[R] How to get predictions, plots, etc. from lmer{lme4}

Michael Kubovy kubovy at virginia.edu
Sat Apr 16 15:52:22 CEST 2005


Any comments on the following strategy:
(1) Take the log-odds of the frequencies of b and a (adding 1/6, as 
Tukey recommended) and run lme.
(2) If the estimates from lme are in line with the estimates I got from 
lmer, then use the results from lme.

On Apr 16, 2005, at 9:16 AM, Douglas Bates wrote:

> Michael Kubovy wrote:
>> I'm having trouble using lmer beyond a first step.
>> My data:
>>  > some(exp1B)
>>     sub   ba amplitude   a   b  c  d
>> 2     1 1.00       1.5  65  63  4  8
>> 41    4 1.15       0.0  92  41  3  4
>> 43    4 1.15       3.0  88  48  2  2
>> 63    6 1.00       3.0  50  72  9  9
>> 77    8 1.15       0.0 112  25  2  1
>> 89   10 1.15       0.0  37  33 36 34
>> 126  13 1.15       1.5  80  50  6  4
>> 140  14 1.15       4.5  12 115  6  7
>> 145  20 1.00       0.0  73  65  0  2
>> 147  20 1.00       3.0  63  72  2  3
>> etc.
>> The output:
>>  > summary(exp1B.both.cont.lmer)
>> Linear mixed-effects model fit by maximum likelihood
>> Formula: cbind(b, a) ~ ba + amplitude + (1 | sub)
>>    Data: exp1B
>>       AIC      BIC    logLik MLdeviance REMLdeviance
>>  768.0086 783.2579 -379.0043   758.0086     766.3104
>> Random effects:
>>  Groups   Name        Variance Std.Dev.
>>  sub      (Intercept) 0.18787  0.43344
>>  Residual             6.36355  2.52261
>> # of obs: 156, groups: sub, 13
>> Fixed effects:
>>               Estimate Std. Error  DF  t value  Pr(>|t|)
>> (Intercept)   4.118806   0.397780 153  10.3545 < 2.2e-16
>> ba           -4.205518   0.330010 153 -12.7436 < 2.2e-16
>> amplitude     0.137958   0.023754 153   5.8076 3.547e-08
>> This is just what I need. But I also need predicted values, plots, 
>> etc, and can't figure out how to proceed. Have I overlooked a more 
>> extended document than the rather terse (for me at least) help page?
>
> I regret to say no.  Methods for generalized linear mixed models fit 
> by lmer are still rather rudimentary.
_____________________________
Professor Michael Kubovy
University of Virginia
Department of Psychology
USPS: 	P.O.Box 400400	Charlottesville, VA 22904-4400
Parcels:	Room 102		Gilmer Hall
		McCormick Road	Charlottesville, VA 22903
Office:	B011	+1-434-982-4729
Lab:		B019	+1-434-982-4751
Fax:		+1-434-982-4766
WWW:	http://www.people.virginia.edu/~mk9y/




More information about the R-help mailing list