Satterthwaite_df: Compute degrees of freedom using Satterthwaite method
In MarioniLab/miloR: Differential neighbourhood abundance testing on a graph

Satterthwaite_df

R Documentation

Compute degrees of freedom using Satterthwaite method

Description

This function is not intended to be called by the user, and is included for reference

Usage

Satterthwaite_df(
  coeff.mat,
  mint,
  cint,
  SE,
  curr_sigma,
  curr_beta,
  V_partial,
  V_a,
  G_inv,
  random.levels
)

Arguments

`coeff.mat`	A `matrix` class object containing the coefficient matrix from the mixed model equations
`mint`	A numeric scalar of the number of fixed effect variables in the model
`cint`	A numeric scalar of the number of random effect variables in the model
`SE`	A `1 x mint` `matrix`, i.e. column vector, containing the standard errors of the fixed effect parameter estimates
`curr_sigma`	A `1 x cint matrix`, i.e. column vector, of the variance component parameter estimates
`curr_beta`	A `1 x mint matrix`, i.e. column vector, of the fixed effect parameter estimates
`V_partial`	A `list` of the partial derivatives for each fixed and random effect variable in the model
`V_a`	A `c+m x c+m` variance-covariance matrix of the fixed and random effect variable parameter estimates
`G_inv`	A `nxc X nxc` inverse matrix containing the variance component estimates
`random.levels`	A `list` containing the mapping between the random effect variables and each respective set of levels for said variable.

Details

The Satterthwaite degrees of freedom are computed, which estimates the numbers of degrees of freedom in the NB-GLMM based on ratio of the squared standard errors and the product of the Jacobians of the variance-covariance matrix from the fixed effect variable parameter estimation with full variance-covariance matrix. For more details see Satterthwaite FE, Biometrics Bulletin (1946) Vol 2 No 6, pp110-114.