I have a split split plot design with a main factor(4 levels) and a subfactor (2 levels) and a sub sub factor (2 levels) with 3 replications:
1-How should I define the method on SPSS?
2- for post Hoc tests for the factors with 2 levels, what do I need to do?
Thank you
SPSS is about the worst choice of software that you can make for a split plot design. R is much better, easier too and the full download is free. Repeat the attached search and see what I mean. Best wishes David Booth
David Eugene Booth, is it safe to assume that you would use lmer in R to estimate a linear mixed model? If so, I wonder why one could not estimate the same model using the MIXED command in SPSS. For example, I have reproduced this SAS example...
https://documentation.sas.com/doc/en/pgmsascdc/9.4_3.4/statug/statug_mixed_examples01.htm
...using the following SPSS code:
NEW FILE.
DATASET CLOSE ALL.
DATA LIST FREE / Block A B Y (4F5.0).
BEGIN DATA
1 1 1 56 1 1 2 41
1 2 1 50 1 2 2 36
1 3 1 39 1 3 2 35
2 1 1 30 2 1 2 25
2 2 1 36 2 2 2 28
2 3 1 33 2 3 2 30
3 1 1 32 3 1 2 24
3 2 1 31 3 2 2 27
3 3 1 15 3 3 2 19
4 1 1 30 4 1 2 25
4 2 1 35 4 2 2 30
4 3 1 17 4 3 2 18
END DATA.
MIXED Y BY A B Block
/FIXED = A B A*B
/RANDOM Block A*Block
/PRINT = SOLUTION G R
.
The results match those obtained via SAS PROC MIXED on that web page. I assume one would get the same results via lmer in R. Perhaps one of you useRs could show us the code for this problem so that we can all see how much simpler it is. ;-)
PS- I do realize that the SAS example is a split-plot design, not a split-split-plot design, which is what the OP asked about.
Then Bruce you might try doing a split split plot and see what happens. I noticed that @Jos Feys who can program just about anything agreed with me. Good luck with it. D. Booth
Point me a worked example, and I might have a go at it (time-permitting). ;-)
I had a few moments today, and found an example using SAS PROC MIXED to analyze a split-split-plot design. See the example by data_null__ on this page (currently at the bottom):
- https://communities.sas.com/t5/SAS-Procedures/sas-code-for-split-split-plot-design-CRD/td-p/241957
Using the following SPSS code, I get the same tests of fixed effects that are shown on the web page--see the attached image. Unfortunately, data_null__ did not show the -2LL for the model, so I cannot compare those values.
* Split-split-plot example from:
* https://communities.sas.com/t5/SAS-Procedures/sas-code-for-split-split-plot-design-CRD/td-p/241957.
NEW FILE.
DATASET CLOSE ALL.
DATA LIST LIST /
fabric rubber rep c1t1p1 c1t1p2 c1t2p1 c1t2p2 c2t1p1 c2t1p2 c2t2p1 c2t2p2 (11F5.0).
BEGIN DATA
1 1 1 29 26 34 29 40 35 76 64
1 1 2 84 54 41 40 136 67 74 58
1 1 3 87 66 47 37 70 56 63 55
1 1 4 80 70 53 27 68 37 49 38
1 1 5 79 48 63 53 66 48 43 29
1 2 1 113 81 88 48 84 37 91 57
1 2 2 108 92 71 49 81 72 55 52
1 2 3 123 108 59 49 121 101 98 42
1 2 4 57 43 47 28 102 56 95 55
1 2 5 82 76 80 35 84 73 50 50
2 1 6 6 6 32 16 7 7 34 22
2 1 7 6 6 26 18 12 12 26 24
2 1 8 12 12 52 20 14 12 41 36
2 1 9 9 6 21 16 9 7 15 14
2 1 10 15 13 21 17 13 9 34 15
2 2 6 6 6 15 12 16 7 41 19
2 2 7 6 6 25 19 8 6 23 23
2 2 8 7 7 50 32 20 19 47 33
2 2 9 13 8 22 20 23 23 64 33
2 2 10 6 6 27 19 16 14 74 31
3 1 11 76 68 34 22 27 20 118 29
3 1 12 16 11 24 15 43 31 49 36
3 1 13 61 50 46 24 36 25 62 21
3 1 14 41 23 29 19 25 13 66 43
3 1 15 8 6 14 11 7 7 28 21
3 2 11 13 11 53 26 37 24 59 52
3 2 12 22 19 22 22 99 17 45 38
3 2 13 22 18 48 18 76 48 30 25
3 2 14 58 32 30 26 72 64 94 43
3 2 15 7 7 11 11 16 12 19 13
END DATA.
* Restructure from WIDE to LONG file.
VARSTOCASES
/ID=id
/MAKE Y FROM c1t1p1 TO c2t2p2
/INDEX=cure(2) temp(2) piece(2)
/KEEP=fabric rubber rep
/NULL=KEEP.
* My translation of the PROC MIXED code.
MIXED Y BY fabric rubber rep cure temp piece
/FIXED =
fabric rubber rubber*fabric
cure cure*fabric cure*rubber cure*rubber*fabric
temp temp*fabric temp*rubber temp*fabric*rubber
temp*cure temp*cure*fabric temp*cure*rubber
temp*cure*fabric*rubber
/RANDOM rep | SUBJECT(fabric)
/RANDOM rubber*rep | SUBJECT(fabric)
/RANDOM cure*rubber*rep | SUBJECT(fabric)
/RANDOM temp*cure*rubber*rep | SUBJECT(fabric)
/PRINT = SOLUTION G R
/EMMEANS=TABLES(rubber*fabric) COMPARE(rubber)
/EMMEANS=TABLES(rubber*fabric) COMPARE(fabric)
.
Dear Colleagues
The report in the following link may be useful for SAS software users
Technical Report Procedure for Analysis of Variance According to Split-split ...
Good Luck
Dear Colleagues
The report in the following link may be useful to SPSS software users to analyze split-split plot design.
Technical Report Procedure for Analysis of Variance According to Split-split ...
Good Luck
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation