Timeline for Conflicting results of summary() and anova() for a mixed model with interactions in lmer+lmerTest
Current License: CC BY-SA 3.0
10 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 16, 2017 at 15:47 | comment | added | Henrik |
For a reference on this issue you might want to check out our chapter on mixed models. An easier code to set the contrasts globally is: afex::set_sum_contrasts() (which uses package afex).
|
|
| May 15, 2017 at 19:13 | comment | added | amoeba | For future references, this is very related: stats.stackexchange.com/a/23198. And this too: stats.stackexchange.com/a/279699. | |
| May 15, 2017 at 19:12 | history | edited | amoeba | CC BY-SA 3.0 |
added 8 characters in body
|
| May 15, 2017 at 9:29 | comment | added | amoeba |
Apologies for getting back to this again. In your question you argue that lmerTest::anova uses Type II SS (even with type=3 argument). But in your answer you say that it produces Type III SS (regardless of the default contrasts). I am not sure I understand how it agrees with each other, can you clarify? If it's Type III, as per your answer, then what about the evidence that you presented in your question that it's Type II - namely, comparison with car::Anova? Thanks.
|
|
| Mar 8, 2017 at 3:06 | comment | added | James S. |
My apologies @amoeba. You are right. I have edited the answer to reflect the real discrepancy, which concern the default SS used by summary() when computing P-values.
|
|
| Mar 8, 2017 at 3:01 | history | edited | James S. | CC BY-SA 3.0 |
Clarified answer
|
| Mar 2, 2017 at 20:21 | history | edited | amoeba | CC BY-SA 3.0 |
added 9 characters in body
|
| Mar 2, 2017 at 17:01 | review | Late answers | |||
| Mar 2, 2017 at 17:11 | |||||
| Mar 2, 2017 at 16:43 | vote | accept | James S. | ||
| Mar 2, 2017 at 16:42 | history | answered | James S. | CC BY-SA 3.0 |