comparing parameter
estimates from more than one data set"). You can use any stat package to
fit the different (full and reduced) models.
Joe Hightower
>Dear fish-ecology netters,
>
>I wonder if anybody can give me some advice on likelihood ratio tests for
>the statistical comparison of
>growth curves. As I gather from the literature this statistical approach
>is somewhat better than other methods
>for comparisons between VBGF parameters (ie Linf, K and t0). I have
>consulted a number of recent papers
>employing this technique, but none of them seems to provide clear details
>on how the growth curves were fitted
>to the different data set/combinations resulting from the linear
>constraints to be satisfied. The original paper
>by Kimura (1980, Fishery Bulletin, 77: 765-776) contains FORTRAN
>subroutines required by BMDP to run the tests,
>although I shudder at the idea of having to go through all that. And I am
>therefore wondering if this could
>be done in another way, perhaps using another statistical package like
>SYSTAT (which I routinely use to fit
>nonlin models).
>
>I would greatly appreciate any feedback on this topic,
>
>Thanks in advance
>
>Lorenzo Vilizzi
>
>
>Department of Zoology
>University of Adelaide
>GPO Box 498
>Adelaide 5005 SA
>Australia
>
>[log in to unmask]
Joseph E. Hightower / NC Cooperative Fish and Wildlife Research Unit
Campus Box 7617, NC State Univ./ Raleigh, NC 27695-7617
phone: (919) 515-8836 email: [log in to unmask]
WWW: http://www4.ncsu.edu/unity/users/j/jhncsu/public/jhightower.html
|
