questions about mixed logit models with R
Yet another occasion when I said I would cc: the list and forgot to. ---------- Forwarded message ---------- From: Douglas Bates <bates at stat.wisc.edu> Date: Tue, Jan 17, 2012 at 10:38 AM Subject: Re: questions about mixed logit models with R To: Angel Tabullo <angeltabullo at yahoo.com> I suggest that you send such a request to the R-SIG-Mixed-Models at R-project.org mailing list, which I am copying on this reply. ?There are several experts who read that list and may be able to provide help more readily than I can.
On Tue, Jan 17, 2012 at 9:04 AM, Angel Tabullo <angeltabullo at yahoo.com> wrote:
Dear professor Bates My name's Angel Tabullo, I'm a phd student and I'm currently working on neurolinguistics and experimental psychology. I'm trying to run a mixed effects model analysis on some behavioral data with R, but I'm quite new to this kind of statistics and I'm having trouble to interpret the results. I'm writing to you because I found your tutorial in the web and it was very helpful. ?I also wrote to the R-lang mailing list. I will be very thankful for any advice you could give me in this matter. In my experiment, subjects were exposed to artificial languages with different word orders (two of them frequent among world languages: SOV, SVO and two of them infrequent: VSO, OSV). After training, subject had to classify new sentences as "correct" or incorrect, according to what they have learned. Sentences could either be correct, contain a syntax violation or a semantic violation (mismatch between a scene and the sentences describing it). Dependent variables were response latency and accuracy (right or wrong answer). I'm trying to analyze the accuracy (1 = right answer, 0 = wrong answer) data using a mixed logit model with "word order (OSV, SVO, SOV, VSO)" and "type of sentence" (correct, semantic violation, syntax violation) as fixed factors, and subject as a random factor. Word order is a between subjects variable, while type of sentences is a repeated measures factor. My questions are: 1) In order to contrast each level of each factor with all the others, as well as their interactions: should I ran different models changing the reference category? Does this mean I should run 4 x 3 = 12 models? 2) Would it be correct to compare interaction levels with post hoc Tukey contrasts (for instance: OSV - correct vs. OSV semantic violation, SVO correct vs. OSV correct and so on?). 3) How do I interpret a significant interaction? For instance: ModeloAngel = lmer(respuest=="1" ~ grupo * tipoF + (1|sujeto), data=DatosAngel, family="binomial") Fixed effects: ? ? ? ? ? ? ? ? ? ? ? ?Estimate Std. Error z value Pr(>|z|) (Intercept) ? ? ? ? ? ? 1.79585 ? ?0.19196 ? 9.356 ?< 2e-16 *** grupoOSV ? ? ? ? ? ? ? ?0.25816 ? ?0.26740 ? 0.965 ? 0.3343 grupoSOV ? ? ? ? ? ? ? ?0.70875 ? ?0.29315 ? 2.418 ? 0.0156 * grupoSVO ? ? ? ? ? ? ? ?0.59607 ? ?0.26769 ? 2.227 ? 0.0260 * tipoFVsemanti ? ? ? ? ?-1.01756 ? ?0.14765 ?-6.892 5.51e-12 *** tipoFVsintact ? ? ? ? ?-1.46088 ? ?0.14566 -10.029 ?< 2e-16 *** grupoOSV:tipoFVsemanti -0.29214 ? ?0.20841 ?-1.402 ? 0.1610 grupoSOV:tipoFVsemanti -0.39714 ? ?0.23265 ?-1.707 ? 0.0878 . grupoSVO:tipoFVsemanti ?0.03181 ? ?0.21459 ? 0.148 ? 0.8821 grupoOSV:tipoFVsintact ?0.83284 ? ?0.21107 ? 3.946 7.95e-05 *** grupoSOV:tipoFVsintact ?0.42079 ? ?0.23408 ? 1.798 ? 0.0722 . grupoSVO:tipoFVsintact ?0.16667 ? ?0.21136 ? 0.789 ? 0.4304 If the reference levels are VSO and "correct": does this mean that performance of OSV in syntax violations trials is better than that of VSO in syntax violation trials. Or does this mean that OSV - syntax violations performance is better than VSO - "correct" performance? Thank you again for your kind attention, I look forward to your answer.