[R-sig-ME] interactions lmer continuous and categorical fixed factor

[Ben Bolker]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 15-06-01 11:18 AM, Thierry Onkelinx wrote:
> Dear Lotte,
> 
> I assume that the "one p-value for the interaction" is the p-value
> from anova(model). Note that this tests a different hypothesis than
> the hypothesis than summary(model) tests (without reporting
> p-values).
> 
> IMHO, p-values of parameters estimates are not that relevant.
> Confidence intervals of those parameter estimates are much more
> relevant. I'd rather report those.
> 
> Best regards,
> 
> ir. Thierry Onkelinx Instituut voor natuur- en bosonderzoek /
> Research Institute for Nature and Forest team Biometrie &
> Kwaliteitszorg / team Biometrics & Quality Assurance Kliniekstraat
> 25 1070 Anderlecht Belgium
> 
> To call in the statistician after the experiment is done may be no
> more than asking him to perform a post-mortem examination: he may
> be able to say what the experiment died of. ~ Sir Ronald Aylmer
> Fisher The plural of anecdote is not data. ~ Roger Brinner The
> combination of some data and an aching desire for an answer does
> not ensure that a reasonable answer can be extracted from a given
> body of data. ~ John Tukey

  To follow up, I would say that
* you probably *should* be reporting just the p-value for the overall
test of the interaction (i.e. the one returned by anova()).
* if you really want the p-values of the individual parameters, look
at ?pvalues and specifically try the lmerTest package.
* if you're going to start looking at tests for lots of different
levels you might want to consider multiple-comparisons corrections,
see e.g.
http://stats.stackexchange.com/questions/5250/multiple-comparisons-on-a-mixed-effects-model

> 
> 2015-06-01 16:56 GMT+02:00 Lotte Schoot <Lotte.Schoot at mpi.nl>:
> 
>> Hi,
>> 
>> I am using the lmer function in lme4 to test a model like this:
>> 
>> DV ~ factor1 * factor2 (simplified for purposes of illustration,
>> so without random effects structure)
>> 
>> DV = continuous (Reaction time) factor1 = continuous factor2 =
>> categorical (3 levels)
>> 
>> summary(model) will give me output like this:
>> 
>> factor2-level1 * Factor1 = xxx factor2-level2 * Factor1 = xxx 
>> factor2-level3 * Factor1 = xxx
>> 
>> If I try to get p-values for this model, however, I only get one
>> p-value for the interaction factor2 * factor 1.
>> 
>> What do you recommend to report in this case? p-values with
>> corresponding F-values and df, or the t-values found in 
>> summary(model), without any p-values?
>> 
>> Thanks in advance, Lotte
>> 
>> _______________________________________________ 
>> R-sig-mixed-models at r-project.org mailing list 
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>> 
> 
> [[alternative HTML version deleted]]
> 
> _______________________________________________ 
> R-sig-mixed-models at r-project.org mailing list 
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
> 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)

iQEcBAEBAgAGBQJVbHq5AAoJEOCV5YRblxUHYrsIAMORxFwdNAJziLg8VODipKKl
IsAJl/jB7YEX93ec04stEqc4/z8QJ8Aj/tXq8szP67+iunrGKUteeSOoIlclc1yE
DfXJPk0OvsFc4OlsX6xeVya5tUebliVYC/xX97H2XSgwI+iiiXZ24t3BJVpZk1ZD
+rsjPY9+lzCmgH/NnuYZjlxhdrY5PJTkG5eRNRNAiGw9taq+atxGFHVr4mcf0e5I
0x3hGIy1iE2f9x2WFz85RpVOdfjTKFgf5pXeQr5LD+57ixfyR5JWRnyAbDhDSkvn
XtatAOHhhESaGlV47xDTzBnn11mUuqtChSa0x4KGktz1yEgBYLxKs3VArJ1k12Q=
=1BtA
-----END PGP SIGNATURE-----