0

I am trying to use intercept and slope effects in my random effects structure (best model fit) and I get the following errors when I do.

MLM <- lmer((sqrt(degrees))~ Condition*CogLoad + (Video|Subject),
data=dataset)

Warning messages: 1: In commonArgs(par, fn, control, environment()) : maxfun < 10 * length(par)^2 is not recommended. 2: Inoptwrap(optimizer, devfun, getStart(start, rho$lower, rho$pp), :
convergence code 1 from bobyqa: bobyqa -- maximum number of function evaluations exceeded 3: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.0704516 (tol = 0.002, component 1)

When I run just intercept effects I do not get these errors though:

Linear mixed model fit by REML ['lmerMod'] Formula: (sqrt(degrees)) ~ Condition * CogLoad + (1 | Video) + (1 | Subject) Data: dataset

REML criterion at convergence: 183048.6

Scaled residuals: Min 1Q Median 3Q Max -3.7418 -0.6048 -0.0377 0.5293 5.9471

Random effects: Groups Name Variance Std.Dev.
Subject (Intercept) 0.01385 0.1177
Video (Intercept) 0.15947 0.3993
Residual 0.67628 0.8224
Number of obs: 74731, groups: Subject, 27; Video, 11

Fixed effects: Estimate Std. Error t value (Intercept) 2.260374 0.123524 18.299 ConditionMap 0.022744 0.050473 0.451 CogLoadNone 0.047609 0.008152 5.840 ConditionMap:CogLoadNone 0.051590 0.013228 3.900

Correlation of Fixed Effects: (Intr) CndtnM CgLdNn ConditionMp -0.121
CogLoadNone -0.032 0.068
CndtnMp:CLN 0.016 -0.130 -0.507

Would really love some input if someone knows why the first model is not converging appropriately.

Werner Hertzog
  • 2,002
  • 3
  • 24
  • 36
Taylo R Simonson
  • 1
  • 1
  • Googling the first warning (`maxfun < 10 * length(par)^2 is not recommended`) got me [this helpful thread](https://stat.ethz.ch/pipermail/r-sig-mixed-models/2014q2/022084.html), which suggests trying `control=glmerControl(optCtrl=list(maxfun=1e5))` – Gregor Thomas Feb 05 '19 at 15:33
  • I have tried using the control command and it does not help with convergence issues. I end up getting the same errors when I use a control command. – Taylo R Simonson Feb 05 '19 at 20:45

0 Answers0