1
$\begingroup$

I conducted a risk factor analysis for which I got some warning messages.

Data:

  • bf= if the piglet got the disease=1 if the piglet didn't=0
  • y= year (categorical data)
  • SOW= random effect of the mother because the piglet can come from the same mother
  • FARM= random effect of the farm because the piglet can come from the same farm

I got this result below with some of my univariate analyses. I wonder what the nearly unidentifiable model and large eigenvalue mean and whether switching from glmer to bglmer to make all my analyses and solve the problem is a good idea? (add a weak prior on my fixed effect?) Most of my results with bglmer remained more or less the same except for some univariate analyses for which the results changed a lot

WARNINGS
1: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  :
  Model failed to converge with max|grad| = 0.251496 (tol = 0.001, component 4)
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  :
  Model is nearly unidentifiable: very large eigenvalue
 - Rescale variables?;Model is nearly unidentifiable: large eigenvalue ratio
 - Rescale variables?


summary(mod1)
Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) [
glmerMod]
 Family: binomial  ( logit )
Formula: bf ~ Y + (1 | SOW) + (1 | FARM)
   Data: pig
Control: glmerControl(optimizer = "bobyqa", optCtrl = list(maxfun = 2e+07))

     AIC      BIC   logLik deviance df.resid 
  2062.1   2147.5  -1018.1   2036.1     5234 

Scaled residuals: 
    Min      1Q  Median      3Q     Max 
-1.8009 -0.0251 -0.0217 -0.0152  3.6741 

Random effects:
 Groups Name        Variance  Std.Dev. 
 SOW    (Intercept) 4.727e+01 6.875e+00
 FARM   (Intercept) 1.542e-14 1.242e-07

Number of obs: 5247, groups:  SOW, 2051; FARM, 155

Fixed effects:

              Estimate Std. Error z value Pr(>|z|) 

(Intercept) -3.5037396  0.0006667   -5255  < 2e-16 *** 
Y2005       -4.8349015  1.3343113      -4 0.000291 ***
Y2006       -5.3733263  0.0006667   -8060  < 2e-16 ***  
Y2007       -4.5012418  0.4806619      -9  < 2e-16 ***
Y2008       -4.5533141  0.7849521      -6 6.60e-09 ***
Y2009       -4.1169916  0.5237818      -8 3.84e-15 ***
Y2010       -3.9853691  0.5144414      -8 9.41e-15 ***    
Y2011       -3.7500203  0.4594190      -8 3.28e-16 ***    
Y2012       -3.9618617  0.4245563      -9  < 2e-16 *** 
Y2013       -3.6838128  0.4442975      -8  < 2e-16 *** 
Y2014       -3.7898477  0.4944937      -8 1.80e-14 ***

---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

[snip]
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.