Random intercepts linear mixed models {Rfast}

Random intercepts linear mixed models

Description

Random intercepts linear mixed models (for balanced data with a single identical covariate).

Usage

rint.reg(y, x, id ,tol = 1e-08, ranef = FALSE, maxiters = 100)
rint.regbx(y, x, id)  

Arguments

y

A numerical vector with the data. The subject values.

x

For the case of "rint.reg" this can be a vector or a numerical matrix with data. In the case of "rint.regbx" this is a numerical vector with the same length as y indicating the fixed predictor variable. Its values are the same for all levels of y. An example of this x is time which is the same for all subjects.

id

A numerical variable with 1, 2, ... indicating the subject.

tol

The tolerance level to terminate the generalised elast squares algorithm.

ranef

If you want to obtain the random effects (random intercepts) set this equal to TRUE.

maxiters

The max number of iterations that can take place in a regression.

Details

Random intercepts linear mixed models with compound covariance structure is fitted in both functions. The "rint.reg" allows any numerical matrix, with balanced or unbalanced data. See Demidenko (2013, pg. 65-67) for more information.

The "rint.regbx" is a special case of a balanced random intercepts model with a compound symmetric covariance matrix and one single covariate which is constant for all replicates. An example, is time, which is the same for all subjects. Maximum likelihood estimation has been performed. In this case the mathematics exist in a closed formula (Demidenko, 2013, pg. 67-69).

Value

A list including:

info

A vector with the random intercepts variance (between), the variance of the errors (within), the log-likelihood, the deviance (twice the log-likelihood) and the BIC. In the case of "rint.reg" it also includes the number of iterations required by the generalised least squares.

be

The estimated regression coefficients, which in the case of "rint.regbx" are simply two: the constant and the slope (time effect).

ranef

The random intercepts effects.

Author(s)

Michail Tsagris.

R implementation and documentation: Michail Tsagris <[e-mail]> and Manos Papadakis <[e-mail]>.

References

Eugene Demidenko (2013). Mixed Models: Theory and Applications with R, 2nd Edition. New Jersey: Wiley & Sons (excellent book).

See Also

rm.lines, varcomps.mom, colvarcomps.mom

Examples

y <- rnorm(100) 
x <- rnorm(10)
x <- rep(x, 10)
id <- rep(1:10, each = 10)
for (i in 1:20) a <- rint.reg(y, x, id)