-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
1 parent
9e0c866
commit c1e042b
Showing
1 changed file
with
111 additions
and
111 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,111 +1,111 @@ | ||
| #/* Run this first piece of code only if you want to create a PDF report for GitHub/OSF | ||
| rmarkdown::render(input = rstudioapi::getSourceEditorContext()$path, | ||
| output_format = rmarkdown::pdf_document(), | ||
| output_dir = "output", | ||
| knit_root_dir = getwd()) #*/ | ||
| #' --- | ||
| #' author: "" | ||
| #' classoption: "landscape" | ||
| #' --- | ||
|
|
||
| ## MCI_STYLE_NEU MIXED MODELS SCRIPT ## | ||
|
|
||
| # Computes linear mixed-effects regression models with simple contrast coding for the fixed effects of semantics and | ||
| # narrative context (the context being emotionally neutral). Thus, in each model, the estimate of the intercept is the | ||
| # grand mean, while the estimates of the slopes contrast "treatment" levels to their respective reference levels | ||
| # (semantics: violation - intuitive, mci - intuitive; narrative context style: fairytale - normal). The maximal random | ||
| # effects structure is used with all by-participant and by-item random slopes and random intercepts. Correlations | ||
| # between random effects are removed if the model fails to converge with two different numerical optimizers. Planned | ||
| # follow-up contrasts are computed for the main effects and the effects of semantics separately within each type of | ||
| # narrative context style. | ||
|
|
||
| ## SETUP ## ------------------------------------------------------------------------------------------------------------ | ||
|
|
||
| # Load packages | ||
| library(MASS) # Version 7.3-51.6 | ||
| library(lme4) # Version 1.1-23 | ||
| library(lmerTest) # Version 3.1-2 | ||
| library(afex) # Version 0.27-2 | ||
| library(emmeans) # Version 1.4.8 | ||
| library(tidyverse) # Version 1.3.0 | ||
| library(magrittr) # Version 1.5 | ||
|
|
||
| # Load preprocessed data | ||
| a1 <- readRDS("EEG/export/a1.RDS") | ||
|
|
||
| # Remove trials with errors or invalid RTs/ERPs | ||
| a1 %<>% filter(!error) %>% na.omit() | ||
|
|
||
| # Center behavioral ratings (valence and arousal) around 0 | ||
| a1 %<>% mutate(rating_1 = Rating1Resp - 2, rating_2 = Rating2Resp - 2) | ||
|
|
||
| # Define simple contrast coding for context narrative style (normal - fairytale) | ||
| # HO(Intercept): (mu1+mu2)/2 = 0 <-> mu1+mu2 = 0 | ||
| # H0(Slope): -mu1 + mu2 = 0 | ||
| # with mu1 = mean of the normal style and mu2 = mean of the fairytale style | ||
| t(contrasts.style <- t(cbind(c("nor" = -1, "ftl" = 1)))) | ||
| contrasts(a1$style) <- ginv(contrasts.style) | ||
|
|
||
| # Define simple contrast coding for semantics (violation - intuitive, mci - intuitive) | ||
| # H0(Intercept): (mu1+mu2+mu3)/3 = 0 <-> mu1+mu2+mu3 = 0 | ||
| # H0(Slope1): -1*mu1 +1*mu2 + 0*mu3 = 0 | ||
| # H0(Slope2): -1*mu1 +0*mu2 + 1*mu3 = 0 | ||
| # with mu1 = mean of intuitive concepts, mu2 = mean of violations, mu3 = mean of MCIs | ||
| t(contrasts.semantics <- t(cbind(c("int" = -1, "vio" = 1, "mci" = 0), | ||
| c("int" = -1, "vio" = 0, "mci" = 1)))) | ||
| contrasts(a1$semantics) <- ginv(contrasts.semantics) | ||
|
|
||
| ## LINEAR MIXED-EFFECTS MODELS ## -------------------------------------------------------------------------------------- | ||
|
|
||
| # Specifiy settings for optimization in lmer | ||
| control_params <- lmerControl(calc.derivs = FALSE, optimizer = "bobyqa", optCtrl = list(maxfun = 2e5)) | ||
|
|
||
| # LMM for rating 1 | ||
| mod_valence <- lmer_alt(rating_1 ~ semantics*style + (semantics*style||participant) + (semantics*style||item), | ||
| data = a1, control = control_params) | ||
|
|
||
| # LMM for rating 2 | ||
| mod_arousal <- lmer_alt(rating_2 ~ semantics*style + (semantics*style||participant) + (semantics*style||item), | ||
| data = a1, control = control_params) | ||
|
|
||
| # LMM for verb-related N400 | ||
| mod_N400_verb <- lmer_alt(N400_verb ~ semantics*style + (semantics*style||participant) + (semantics*style||item), | ||
| data = a1, control = control_params) | ||
|
|
||
| # LMM for picture-related N400 | ||
| mod_N400_pict <- lmer_alt(N400_pict ~ semantics*style + (semantics*style||participant) + (semantics*style||item), | ||
| data = a1, control = control_params) | ||
|
|
||
| # Create a list of all four models | ||
| models <- list("RATING_1" = mod_valence, "RATING_2" = mod_arousal, | ||
| "N400_VERB" = mod_N400_verb, "N400_PICT" = mod_N400_pict) | ||
|
|
||
| # F-tests (type III tests) | ||
| (tests <- map(models, anova)) | ||
|
|
||
| ## PLANNED FOLLOW-UP CONTRASTS ## -------------------------------------------------------------------------------------- | ||
|
|
||
| # Allow emmeans to compute Satterthwaites p-values | ||
| emm_options(lmer.df = "Satterthwaite", lmerTest.limit = Inf) | ||
|
|
||
| # Follow-up contrasts for the main effect of semantics | ||
| (means.semantics <- map(models,function(x){ | ||
| emmeans(x, trt.vs.ctrl ~ semantics, infer = TRUE, adjust = "bonferroni")$contrasts | ||
| })) | ||
|
|
||
| # Follow-up contrasts for the main effect of context style | ||
| (means.style <- map(models, function(x){ | ||
| emmeans(x, trt.vs.ctrl ~ style, infer = TRUE, adjust = "bonferroni")$contrasts | ||
| })) | ||
|
|
||
| # Follow-up contrasts for semantics within each context style | ||
| (means.nested <- map(models, function(x){ | ||
| emmeans(x, trt.vs.ctrl ~ semantics|style, infer = TRUE, adjust = "bonferroni")$contrasts | ||
| })) | ||
|
|
||
| # Backup results | ||
| save(models, tests, means.semantics, means.style, means.nested, file = "EEG/export/stats.RData") | ||
|
|
||
| # System specs and package versions | ||
| sessionInfo() | ||
|
|
||
| #/* Run this first piece of code only if you want to create a PDF report for GitHub/OSF | ||
| rmarkdown::render(input = rstudioapi::getSourceEditorContext()$path, | ||
| output_format = rmarkdown::pdf_document(), | ||
| output_dir = "output", | ||
| knit_root_dir = getwd()) #*/ | ||
| #' --- | ||
| #' author: "" | ||
| #' classoption: "landscape" | ||
| #' --- | ||
|
|
||
| ## MCI_STYLE_NEU MIXED MODELS SCRIPT ## | ||
|
|
||
| # Computes linear mixed-effects regression models with simple contrast coding for the fixed effects of semantics and | ||
| # narrative context (the context being emotionally neutral). Thus, in each model, the estimate of the intercept is the | ||
| # grand mean, while the estimates of the slopes contrast "treatment" levels to their respective reference levels | ||
| # (semantics: violation - intuitive, mci - intuitive; narrative context style: fairytale - normal). The maximal random | ||
| # effects structure is used with all by-participant and by-item random slopes and random intercepts. Correlations | ||
| # between random effects are removed if the model fails to converge with two different numerical optimizers. Planned | ||
| # follow-up contrasts are computed for the main effects and the effects of semantics separately within each type of | ||
| # narrative context style. | ||
|
|
||
| ## SETUP ## ------------------------------------------------------------------------------------------------------------ | ||
|
|
||
| # Load packages | ||
| library(MASS) # Version 7.3-51.6 | ||
| library(lme4) # Version 1.1-23 | ||
| library(lmerTest) # Version 3.1-2 | ||
| library(afex) # Version 0.27-2 | ||
| library(emmeans) # Version 1.4.8 | ||
| library(tidyverse) # Version 1.3.0 | ||
| library(magrittr) # Version 1.5 | ||
|
|
||
| # Load preprocessed data | ||
| a1 <- readRDS("EEG/export/a1.RDS") | ||
|
|
||
| # Remove trials with errors or invalid RTs/ERPs | ||
| a1 %<>% filter(!error) %>% na.omit() | ||
|
|
||
| # Center behavioral ratings (valence and arousal) around 0 | ||
| a1 %<>% mutate(rating_1 = Rating1Resp - 2, rating_2 = Rating2Resp - 2) | ||
|
|
||
| # Define simple contrast coding for context narrative style (normal - fairytale) | ||
| # HO(Intercept): (mu1+mu2)/2 = 0 <-> mu1+mu2 = 0 | ||
| # H0(Slope): -mu1 + mu2 = 0 | ||
| # with mu1 = mean of the normal style and mu2 = mean of the fairytale style | ||
| t(contrasts.style <- t(cbind(c("nor" = -1, "ftl" = 1)))) | ||
| contrasts(a1$style) <- ginv(contrasts.style) | ||
|
|
||
| # Define simple contrast coding for semantics (violation - intuitive, mci - intuitive) | ||
| # H0(Intercept): (mu1+mu2+mu3)/3 = 0 <-> mu1+mu2+mu3 = 0 | ||
| # H0(Slope1): -1*mu1 +1*mu2 + 0*mu3 = 0 | ||
| # H0(Slope2): -1*mu1 +0*mu2 + 1*mu3 = 0 | ||
| # with mu1 = mean of intuitive concepts, mu2 = mean of violations, mu3 = mean of MCIs | ||
| t(contrasts.semantics <- t(cbind(c("int" = -1, "vio" = 1, "mci" = 0), | ||
| c("int" = -1, "vio" = 0, "mci" = 1)))) | ||
| contrasts(a1$semantics) <- ginv(contrasts.semantics) | ||
|
|
||
| ## LINEAR MIXED-EFFECTS MODELS ## -------------------------------------------------------------------------------------- | ||
|
|
||
| # Specifiy settings for optimization in lmer | ||
| control_params <- lmerControl(calc.derivs = FALSE, optimizer = "bobyqa", optCtrl = list(maxfun = 2e5)) | ||
|
|
||
| # LMM for rating 1 | ||
| mod_rating1 <- lmer_alt(rating_1 ~ semantics*style + (semantics*style||participant) + (semantics*style||item), | ||
| data = a1, control = control_params) | ||
|
|
||
| # LMM for rating 2 | ||
| mod_rating2 <- lmer_alt(rating_2 ~ semantics*style + (semantics*style||participant) + (semantics*style||item), | ||
| data = a1, control = control_params) | ||
|
|
||
| # LMM for verb-related N400 | ||
| mod_N400_verb <- lmer_alt(N400_verb ~ semantics*style + (semantics*style||participant) + (semantics*style||item), | ||
| data = a1, control = control_params) | ||
|
|
||
| # LMM for picture-related N400 | ||
| mod_N400_pict <- lmer_alt(N400_pict ~ semantics*style + (semantics*style||participant) + (semantics*style||item), | ||
| data = a1, control = control_params) | ||
|
|
||
| # Create a list of all four models | ||
| models <- list("RATING_1" = mod_rating, "RATING_2" = mod_rating2, | ||
| "N400_VERB" = mod_N400_verb, "N400_PICT" = mod_N400_pict) | ||
|
|
||
| # F-tests (type III tests) | ||
| (tests <- map(models, anova)) | ||
|
|
||
| ## PLANNED FOLLOW-UP CONTRASTS ## -------------------------------------------------------------------------------------- | ||
|
|
||
| # Allow emmeans to compute Satterthwaites p-values | ||
| emm_options(lmer.df = "Satterthwaite", lmerTest.limit = Inf) | ||
|
|
||
| # Follow-up contrasts for the main effect of semantics | ||
| (means.semantics <- map(models,function(x){ | ||
| emmeans(x, trt.vs.ctrl ~ semantics, infer = TRUE, adjust = "bonferroni")$contrasts | ||
| })) | ||
|
|
||
| # Follow-up contrasts for the main effect of context style | ||
| (means.style <- map(models, function(x){ | ||
| emmeans(x, trt.vs.ctrl ~ style, infer = TRUE, adjust = "bonferroni")$contrasts | ||
| })) | ||
|
|
||
| # Follow-up contrasts for semantics within each context style | ||
| (means.nested <- map(models, function(x){ | ||
| emmeans(x, trt.vs.ctrl ~ semantics|style, infer = TRUE, adjust = "bonferroni")$contrasts | ||
| })) | ||
|
|
||
| # Backup results | ||
| save(models, tests, means.semantics, means.style, means.nested, file = "EEG/export/stats.RData") | ||
|
|
||
| # System specs and package versions | ||
| sessionInfo() | ||
|
|