Calculates contribution of one subject to the log-likelihood for lme object

This is method for logLik1() generic function.

# S3 method for class 'lme'
logLik1 (modfit, dt1, dtInit)

Arguments

modfit

an lme object representing model fitted using maximum likelihood.

dt1

a data frame with data for one subject, for whom the log-likelihood function is to be evaluated

dtInit

an optional auxiliary data frame.

Value

numeric scalar value representing contribution of a given subject to the overall log-likelihood returned by logLik() function applied to lme object defined by modfit argument.

Details

Calculates profile likelihood (with beta profiled out) for *one* subject. Data with *one* level of grouping only. correlation component in modelStruct not implemented.

Author

Andrzej Galecki and Tomasz Burzykowski

Examples

require(nlme)
lm3.form <- visual ~ visual0 + time + treat.f
(fm16.5ml <-                                # M16.5
   lme(lm3.form,
   random = list(subject = pdDiag(~time)),
   weights = varPower(form = ~time),
   data = armd, method = "ML"))
#> Linear mixed-effects model fit by maximum likelihood
#>   Data: armd 
#>   Log-likelihood: -3210.685
#>   Fixed: lm3.form 
#>   (Intercept)       visual0          time treat.fActive 
#>     5.4472103     0.8997320    -0.2415519    -2.6563784 
#> 
#> Random effects:
#>  Formula: ~time | subject
#>  Structure: Diagonal
#>         (Intercept)      time Residual
#> StdDev:    7.170457 0.2801255 5.022971
#> 
#> Variance function:
#>  Structure: Power of variance covariate
#>  Formula: ~time 
#>  Parameter estimates:
#>     power 
#> 0.1117254 
#> Number of Observations: 867
#> Number of Groups: 234 
 df1 <- subset(armd, subject == "1")       # Panel R20.7
 logLik1(fm16.5ml, df1)
#> Error in eval(mCall$fixed): object 'lm3.form' not found