Hi, I asked this on mixed model mailing list, but that list is not very active,
so I'd like to try the general R mailing list. Sorry if anyone receives the
double post.
Hi, I have a dataset of animals receiving some eye treatments. There are 8
treatments, each animal's right and left eye was measured with some scores
(ranging from 0 to 7) 4 times after treatment. So there are nesting groups eyes
within animal. Dataset attached
> dat<-read.table("dat.txt",sep='\t',header=T,row.names=1)
> dat$id<-factor(dat$id)
> str(dat)
'data.frame': 640 obs. of 5 variables:
$ score: int 0 2 0 7 4 7 0 2 0 7 ...
$ id : Factor w/ 80 levels "1","3","6","10",..: 7 48 66 54 18 26 38 52 39 63
...
$ rep : int 1 1 1 1 1 1 1 1 1 1 ...
$ eye : Factor w/ 2 levels "L","R": 2 2 2 2 2 2 2 2 2 2 ...
$ trt : Factor w/ 8 levels "A","B","C","Control",..: 1 1 1 1 1 1 1 1 1 1 ...
I fit a mixed model using both lmer() from lme4 package and lme() from nlme
package:
> lmer(score~trt+(1|id/eye),dat)
Linear mixed model fit by REML
Formula: score ~ trt + (1 | id/eye)
Data: dat
AIC BIC logLik deviance REMLdev
446.7 495.8 -212.4 430.9 424.7
Random effects:
Groups Name Variance Std.Dev.
eye:id (Intercept) 6.9208e+00 2.630742315798
id (Intercept) 1.4471e-16 0.000000012030
Residual 1.8750e-02 0.136930641909
Number of obs: 640, groups: eye:id, 160; id, 80
> summary(lme(score~trt, random=(~1|id/eye), dat))
Linear mixed-effects model fit by REML
Data: dat
AIC BIC logLik
425.1569 474.0947 -201.5785
Random effects:
Formula: ~1 | id
(Intercept)
StdDev: 1.873576
Formula: ~1 | eye %in% id
(Intercept) Residual
StdDev: 1.896126 0.1369306
As you can see, the variance components estimates of random effects are quite
different between the 2 model fits. From the data, I know that the variance
component for "id" can't be near 0, which is what lmer() fit produced, so I
think the lme() fit is correct while lmer() fit is off. This can also be seen
from AIC, BIC etc. lme() fit has better values than lmer() fit.
I guess this might be due to lmer() didn't converge very well, is there anyway
to adjust to make lmer() converge better to get similar results as lme()?
Thanks
John
score id rep eye trt
1 0 15 1 R A
2 2 72 1 R A
3 0 102 1 R A
4 7 82 1 R A
5 4 28 1 R A
6 7 42 1 R A
7 0 60 1 R A
8 2 79 1 R A
9 0 61 1 R A
10 7 95 1 R A
11 0 15 2 R A
12 2 72 2 R A
13 0 102 2 R A
14 7 82 2 R A
15 4 28 2 R A
16 7 42 2 R A
17 0 60 2 R A
18 2 79 2 R A
19 0 61 2 R A
20 7 95 2 R A
21 0 15 3 R A
22 2 72 3 R A
23 0 102 3 R A
24 7 82 3 R A
25 4 28 3 R A
26 7 42 3 R A
27 0 60 3 R A
28 2 79 3 R A
29 0 61 3 R A
30 7 95 3 R A
31 0 15 4 R A
32 2 72 4 R A
33 0 102 4 R A
34 7 82 4 R A
35 4 28 4 R A
36 7 42 4 R A
37 0 60 4 R A
38 2 79 4 R A
39 0 61 4 R A
40 7 95 4 R A
41 3 15 1 L A
42 0 72 1 L A
43 0 102 1 L A
44 5 82 1 L A
45 0 28 1 L A
46 7 42 1 L A
47 0 60 1 L A
48 0 79 1 L A
49 0 61 1 L A
50 7 95 1 L A
51 3 15 2 L A
52 0 72 2 L A
53 0 102 2 L A
54 5 82 2 L A
55 0 28 2 L A
56 7 42 2 L A
57 0 60 2 L A
58 0 79 2 L A
59 0 61 2 L A
60 7 95 2 L A
61 3 15 3 L A
62 0 72 3 L A
63 0 102 3 L A
64 5 82 3 L A
65 0 28 3 L A
66 7 42 3 L A
67 0 60 3 L A
68 0 79 3 L A
69 0 61 3 L A
70 7 95 3 L A
71 3 15 4 L A
72 0 72 4 L A
73 0 102 4 L A
74 5 82 4 L A
75 0 28 4 L A
76 7 42 4 L A
77 0 60 4 L A
78 0 79 4 L A
79 0 61 4 L A
80 7 95 4 L A
81 4 115 1 R B
82 2 58 1 R B
83 0 113 1 R B
84 7 34 1 R B
85 3 103 1 R B
86 7 40 1 R B
87 0 3 1 R B
88 0 68 1 R B
89 7 57 1 R B
90 3 1 1 R B
91 4 115 2 R B
92 2 58 2 R B
93 0 113 2 R B
94 7 34 2 R B
95 3 103 2 R B
96 7 40 2 R B
97 0 3 2 R B
98 0 68 2 R B
99 7 57 2 R B
100 3 1 2 R B
101 4 115 3 R B
102 2 58 3 R B
103 0 113 3 R B
104 7 34 3 R B
105 3 103 3 R B
106 7 40 3 R B
107 0 3 3 R B
108 0 68 3 R B
109 7 57 3 R B
110 3 1 3 R B
111 5 115 4 R B
112 3 58 4 R B
113 1 113 4 R B
114 7 34 4 R B
115 3 103 4 R B
116 7 40 4 R B
117 0 3 4 R B
118 0 68 4 R B
119 7 57 4 R B
120 3 1 4 R B
121 2 115 1 L B
122 4 58 1 L B
123 1 113 1 L B
124 7 34 1 L B
125 0 103 1 L B
126 7 40 1 L B
127 4 3 1 L B
128 0 68 1 L B
129 7 57 1 L B
130 5 1 1 L B
131 2 115 2 L B
132 4 58 2 L B
133 1 113 2 L B
134 7 34 2 L B
135 0 103 2 L B
136 7 40 2 L B
137 4 3 2 L B
138 0 68 2 L B
139 7 57 2 L B
140 5 1 2 L B
141 2 115 3 L B
142 4 58 3 L B
143 1 113 3 L B
144 7 34 3 L B
145 0 103 3 L B
146 7 40 3 L B
147 4 3 3 L B
148 0 68 3 L B
149 7 57 3 L B
150 5 1 3 L B
151 2 115 4 L B
152 4 58 4 L B
153 1 113 4 L B
154 7 34 4 L B
155 0 103 4 L B
156 7 40 4 L B
157 4 3 4 L B
158 0 68 4 L B
159 7 57 4 L B
160 5 1 4 L B
161 5 35 1 R C
162 0 111 1 R C
163 3 118 1 R C
164 1 38 1 R C
165 5 36 1 R C
166 1 22 1 R C
167 7 67 1 R C
168 6 105 1 R C
169 5 112 1 R C
170 6 81 1 R C
171 5 35 2 R C
172 0 111 2 R C
173 3 118 2 R C
174 1 38 2 R C
175 5 36 2 R C
176 1 22 2 R C
177 7 67 2 R C
178 6 105 2 R C
179 5 112 2 R C
180 6 81 2 R C
181 5 35 3 R C
182 0 111 3 R C
183 3 118 3 R C
184 1 38 3 R C
185 5 36 3 R C
186 1 22 3 R C
187 7 67 3 R C
188 6 105 3 R C
189 5 112 3 R C
190 6 81 3 R C
191 5 35 4 R C
192 0 111 4 R C
193 3 118 4 R C
194 1 38 4 R C
195 5 36 4 R C
196 1 22 4 R C
197 7 67 4 R C
198 6 105 4 R C
199 5 112 4 R C
200 6 81 4 R C
201 6 35 1 L C
202 0 111 1 L C
203 3 118 1 L C
204 6 38 1 L C
205 7 36 1 L C
206 0 22 1 L C
207 1 67 1 L C
208 5 105 1 L C
209 1 112 1 L C
210 0 81 1 L C
211 6 35 2 L C
212 0 111 2 L C
213 3 118 2 L C
214 6 38 2 L C
215 7 36 2 L C
216 0 22 2 L C
217 1 67 2 L C
218 5 105 2 L C
219 1 112 2 L C
220 0 81 2 L C
221 6 35 3 L C
222 0 111 3 L C
223 3 118 3 L C
224 6 38 3 L C
225 7 36 3 L C
226 0 22 3 L C
227 1 67 3 L C
228 5 105 3 L C
229 1 112 3 L C
230 0 81 3 L C
231 6 35 4 L C
232 0 111 4 L C
233 4 118 4 L C
234 6 38 4 L C
235 7 36 4 L C
236 0 22 4 L C
237 1 67 4 L C
238 5 105 4 L C
239 1 112 4 L C
240 0 81 4 L C
241 6 17 1 R D
242 1 51 1 R D
243 0 19 1 R D
244 6 71 1 R D
245 5 110 1 R D
246 6 86 1 R D
247 6 18 1 R D
248 6 44 1 R D
249 7 66 1 R D
250 7 46 1 R D
251 6 17 2 R D
252 1 51 2 R D
253 0 19 2 R D
254 6 71 2 R D
255 5 110 2 R D
256 6 86 2 R D
257 6 18 2 R D
258 6 44 2 R D
259 7 66 2 R D
260 7 46 2 R D
261 6 17 3 R D
262 1 51 3 R D
263 0 19 3 R D
264 6 71 3 R D
265 5 110 3 R D
266 6 86 3 R D
267 6 18 3 R D
268 6 44 3 R D
269 7 66 3 R D
270 7 46 3 R D
271 6 17 4 R D
272 1 51 4 R D
273 0 19 4 R D
274 6 71 4 R D
275 5 110 4 R D
276 6 86 4 R D
277 6 18 4 R D
278 6 44 4 R D
279 7 66 4 R D
280 7 46 4 R D
281 6 17 1 L D
282 7 51 1 L D
283 7 19 1 L D
284 0 71 1 L D
285 2 110 1 L D
286 7 86 1 L D
287 0 18 1 L D
288 7 44 1 L D
289 7 66 1 L D
290 7 46 1 L D
291 5 17 2 L D
292 7 51 2 L D
293 7 19 2 L D
294 0 71 2 L D
295 2 110 2 L D
296 7 86 2 L D
297 0 18 2 L D
298 7 44 2 L D
299 7 66 2 L D
300 7 46 2 L D
301 5 17 3 L D
302 7 51 3 L D
303 7 19 3 L D
304 0 71 3 L D
305 2 110 3 L D
306 7 86 3 L D
307 0 18 3 L D
308 7 44 3 L D
309 7 66 3 L D
310 7 46 3 L D
311 5 17 4 L D
312 7 51 4 L D
313 7 19 4 L D
314 0 71 4 L D
315 2 110 4 L D
316 7 86 4 L D
317 0 18 4 L D
318 7 44 4 L D
319 7 66 4 L D
320 7 46 4 L D
321 7 12 1 R E
322 6 89 1 R E
323 0 87 1 R E
324 6 96 1 R E
325 0 117 1 R E
326 5 106 1 R E
327 1 64 1 R E
328 5 88 1 R E
329 0 90 1 R E
330 7 23 1 R E
331 7 12 2 R E
332 6 89 2 R E
333 0 87 2 R E
334 6 96 2 R E
335 0 117 2 R E
336 5 106 2 R E
337 1 64 2 R E
338 5 88 2 R E
339 0 90 2 R E
340 7 23 2 R E
341 7 12 3 R E
342 6 89 3 R E
343 0 87 3 R E
344 6 96 3 R E
345 0 117 3 R E
346 5 106 3 R E
347 1 64 3 R E
348 5 88 3 R E
349 0 90 3 R E
350 7 23 3 R E
351 7 12 4 R E
352 6 89 4 R E
353 0 87 4 R E
354 6 96 4 R E
355 0 117 4 R E
356 5 106 4 R E
357 1 64 4 R E
358 5 88 4 R E
359 0 90 4 R E
360 7 23 4 R E
361 7 12 1 L E
362 5 89 1 L E
363 0 87 1 L E
364 6 96 1 L E
365 0 117 1 L E
366 6 106 1 L E
367 4 64 1 L E
368 7 88 1 L E
369 0 90 1 L E
370 7 23 1 L E
371 7 12 2 L E
372 6 89 2 L E
373 0 87 2 L E
374 6 96 2 L E
375 0 117 2 L E
376 6 106 2 L E
377 4 64 2 L E
378 7 88 2 L E
379 0 90 2 L E
380 7 23 2 L E
381 7 12 3 L E
382 6 89 3 L E
383 0 87 3 L E
384 6 96 3 L E
385 0 117 3 L E
386 6 106 3 L E
387 4 64 3 L E
388 7 88 3 L E
389 0 90 3 L E
390 7 23 3 L E
391 7 12 4 L E
392 6 89 4 L E
393 0 87 4 L E
394 6 96 4 L E
395 0 117 4 L E
396 6 106 4 L E
397 4 64 4 L E
398 7 88 4 L E
399 0 90 4 L E
400 7 23 4 L E
401 6 11 1 R F
402 2 6 1 R F
403 0 41 1 R F
404 3 75 1 R F
405 5 45 1 R F
406 5 69 1 R F
407 6 120 1 R F
408 7 53 1 R F
409 0 94 1 R F
410 7 39 1 R F
411 7 11 2 R F
412 2 6 2 R F
413 0 41 2 R F
414 3 75 2 R F
415 5 45 2 R F
416 5 69 2 R F
417 6 120 2 R F
418 7 53 2 R F
419 0 94 2 R F
420 7 39 2 R F
421 6 11 3 R F
422 2 6 3 R F
423 0 41 3 R F
424 3 75 3 R F
425 5 45 3 R F
426 5 69 3 R F
427 6 120 3 R F
428 7 53 3 R F
429 0 94 3 R F
430 7 39 3 R F
431 6 11 4 R F
432 2 6 4 R F
433 0 41 4 R F
434 3 75 4 R F
435 5 45 4 R F
436 5 69 4 R F
437 6 120 4 R F
438 7 53 4 R F
439 0 94 4 R F
440 7 39 4 R F
441 6 11 1 L F
442 2 6 1 L F
443 6 41 1 L F
444 6 75 1 L F
445 7 45 1 L F
446 7 69 1 L F
447 4 120 1 L F
448 7 53 1 L F
449 0 94 1 L F
450 7 39 1 L F
451 6 11 2 L F
452 3 6 2 L F
453 6 41 2 L F
454 6 75 2 L F
455 7 45 2 L F
456 7 69 2 L F
457 4 120 2 L F
458 7 53 2 L F
459 0 94 2 L F
460 7 39 2 L F
461 7 11 3 L F
462 3 6 3 L F
463 6 41 3 L F
464 6 75 3 L F
465 7 45 3 L F
466 7 69 3 L F
467 4 120 3 L F
468 7 53 3 L F
469 0 94 3 L F
470 7 39 3 L F
471 7 11 4 L F
472 3 6 4 L F
473 6 41 4 L F
474 6 75 4 L F
475 7 45 4 L F
476 7 69 4 L F
477 4 120 4 L F
478 7 53 4 L F
479 0 94 4 L F
480 7 39 4 L F
481 0 21 1 R G
482 0 20 1 R G
483 0 43 1 R G
484 0 70 1 R G
485 0 101 1 R G
486 0 62 1 R G
487 0 16 1 R G
488 0 55 1 R G
489 0 47 1 R G
490 0 54 1 R G
491 0 21 2 R G
492 0 20 2 R G
493 0 43 2 R G
494 0 70 2 R G
495 0 101 2 R G
496 0 62 2 R G
497 0 16 2 R G
498 0 55 2 R G
499 0 47 2 R G
500 0 54 2 R G
501 0 21 3 R G
502 0 20 3 R G
503 0 43 3 R G
504 0 70 3 R G
505 0 101 3 R G
506 0 62 3 R G
507 0 16 3 R G
508 0 55 3 R G
509 0 47 3 R G
510 0 54 3 R G
511 0 21 4 R G
512 0 20 4 R G
513 0 43 4 R G
514 0 70 4 R G
515 0 101 4 R G
516 0 62 4 R G
517 0 16 4 R G
518 0 55 4 R G
519 0 47 4 R G
520 0 54 4 R G
521 0 21 1 L G
522 0 20 1 L G
523 0 43 1 L G
524 0 70 1 L G
525 0 101 1 L G
526 0 62 1 L G
527 0 16 1 L G
528 0 55 1 L G
529 0 47 1 L G
530 0 54 1 L G
531 0 21 2 L G
532 0 20 2 L G
533 0 43 2 L G
534 0 70 2 L G
535 0 101 2 L G
536 0 62 2 L G
537 0 16 2 L G
538 0 55 2 L G
539 0 47 2 L G
540 0 54 2 L G
541 0 21 3 L G
542 0 20 3 L G
543 0 43 3 L G
544 0 70 3 L G
545 0 101 3 L G
546 0 62 3 L G
547 0 16 3 L G
548 0 55 3 L G
549 0 47 3 L G
550 0 54 3 L G
551 0 21 4 L G
552 0 20 4 L G
553 0 43 4 L G
554 0 70 4 L G
555 0 101 4 L G
556 0 62 4 L G
557 0 16 4 L G
558 0 55 4 L G
559 0 47 4 L G
560 0 54 4 L G
561 6 119 1 R Control
562 7 104 1 R Control
563 0 25 1 R Control
564 0 85 1 R Control
565 0 91 1 R Control
566 7 10 1 R Control
567 7 108 1 R Control
568 0 73 1 R Control
569 6 26 1 R Control
570 5 76 1 R Control
571 6 119 2 R Control
572 7 104 2 R Control
573 0 25 2 R Control
574 0 85 2 R Control
575 0 91 2 R Control
576 7 10 2 R Control
577 7 108 2 R Control
578 0 73 2 R Control
579 6 26 2 R Control
580 5 76 2 R Control
581 5 119 3 R Control
582 7 104 3 R Control
583 0 25 3 R Control
584 0 85 3 R Control
585 0 91 3 R Control
586 7 10 3 R Control
587 7 108 3 R Control
588 0 73 3 R Control
589 6 26 3 R Control
590 5 76 3 R Control
591 5 119 4 R Control
592 7 104 4 R Control
593 0 25 4 R Control
594 0 85 4 R Control
595 0 91 4 R Control
596 7 10 4 R Control
597 7 108 4 R Control
598 0 73 4 R Control
599 6 26 4 R Control
600 5 76 4 R Control
601 4 119 1 L Control
602 6 104 1 L Control
603 3 25 1 L Control
604 4 85 1 L Control
605 5 91 1 L Control
606 0 10 1 L Control
607 7 108 1 L Control
608 0 73 1 L Control
609 0 26 1 L Control
610 7 76 1 L Control
611 4 119 2 L Control
612 6 104 2 L Control
613 3 25 2 L Control
614 4 85 2 L Control
615 5 91 2 L Control
616 0 10 2 L Control
617 7 108 2 L Control
618 0 73 2 L Control
619 0 26 2 L Control
620 7 76 2 L Control
621 5 119 3 L Control
622 6 104 3 L Control
623 3 25 3 L Control
624 4 85 3 L Control
625 5 91 3 L Control
626 0 10 3 L Control
627 7 108 3 L Control
628 0 73 3 L Control
629 0 26 3 L Control
630 7 76 3 L Control
631 5 119 4 L Control
632 6 104 4 L Control
633 3 25 4 L Control
634 4 85 4 L Control
635 5 91 4 L Control
636 0 10 4 L Control
637 7 108 4 L Control
638 0 73 4 L Control
639 0 26 4 L Control
640 7 76 4 L Control
_______________________________________________
[email protected] mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
______________________________________________
[email protected] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
