[R-sig-ME] glmer and influence.me - complaining about nAGQ==0

[Cátia Ferreira De Oliveira]

Dear Professor Bolker,

I am currently running the allfit function for the models I mentioned in
the previous email and I am now doing the same for another model using the
same data and this seems much worse. This is the model that seems to give
even worse fit with this value "Model failed to converge with max|grad| *=
0.167262* (tol = 0.002, component 1)" being much higher. Are there any
other suggestions I could take? This model I am running without the logRT
and the nAGQ==0.

Thank you!

## Generalized linear mixed model fit by maximum likelihood (Laplace
##   Approximation) [glmerMod]
##  Family: Gamma  ( log )
## Formula:
## RT ~ Probability * Session * Group * Age + (1 + Session *
Probability |  Participant)
##    Data: Data.trimmed
## Control: glmerControl(optimizer = "bobyqa", optCtrl = list(maxfun = 2e+05))
##
##       AIC       BIC    logLik  deviance  df.resid
##  938050.5  938301.3 -468998.3  937996.5     79890
##
## Scaled residuals:
##    Min     1Q Median     3Q    Max
## -4.258 -0.597 -0.151  0.410 33.938
##
## Random effects:
##  Groups      Name                  Variance  Std.Dev. Corr
##  Participant (Intercept)           1.700e-03 0.041232
##              Session1              2.583e-04 0.016073  0.13
##              Probability1          1.121e-04 0.010589 -0.06  0.11
##              Session1:Probability1 6.176e-05 0.007859  0.02 -0.02 -0.02
##  Residual                          5.448e-02 0.233418
## Number of obs: 79917, groups:  Participant, 45
##
## Fixed effects:
##                                    Estimate Std. Error t value Pr(>|z|)
## (Intercept)                       5.9334448  0.0739725  80.211  < 2e-16 ***
## Probability1                     -0.0162354  0.0100753  -1.611  0.10709
## Session1                          0.0631180  0.0209673   3.010  0.00261 **
## Group1                           -0.0331517  0.0740287  -0.448  0.65428
## Age                               0.0035260  0.0023510   1.500  0.13367
## Probability1:Session1             0.0059755  0.0075507   0.791  0.42872
## Probability1:Group1               0.0060958  0.0101003   0.604  0.54616
## Session1:Group1                   0.0215458  0.0210650   1.023  0.30639
## Probability1:Age                 -0.0003351  0.0003189  -1.051  0.29331
## Session1:Age                     -0.0006021  0.0006607  -0.911  0.36220
## Group1:Age                       -0.0009086  0.0023529  -0.386  0.69938
## Probability1:Session1:Group1      0.0067270  0.0075839   0.887  0.37507
## Probability1:Session1:Age        -0.0001585  0.0002380  -0.666  0.50544
## Probability1:Group1:Age          -0.0002775  0.0003196  -0.868  0.38519
## Session1:Group1:Age              -0.0007188  0.0006636  -1.083  0.27877
## Probability1:Session1:Group1:Age -0.0001956  0.0002391  -0.818  0.41320
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

##
## Correlation matrix not shown by default, as p = 16 > 12.
## Use print(x, correlation=TRUE)  or
##     vcov(x)        if you need it

## optimizer (bobyqa) convergence code: 0 (OK)
## Model failed to converge with max|grad| = 0.167262 (tol = 0.002, component 1)
## Model is nearly unidentifiable: very large eigenvalue
##  - Rescale variables?
## Model is nearly unidentifiable: large eigenvalue ratio
##  - Rescale variables?


-- 
Cátia Margarida Ferreira de Oliveira
Psychology PhD Student
Department of Psychology, Room B214
University of York, YO10 5DD

	[[alternative HTML version deleted]]