Select an LLM

Description

Without using covariates (i.e., with capture probabilities assumed flat over the covariate space), select the best log-linear model for the marginal contingency table of capture pattern counts.

Usage

1
2
flat.IC(pop, models = make.hierarchical.term.sets(k = attributes(dt)$k),
  rasch = FALSE, ic = "AICc", adjust = FALSE, averaging = FALSE)

Arguments

pop

A data.frame containing CRC data as output of formatdata.

models

A list of models – or an expression that returns a list of models – to be considered in local model search. The default is NULL, and in this case make.hierarchical.term.sets(k = attributes(dat)$k) is called to generate all hierarchical models that include all main effects.

rasch

Logical: Should the Rasch model (most basic version, Darroch et. al. 1993) be considered, in addition to standard models? FALSE by default.

ic

Character string specifying the information criterion to use for model selection. Currently AIC, AICc, BIC, and BICpi are implemented.

adjust

Logical: Should we adjust the cells as in Evans and Bonett (1995)?

averaging

Logical: Should we use model averaging based on the information criterion scores?

Value

pred

The point estimate of the population size

form

The log-linear terms in the chosen model

Author(s)

Zach Kurtz

References

Fienberg SE (1972). "The Multiple Recapture Census for Closed Populations and Incomplete $2^k$ Contingency Tables." Biometrika, 59(3), pp. 591.

Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.