Change-in-ratio methods have been developed to estimate the size of populations with two or three population subclasses. Most of these methods require the often unreasonable assumption of equal sampling probabilities for individuals in all subclasses. This paper presents new models based on the weaker assumption that ratios of sampling probabilities are constant over time for populations with three or more subclasses. Estimation under these models requires that a value be assumed for one of these ratios when there are two samples; no assumptions about the values of these ratios are required when there are more than two samples. Explicit expressions are given for the maximum likelihood estimators under models for two samples with three or more subclasses and for three samples with two subclasses. A numerical method using readily available statistical software is described for obtaining the estimators and their standard errors under all of the models. Likelihood ratio tests that can be used in model selection are discussed. Emphasis is on the two-sample, three-subclass models for which Monte Carlo simulation results and an illustrative example are presented.
[1]
R. Davidson,et al.
Moment-type estimation in the exponential family
,
1974
.
[2]
Douglas G. Chapman,et al.
Estimates of Mortality and Population from Survey-Removal Records
,
1965
.
[3]
D. S. Robson,et al.
Statistical Calculations for Change-in-Ratio Estimators of Population Parameters
,
1969
.
[4]
Kenneth H. Pollock,et al.
Precision of the change-in-ratio technique for deer population management
,
1986
.
[5]
G. Seber.
The estimation of animal abundance and related parameters
,
1974
.
[6]
D. G. Chapman,et al.
POPULATION ESTIMATION BASED ON CHANGE OF COMPOSITION CAUSED BY A SELECTIVE REMOVAL
,
1955
.
[7]
David L. Otis.
An Extension of the Change-in-Ratio Method
,
1980
.
[8]
K H Pollock,et al.
A new change-in-ratio procedure robust to unequal catchability of types of animal.
,
1985,
Biometrics.