ID
int64 1
1.07k
| Comment
stringlengths 8
1.13k
| Code
stringlengths 10
4.28k
| Label
stringclasses 4
values | Source
stringlengths 21
21
| File
stringlengths 4
82
|
|---|---|---|---|---|---|
401 | Analysis Linear Probability Model on Impact Factor | fit_ev_if <- lm(X1.year.Impact.Factor ~ diff_ev + factor(year), data=FullDataset) summary(fit_ev_if) # show results fit_ex_if <- lm(X1.year.Impact.Factor ~ diff_ex + factor(year), data=FullDataset) summary(fit_ex_if) # show results | Statistical Modeling | https://osf.io/jh47m/ | Gatekeeper analysis.r |
402 | Tobit Model as robusntess check assign value 0 to Impact factors for unpublished studies | FullDataset$X1.year.Impact.Factor[FullDataset$published == "No"] = 0 tobit_ev_if <- censReg( X1.year.Impact.Factor ~ diff_ev + factor(year), data = FullDataset, left = 0) summary(tobit_ev_if) tobit_ex_if <- censReg( X1.year.Impact.Factor ~ diff_ex + factor(year), data = FullDataset, left = 0) summary(tobit_ex_if) | Statistical Modeling | https://osf.io/jh47m/ | Gatekeeper analysis.r |
403 | initialize standard error function | stderr <- function(x, na.rm=FALSE) { if (na.rm) x <- na.omit(x) sqrt(var(x)/length(x)) } | Data Variable | https://osf.io/wyrav/ | AnalysisScript-ResearchQuestion1.R |
404 | 2019 data zstandardize mood scores for each person | dat2019_scaled <- dat2019_complete %>% dplyr::select(identity_id, country, doy, mood, weekday) %>% group_by(identity_id) %>% mutate_at(vars(-identity_id,-doy,-country, -weekday), scale) | Data Variable | https://osf.io/wyrav/ | AnalysisScript-ResearchQuestion1.R |
405 | Run regression anlayses and save as table (make one table with mood and depression see depression section) | m1_het <- lm(mood_change ~ gender + age + education_level + race, data = dat_het_prepost) m2_het <- lm(depression_change ~ gender + age + education_level + race, data = dat_het_prepost) tab_model(m1_het, m2_het, show.est = F, show.std = T, digits = 3, file="Mood_Depression_PrePost_Demog_std.doc") | Statistical Modeling | https://osf.io/wyrav/ | AnalysisScript-ResearchQuestion1.R |
406 | Run multilevel model analyses predicting mood scores from the the dummycoded month variable | dat2020_complete$month <- as.factor(dat2020_complete$month) summary(m1 <- lmer(depression ~ month + (1|identity_id), data = subset(dat2020_complete, country == "United States"))) summary(m2 <- lmer(depression ~ month + (1|identity_id), data = subset(dat2020_complete, country == "Germany"))) summary(m3 <- lmer(depression ~ month + (1|identity_id), data = subset(dat2020_complete, country == "United Kingdom"))) tab_model(m1,m2,m3, show.est = F, show.std = T, digits = 3, file="Depression_2020_Months.doc") | Statistical Modeling | https://osf.io/wyrav/ | AnalysisScript-ResearchQuestion1.R |
407 | Step 2: Create onehot encodings (dummy variables) | testData3 <- predict(dummies_model, testData2) testData3 <- predict(dummies_model, testData2) testData3 <- predict(dummies_model, testData2) testData3 <- predict(dummies_model, testData2) testData3 <- predict(dummies_model, testData2) | Data Variable | https://osf.io/wyrav/ | ThesisMLROCCode.R |
408 | function to generate output csv files for a list of response IDs inputs: dataset, list of response IDs outputs: csv files named with response IDs | multi_response <- function(dataset, IDs_list) { for (i in IDs_list) { participant_response <- single_response(dataset, i) csv_name <- str_c(i, ".csv", sep = "") write.csv(participant_response, csv_name) } } | Visualization | https://osf.io/3bn9u/ | 4_4_identification.R |
409 | Means and SDs of proportional looking time to novel object in the preferential looking phase condition 1 | mean(dfX_base$PrefLook_LT_Object_Nov_PROP[which(dfX_base$Condition == "Con1")], na.rm = TRUE) | Data Variable | https://osf.io/yfegm/ | PREPROC_script_Experiment1.r |
410 | calculate mode of a vector | mode.knn = function(x){ uniq.x = unique(x) uniq.x = uniq.x[which(!is.na(uniq.x))] knn = uniq.x[which.max(tabulate(match(x, uniq.x)))] return(knn) } | Statistical Modeling | https://osf.io/b7krz/ | helper_variables.R |
411 | find [k] neigherst neigbors and impute by their mode | impute.knn = function(y, k){ t = which(is.na(y)) if(length(t) == 0){ return(y) }else{ is = 1:length(t) for(i in is){ if(i > k){ look.at = y[(t[i]-k):(t[i]+k)] y[t[i]] = mode.knn(look.at) } if(i <= k){ look.at = y[1:(t[i]+k)] y[t[i]] = mode.knn(look.at) } } return(y) } } | Statistical Modeling | https://osf.io/b7krz/ | helper_variables.R |
412 | get categorical data information then look up the variables in the process in categorical_data so we can get the number of categories for each variable in the process this would be achieved by categories[cat_indices[i]] for variable i in the process | categories <- as.integer(mplus.get.group.attribute(file,'categorical_data','categories')) catvars <- mplus.get.group.attribute(file,'categorical_data','var_names') vartypes <- as.integer(mplus.get.group.attribute(file,'categorical_data','vtype')) if (series) { categories <- as.integer(mplus.get.group.attribute(file,'categorical_data','categories')) catvars <- mplus.get.group.attribute(file,'categorical_data','var_names') vartypes <- as.integer(mplus.get.group.attribute(file,'categorical_data','vtype')) if (series) { | Data Variable | https://osf.io/nxyh3/ | mplus.R |
413 | get indices and names of the variables in the series | var_indices <- mplus.get.group.attribute(file,cstr2,'var_indices') var_names <- mplus.get.group.attribute(file,cstr2,'var_names') cat_indices <- pmatch(var_names, catvars, nomatch=0) cat_indices <- as.integer(cat_indices) var_indices <- mplus.get.group.attribute(file,cstr2,'var_indices') var_names <- mplus.get.group.attribute(file,cstr2,'var_names') cat_indices <- as.integer(pmatch(var_names, catvars, nomatch=0)) | Data Variable | https://osf.io/nxyh3/ | mplus.R |
414 | extract correlations between selfreported scores (observational level!) | self.reported.scores = c("duty", "intellect", "mating", "positivity", "sociality") l = "mating" for(i in self.reported.scores[1:length(self.reported.scores)-1]){ data_phi = ftable(data[,c(paste0("diamonds_", l), paste0("diamonds_", self.reported.scores[which(self.reported.scores == i)+1]))]) print(paste(l, "and", self.reported.scores[which(self.reported.scores == i)+1])) print(phi(data_phi)) } | Data Variable | https://osf.io/b7krz/ | 02_IML_LASSO_FeatImp.R |
415 | generate unique study identifier, format and annotate data | d$id <- paste0(d$First_Author,", ", d$Year) d$Year <- as.numeric(substr(d$Year,1,4)) d$ids <- NA for(i in d$id) d[d$id == i,"ids"] <- 1:dim(d[d$id == i,"ids"])[1] d$ids <- as.character(d$ids) d$Age <- (as.numeric(d$Age_M) - 30) / 20 d$energy_renew <- d$energy_renew1 + d$energy_renew2 find_cCode <- Vectorize(function(i) which(unlist(Map(function(x) sum(d[i,grepl("SVS",names(d))] - f[x,grepl("SVS",names(f))]), 1:dim(f)[1])) == 0)) d$Ccode <- f$Ccode[find_cCode(1:dim(d)[1])] | Data Variable | https://osf.io/qxf5t/ | TSST_Meta.R |
416 | to make sure that effects are tested from complex to simple, we reverse the order of the vector containing the to be tested effects: | effects_to_test = rev(effects_to_test) | Data Variable | https://osf.io/dpkyb/ | create_model_formulas.R |
417 | combine model matrix and data to include participant id and DV: | new_data = data.frame(dplyr::select(data, contains(c(group, DV_variables, by))), model_matrix) | Data Variable | https://osf.io/dpkyb/ | create_model_formulas.R |
418 | check if retention is a number from 0 to 1 | if (retention < 0 || retention > 1) { stop('Retention value is not a number from 0 to 1.') } | Data Variable | https://osf.io/a9bv6/ | sanet_2.R |
419 | construct a maximal glmer() model This model contains code for Ambiguity effect, plus random effects by participants and items. | Acc.Modality.max <- glmer(Correct ~ 1 + Ambiguity.code + (1 + Ambiguity.code | ï..ID) + (1 | Item), data = Data.DisambTask.AmbvsUnamb, family = "binomial", control = glmerControl(optimizer ="bobyqa")) | Statistical Modeling | https://osf.io/m87vg/ | Exp2_BehaviouralAnalyses_Code.R |
420 | create a histogram of the residuals | hist(rawResiduals) hist(invResiduals) hist(logResiduals) | Visualization | https://osf.io/m87vg/ | Exp2_BehaviouralAnalyses_Code.R |
421 | construct a maximal lmer() model This model contains a fixed withinsubjects effect of Ambiguity (effectcoded with 0.5 amb) plus random effects by participants and items. | RT.max <- lmer(logRT ~ 1 + Ambiguity.code + (1 + Ambiguity.code | ï..ID) + (1 | Item), data = Data.DisambTask.AmbvsUnamb, REML=FALSE) | Statistical Modeling | https://osf.io/m87vg/ | Exp2_BehaviouralAnalyses_Code.R |
422 | construct a maximal glmer() model This model contains codes for Modality effects, plus random effects by participants and items. | Acc.Modality.max <- glmer(Correct ~ 1 + Modality.code1 + Modality.code2 + (1 + Modality.code1 + Modality.code2 | ï..ID) + (1 | Item), data = Data.DisambTask.Amb, family = "binomial", control = glmerControl(optimizer ="bobyqa")) Acc.Modality.max <- glmer(True.Positive ~ 1 + Modality.code2 + (1 + Modality.code2 | ID) + (1 | Item), data = Data.ListAndRead, family = "binomial", control = glmerControl(optimizer ="bobyqa")) Acc.Modality.max <- glmer(True.Positive ~ 1 + Modality.code3 + (1 + Modality.code3 | ID) + (1 | Item), data = Data.ListAndRSVP, family = "binomial", control = glmerControl(optimizer ="bobyqa")) Acc.Modality.max <- glmer(True.Positive ~ 1 + Modality.code3 + (1 + Modality.code3 | ID) + (1 | Item), data = Data.ReadAndRSVP, family = "binomial", control = glmerControl(optimizer ="bobyqa")) | Statistical Modeling | https://osf.io/m87vg/ | Exp2_BehaviouralAnalyses_Code.R |
423 | construct a maximal glmer() model This model contains codes for run effects, plus random effects by participants and items. | Acc.Run.max <- glmer(Correct ~ 1 + Run.code1 + Run.code2 + (1 + Run.code1 + Run.code2 | ï..ID) + (1 | Item), data = Data.DisambTask.Amb, family = "binomial", control = glmerControl(optimizer ="bobyqa")) Acc.Run.max <- glmer(True.Positive ~ 1 + Run.code1 + Run.code2 + (1 + Run.code1 + Run.code2 | ID) + (1 | Item), data = Data.RecMem, family = "binomial", control = glmerControl(optimizer ="bobyqa")) | Statistical Modeling | https://osf.io/m87vg/ | Exp2_BehaviouralAnalyses_Code.R |
424 | run a onesample ttest comparing accuracy to chance level (0.5) | t.test(Data.List$True.Positive, mu=0.5) t.test(Data.Read$True.Positive, mu=0.5) t.test(Data.RSVP$True.Positive, mu=0.5) | Statistical Test | https://osf.io/m87vg/ | Exp2_BehaviouralAnalyses_Code.R |
425 | abbreviate first names to match with abbreviations in publication list | stats_persons <- lapply(stats_persons, function(x){ x[2]<-str_sub(x[2], 1, 1) x }) stats_persons <- lapply(stats_persons, function(x)paste(x[1], x[2], sep = ",")) | Data Variable | https://osf.io/rf6zu/ | scrape_web_pages.R |
426 | Load the data | load('../Relative Effectiveness Data - final.Rdata') | Data Variable | https://osf.io/3aryn/ | 6speciesismgraphs.R |
427 | Print a txt and csv file with the results ' ' @param stats_to_print the tibble/table to print ' @param name_file name of the file to write ' ' @return 2 files with the results (txt and csv) | print_result <- function(stats_to_print, name_file){ knitr::kable( stats_to_print, format = "rst") %>% cat( file = here('results', str_c(name_file, '.txt', sep = "")), sep = "\n") stats_to_print %>% write_csv(file = here('results', str_c(name_file, '.csv', sep = ""))) } | Visualization | https://osf.io/4fvwe/ | print_result.R |
428 | convert degrees of freedom into numeric variables and store in new variable | ScienceStatus$df.numerator.value = as.numeric(as.character(ScienceStatus$df.numerator)) ScienceStatus$df.denominator.value = as.numeric(as.character(ScienceStatus$df.denominator)) | Data Variable | https://osf.io/he8mu/ | Study2_Load_Analysis_Post_Review_11-14-16_Final.R |
429 | Assigning numeric values to variable labels which were stored in CSV | ScienceStatus$coding.difficulty_r = revalue(ScienceStatus$coding.difficulty, c("Very Easy"="1", "Moderately Easy"="2", "Slightly Easy"="3", "Neither Difficult nor Easy"="4", "Slightly Difficult"="5", "Moderately Difficult"="6", "Very Difficult"="7")) ScienceStatus$coding.difficulty_r = as.numeric(as.character(ScienceStatus$coding.difficulty_r)) | Data Variable | https://osf.io/he8mu/ | Study2_Load_Analysis_Post_Review_11-14-16_Final.R |
430 | Calcuate EXACT pvalue from the stats reported in the paper (Posthoc) | ScienceStatus_J$Calc.Pvalue<-(rowSums(cbind(ScienceStatus_J$T.pvalue_calc, ScienceStatus_J$F.pvalue_calc, ScienceStatus_J$rg.pvalue_calc, ScienceStatus_J$r.pvalue_calc,ScienceStatus_J$chi.pvalue_calc), na.rm = TRUE) + ifelse(is.na(ScienceStatus_J$T.pvalue_calc) & is.na(ScienceStatus_J$F.pvalue_calc) & is.na(ScienceStatus_J$rg.pvalue_calc) & is.na(ScienceStatus_J$r.pvalue_calc) & is.na(ScienceStatus_J$chi.pvalue_calc), NA, 0)) | Statistical Test | https://osf.io/he8mu/ | Study2_Load_Analysis_Post_Review_11-14-16_Final.R |
431 | PCurve Graph Calculated from statistic and DF By Paper | P.By.Paper<-ddply(ScienceStatus_SP, .(article.id,yearcat), summarize, Pmean.calc = stouffer.P(Calc.Pvalue.Clean), PMedian.calc = median(Calc.Pvalue.Clean)) | Visualization | https://osf.io/he8mu/ | Study2_Load_Analysis_Post_Review_11-14-16_Final.R |
432 | RIndex Calcuate Mean, Median, Peak | Rindex.Reults<-BCa.Boot.CI(ScienceStatus_SP,Calc.Z,r.index.calc.boot,LogT=FALSE,splitter=yearcat, StatType="AR") Rindex.Reults | Statistical Modeling | https://osf.io/he8mu/ | Study2_Load_Analysis_Post_Review_11-14-16_Final.R |
433 | FIGURE 3 Length in phonemes RD fit a model with fixed effects and with the random effect of subjects on intercepts and the slopes for length | Cue.Subj.lmer = glmer(ACC ~ CueCondition + zLengthPh + zFreq + (1+zLengthPh|Subject), data = NamingData, family = "binomial", control = glmerControl(optimizer="bobyqa")) summary(Cue.Subj.lmer) | Statistical Modeling | https://osf.io/bfq39/ | Code_LMMs_Code_BestPractice_Example.R |
434 | extract fitted values and align with original data for plotting | a<-fitted.values(Cue.Subj.lmer) length(a) Dat<-na.omit(NamingData) #remove NAs from original data so aligns with model values Dat$fitted<-a names(Dat) a<-fitted.values(Cue.Subj.lmer) length(a) Dat<-na.omit(NamingData) #remove NAs from original data so aligns with model values Dat$fitted<-a names(Dat) | Visualization | https://osf.io/bfq39/ | Code_LMMs_Code_BestPractice_Example.R |
435 | FIGURE 4a Plot average slope for effect of Frequency | p <- ggplot(data=Dat,aes(x=zFreq, y=fitted)) p <- p + geom_smooth(method = "glm") p <- p + xlab("Frequency (z score)") + ylab("Accuracy (model fit)") p <- p + ggtitle("4a) Average (group) effect of Frequency") p <- p + ylim(0,1) p <- p + theme_bw() + theme(text=element_text(size=12)) p | Visualization | https://osf.io/bfq39/ | Code_LMMs_Code_BestPractice_Example.R |
436 | Main effects model for predicting inversion questions during alignment sessions. | inversion_main_effects_model<-glmer(inversion_dv ~ prime_type + WMC + modality + inversion_production_pre + (1 + prime_type + trial_order | subject) + (1|verb), priming, family="binomial"(link="logit"),glmerControl(optimizer = "bobyqa", optCtrl = list(maxfun = 100000))) summary(inversion_main_effects_model) | Statistical Modeling | https://osf.io/f3qrh/ | Kim Skalicky and Jung - R syntax.R |
437 | Odds ratios for main effects model | inversion.me.CI<-confint(inversion_main_effects_model, parm="beta_", level=0.90, method="Wald") inversion.me.tab <- cbind(est = fixef(inversion_main_effects_model),inversion.me.CI) inversion.me.tab <- exp(inversion.me.tab) inversion.me.tab <- as.data.frame(inversion.me.tab) inversion.me.tab <- rownames_to_column(inversion.me.tab) inversion.me.tab | Statistical Modeling | https://osf.io/f3qrh/ | Kim Skalicky and Jung - R syntax.R |
438 | Visualize the interaction between prime type and modality. save the specific effect to a variable | effect2 <-effect("prime_type*modality",inversion_full_model_sig_int_only) summary(effect2) plot(effect2) | Visualization | https://osf.io/f3qrh/ | Kim Skalicky and Jung - R syntax.R |
439 | post hoc analyses to compare all levels of test order between groups indirect production main effect only | ind_prod_me_between_groups <- emmeans(indirect_production_me,c("test_order"), type = "response") pairs(ind_prod_me_between_groups, reverse = F, type = 'response', adjust = 'none') plot(ind_prod_me_between_groups) | Statistical Test | https://osf.io/f3qrh/ | Kim Skalicky and Jung - R syntax.R |
440 | Test assumptions MANOVA Test whether residuals are normally distributed | df$pc1.residuals = lm(pc1~condition.socaccount, data=df)$residuals df$pc2.residuals = lm(pc2~condition.socaccount, data=df)$residuals df$pc3.residuals = lm(pc3~condition.socaccount, data=df)$residuals df$pc4.residuals = lm(pc4~condition.socaccount, data=df)$residuals shapiro.test(df$pc1.residuals) shapiro.test(df$pc2.residuals) shapiro.test(df$pc3.residuals) shapiro.test(df$pc4.residuals) | Statistical Test | https://osf.io/qj86m/ | 9_manova_fda_socaccount.R |
441 | Only include tweets with at least 3 words | final_twitter_data_ex <- final_twitter_data %>% filter(wc>=3) final_twitter_data_ex <- final_twitter_data_ex %>% filter(bryscore>=0.0001) | Data Variable | https://osf.io/qxwsz/ | anxiety_abstraction_s2.R |
442 | Means and SDs in feartweets and anxietytweets | final_twitter_data_ex %>% group_by(fearVSanx) %>% summarise_at(vars(bryscore,i,we,they,focusfuture), list(mean=mean, sd=sd)) | Data Variable | https://osf.io/qxwsz/ | anxiety_abstraction_s2.R |
443 | Independent samples ttest: Brysbaert concreteness score as DV | t.test_bci <- final_twitter_data_ex %>% rstatix::t_test(bryscore ~ fearVSanx, var.equal = TRUE, detailed=TRUE) %>% rstatix::add_significance() t.test_bci | Statistical Test | https://osf.io/qxwsz/ | anxiety_abstraction_s2.R |
444 | convert age from months to years | demog.tab$interview_age <- demog.tab$interview_age / 12 | Data Variable | https://osf.io/5y27d/ | load_all_tables.R |
445 | multinomial CI (these are simulataneous CI although assume full pooling) | obs_p_CI <- DescTools::MultinomCI(pl_d$N, conf.level=1-0.05/2, method="goodman") *4 pl_d$obs_p_lb <- obs_p_CI[,2] pl_d$obs_p_ub <- obs_p_CI[,3] pl_d %>% mutate(resp_true = ifelse(k>=4,1,0), fill_group = str_c(item, resp_true, type), line_group = str_c(item, type), item = factor(item, levels=c("true","false"))) %>% ggplot(aes(x=k,y=obs_p))+ facet_grid(item ~ type) + nice_theme + geom_col(aes(fill=fill_group))+ geom_errorbar(aes(y=obs_p,ymin=obs_p_lb, ymax=obs_p_ub),color="dark grey",width=0,lwd=0.8)+ | Statistical Modeling | https://osf.io/nd9yr/ | make_fig1_main_text.R |
446 | Mratio posterior density plot | log_mu_sci <- unlist(c(output[,grep('mu_logMratio_1',varnames(output))])) log_mu_cov <- unlist(c(output[,grep('mu_logMratio_2',varnames(output))])) d_groupM <- data.frame(logM =c(log_mu_sci,log_mu_cov), type=c(rep('science',length(log_mu_sci)),rep('Covid-19',length(log_mu_sci)))) d_groupM %>% mutate(M=exp(logM)) %>% ggplot(aes(x=M,y=type,fill=type))+ nice_theme + geom_vline(xintercept = 1, lty=2,size=0.4)+ stat_halfeye(.width = c(.95, .95),aes(fill=type),alpha=0.8)+ | Visualization | https://osf.io/nd9yr/ | make_fig1_main_text.R |
447 | Convert `campaign_week` from a character string to a factor with 11 levels to order weeks chronologically. | Facebook_Users$campaign_week <- factor(Facebook_Users$campaign_week, levels = c("Week 1", "Week 2", "Week 3", "Week 4", "Week 5", "Week 6", "Week 7", "Week 8", "Week 9", "Week 10", "Week 11", "Election Day")) | Data Variable | https://osf.io/3fnjq/ | facebook_users.R |
448 | Convert `candidate_page` from a character string to a factor with 3 levels.(Facilitates data visualization and data exploration). | Facebook_Users$candidate_page <- factor(Facebook_Users$candidate_page, levels = c ("Harper", "Trudeau", "Mulcair")) | Visualization | https://osf.io/3fnjq/ | facebook_users.R |
449 | Adds a new column to identify the Facebook page being `liked`. | Harper_User_Likes$partisanship <- "Conservative" Trudeau_User_Likes$partisanship <- "Liberal" Mulcair_User_Likes$partisanship <- "Social Democrat" | Data Variable | https://osf.io/3fnjq/ | facebook_users.R |
450 | Match partisan assignment from Facebook_User_Likes to the Facebook Users Dataset using `user_id`. | Facebook_Users$partisanship <- NA Facebook_Users$partisanship <- Facebook_User_Likes$partisanship[match(Facebook_Users$user_id, Facebook_User_Likes$user_id)] | Data Variable | https://osf.io/3fnjq/ | facebook_users.R |
451 | Convert `partisanship` from a character string to a factor with 3 levels.(Facilitates data visualization and data exploration). | Facebook_Users$partisanship <- factor(Facebook_Users$partisanship, levels = c ("Conservative", "Liberal", "Social Democrat")) | Data Variable | https://osf.io/3fnjq/ | facebook_users.R |
452 | split the data and transform it into integers | ARTE <- as.numeric(data[which(data$TV.Station == 'ARTE'),]$Rating) Degeto <- as.numeric(data[which(data$TV.Station == 'Degeto Film'),]$Rating) | Data Variable | https://osf.io/8fsbd/ | IMDB_analysis.r |
453 | run the onesided ttest with alpha 0.1 | t.test(ARTE, Degeto, alternative = "greater", conf.level = 0.90) | Statistical Test | https://osf.io/8fsbd/ | IMDB_analysis.r |
454 | Create list of dataframes per country | countries <- unique(data8$cntry_full) index = 0 listofdfs <- list() | Data Variable | https://osf.io/k853j/ | ESS_openness_2016_perCountry.R |
455 | Draw correlation plots | char.col<-c("#228822","#663388","#006688") groupn<-c(4,11,6,3) left<--8 top<-25 corrplot.mixed(cortabNB, upper = "ellipse",tl.pos="lt",tl.col=rep(char.col,groupn[1:3]),number.cex = .7) text(left,top,"Non-biological\nfathers",cex=1.2,pos=4) nc<-0.5 w<-16 tiff("Figure_S3_corr1.tif",width=w,height=2*w,units="cm",res=600,compression="lzw") par(mfrow=c(2,1)) corrplot.mixed(cortabNB, upper = "ellipse",tl.pos="lt",number.cex = nc,tl.col=rep(char.col,groupn[1:3])) text(left,top,"Non-biological\nfathers",cex=1.2,pos=4) corrplot.mixed(cortabPA, upper = "ellipse",tl.pos="lt",number.cex = nc,tl.col=rep(char.col,groupn[1:3])) text(left,top,"Partners",cex=1.2,pos=4) dev.off() tiff("Figure_S4_corr2.tif",width=w,height=2*w,units="cm",res=600,compression="lzw") par(mfrow=c(2,1)) corrplot.mixed(cortabLO, upper = "ellipse",tl.pos="lt",number.cex = nc,tl.col=rep(char.col,groupn[1:3])) text(left,top,"Sensitive\nperiod\nlog-odds",cex=1.2,pos=4) corrplot.mixed(cortabD, upper = "ellipse",tl.pos="lt",number.cex = nc,tl.col=rep(char.col,groupn[1:3])) text(left,top,"Relative\nsimilarity\nt value",cex=1.2,pos=4) dev.off() | Visualization | https://osf.io/greqt/ | 07_correlation_structure.R |
456 | function to calculate proporiton of studies within each time period for continuous variables that were transformed in a cutoff x character string of the variable name character string of the name that will appear in the table time character string of the time period variable method either "chisq" or "fisher" depending on the type of the test wanted | CAT<-function(x, name, time, method) { tabtot <- table(DATA[,x]) sumtot <- sum(tabtot) tabtime <- table(DATA[,x], DATA[,time]) sumtime <- apply(tabtime, 2, sum) proptot <- tabtot[2]/sum(tabtot) CItot <- exactci(tabtot[2], sum(tabtot), conf.level = 0.95) proptime <- tabtime[2,]/sumtime CItime <- map2(tabtime[2,], sumtime, function(x, y){exactci(x, y, conf.level = 0.95)}) propCItime <- map2(proptime, CItime, function(x,y){paste0(round(x*100,0)," (", round(y[["conf.int"]][1]*100,0),"-", round(y[["conf.int"]][2]*100,0),")")}) if(method == "fisher"){ p <- fisher.test(tabtime) } else if(method == "chisq"){ p<-chisq.test(tabtime) } N_propCItime <- NULL for(i in 1:length(propCItime)){ N_propCItime <- c(N_propCItime, c(sumtime[i], propCItime[[i]])) } res=c(name, sumtot, paste0(round(proptot*100,0)," (",round(CItot$conf.int[1]*100,0),"-", round(CItot$conf.int[2]*100,0),")"), N_propCItime, round(p$p.value[1],4)) return(res) } | Statistical Test | https://osf.io/cxv5k/ | R_functions_Kimmoun_et_al_final.R |
457 | function to calculate overall effect size for death or readmission rates (output has results + model to do model checks) var character string of the variable to calculate the effect size for | rate_ES <- function(var){ dat <- na.omit(DATA[,c("number_follow_up", var)]) colnames(dat)<-c("Ni","Ei") dat_es <- escalc( xi= Ei, ni =Ni, data = dat, measure = "PLO", to="if0all") | Statistical Modeling | https://osf.io/cxv5k/ | R_functions_Kimmoun_et_al_final.R |
458 | function to get weighted logistic regression with continuous X variable and proportion Y variable and ggplot showing the predicted relationship x X variable as character y Y variable as character Xpred vector of X values to get predictions for labX X label as character labY Y label as character limY axis limits for Y axis, vector of 2 values title plot title as character ypos y relative position of text with OR and pvalue | wtlogis_plot <- function(NB, x, y, Xpred, labX, labY, limY, title, ypos, unit){ dat <- data.frame(count=as.integer(DATA[,y]), Ntot=DATA[,NB], t=DATA[,x]) %>% mutate(prop = count/Ntot, perc = prop*100, wt = log(Ntot), t_div = t/10) dat <- na.omit(dat) | Visualization | https://osf.io/cxv5k/ | R_functions_Kimmoun_et_al_final.R |
459 | model with time divided by 10 to get OR for 10yr increment | timediv_mod <- glm(prop ~ t_div, family = "binomial", weights = wt, data = dat) ORdiv <- tidy(timediv_mod, conf.int = TRUE) %>% mutate(OR_CI = paste0(round(exp(estimate),2), " (", round(exp(conf.low),2),"-", round(exp(conf.high),2), ")")) | Statistical Modeling | https://osf.io/cxv5k/ | R_functions_Kimmoun_et_al_final.R |
460 | function to get weighted linear regression with continuous X variable and continuous Y variable and ggplot showing the predicted relationship x X variable as character y Y variable as character Xpred vector of X values to get predictions for labX X label as character labY Y label as character limY axis limits for Y axis, vector of 2 values title plot title as character ypos y relative position of text with OR and pvalue | wtlin_plot <- function(NB, x, y, Xpred, labX, labY, limY, title, ypos, unit){ dat <- data.frame(y=as.integer(DATA[,y]), Ntot=DATA[,NB], t=DATA[,x]) %>% mutate(wt = log(Ntot), t_div = t/10) dat <- na.omit(dat) | Statistical Modeling | https://osf.io/cxv5k/ | R_functions_Kimmoun_et_al_final.R |
461 | function to get OR and CI from the model with 10 year increment obj model with x variable divided by 10 p pvalue from the LRT | OR_div <- function(obj, p){ OR_CI <- paste0(round(exp(obj$beta[2]),2), " (", round(exp(obj$ci.lb[2]),2),"-", round(exp(obj$ci.ub[2]),2), ")") pval_clean <- ifelse(p[["pval"]] < 0.001, "p<0.001", paste0("p=",round(p[["pval"]],3))) OR_p <- c(OR_CI, p, pval_clean) return(OR_p) } | Statistical Modeling | https://osf.io/cxv5k/ | R_functions_Kimmoun_et_al_final.R |
462 | function to get metaregression model with continuous X variable and proportion Y variable and ggplot showing the predicted relationship x X variable as character y Y variable as character Xpred vector of X values to get predictions for labX X label as character labY Y label as character limY axis limits for Y axis, vector of 2 values title plot title as character ypos y relative position of text with OR and pvalue | MA_prop_plot <- function(NB, x, y, Xpred, labX, labY, limY, title, ypos){ prop_dat <- na.omit(DATA[,c(NB, y, x)]) colnames(prop_dat)<-c("Ni","Ei","X") prop_dat$prop <- (prop_dat$Ei/prop_dat$Ni) * 100 prop_dat$X_div <- prop_dat$X/10 es_prop_dat <- escalc( xi= Ei, ni =Ni, data = prop_dat, measure = "PLO", to="if0all") | Visualization | https://osf.io/cxv5k/ | R_functions_Kimmoun_et_al_final.R |
463 | function to get OR for subgroup variables with 2 categories x character string for the x variable y character string for the y variable group chracter string for the group variable newref character string for the level of the group variable that is not the reference level | OR_group_fun <- function(x, y, group, newref){ OR_data <- data.frame(Group = rep(NA,2), N = rep(NA,2), OR = rep(NA,2), lowCI = rep(NA,2), upCI = rep(NA,2), p = rep(NA,2), I2 = rep(NA,2)) group_dat <- na.omit(DATA[,c("number_follow_up", y, x, group)]) colnames(group_dat)<-c("Ni","Ei","X","group") group_dat$X_div <- group_dat$X/10 OR_data$Group <- levels(group_dat$group) OR_data$N <- as.integer(table(group_dat$group)) es_group_dat <- escalc( xi= Ei, ni =Ni, data = group_dat, measure = "PLO", to="if0all") inter_mod1 <- rma(yi=yi,mods = ~ X_div * group, vi=vi, data=es_group_dat, method = "REML") inter_mod2 <- rma(yi=yi,mods = ~ X_div * I(relevel(group, newref)) , vi=vi, data=es_group_dat, method = "REML") | Data Variable | https://osf.io/cxv5k/ | R_functions_Kimmoun_et_al_final.R |
464 | function to get metaregression model with continuous X1 and X2 variables and 2 group variables and proportion Y variable and ggplot showing the predicted relationship with x1 for a median value of x2 x1 X1 variable as character x2 X2 variable as character group1 group1 variable (2level factor) as character group2 group2 variable (2level factor) as character y Y variable as character x1pred vector of values for which to predict x1 x2val median value of X2 for which to get predictions labX X label as character labY Y label as character limY axis limits for Y axis, vector of 2 values title plot title as character ypos y relative for text of OR and pvalue | MA_prop_plot_adj <- function(NB, x1, x2, group1, group2, y, x1pred, x2val, labX, labY, limY, title, ypos){ prop_dat <- na.omit(DATA[,c(NB, y, x1, x2, group1, group2)]) colnames(prop_dat)<-c("Ni","Ei","X1", "X2", "group1", "group2") prop_dat$prop <- (prop_dat$Ei/prop_dat$Ni) * 100 prop_dat$X1_div <- prop_dat$X1/10 es_prop_dat <- escalc( xi= Ei, ni =Ni, data = prop_dat, measure = "PLO", to="if0all") | Statistical Modeling | https://osf.io/cxv5k/ | R_functions_Kimmoun_et_al_final.R |
465 | rayleigh_test Rayleigh test for the uniformity of directional data, as described by Mardia et al. (1979: chapter 15) and Mardia & Jupp (1999) Note that there would probably be better alternatives on CRAN | rayleigh_test <- function(X, correction = TRUE, check = TRUE, convert = TRUE) { if(check) { if(!isTRUE(all.equal(diag(tcrossprod(X)), 1))) { if(convert) { X / sqrt(diag(tcrossprod(X))) warning("X was converted into directional data") } else { stop("Directional data (in coordinates) are assumed as X") } } } p <- ncol(X) n <- nrow(X) Mean <- colMeans(X) Norm_Mean_2 <- drop(crossprod(Mean)) S <- p * n * Norm_Mean_2 if(correction) { S <- S * (1 - 1 / 2 / n) + S ^ 2 / (2 * n * (p + 2)) } P_value <- pchisq(S, p, lower.tail = FALSE) list(Norm_Mean = sqrt(Norm_Mean_2), n = n, p = p, Statistics = S, P_value = P_value) } | Statistical Test | https://osf.io/6ukwg/ | utility_functions.R |
466 | Compute ICC run null/unconditional model and use intercept and residual variances to solve for ICC (i.e., Intercept Variance / (Intercept Variance + Residual Variance)) | Model.Null <- lmer(VerbPhysAggSum ~ 1 + (1 | ID), data = LineBisectionStudy) summary(Model.Null) ICC <- (18.69/(18.69+7.76)) print(ICC) | Statistical Modeling | https://osf.io/vprwb/ | MLM--lmerTest--JPNpsyUpdated.R |
467 | Function to extract legend (from: https://gist.github.com/crsh/be88be19233f1df4542aca900501f0fbfilegglegendrL7) otherwise plots without tracks would have no legend | gglegend <- function(x){ tmp <- ggplot_gtable(ggplot_build(x)) leg <- which(sapply(tmp$grobs, function(y) y$name) == "guide-box") tmp$grobs[[leg]] } legend = gglegend(gi) | Visualization | https://osf.io/amd3r/ | D_Wind_support_and_track_animation.R |
468 | define function doing the bootstrap: | boot.fun<-function(x, xcall., data., rv.name., m., discard.non.conv., save.path., extract.all.){ xdone=F while(!xdone){ done2=F while(!done2){ data.[, rv.name.]=simulate(object=m.)[, 1] if(xcall[["xfam"]][["family"]]!="beta" | (xcall[["xfam"]][["family"]]=="beta" & min(data.[, rv.name.])>0 & max(data.[, rv.name.])<1)){ done2=T } } i.res=try(update(m., data=data.), silent=T) | Data Variable | https://osf.io/vjeb3/ | boot_glmm.r |
469 | prepare model frame for conditional model for prediction: | model=attr(terms(as.formula(xcall[["cond.form"]])), "term.labels") model=model[!grepl(x=model, pattern="|", fixed=T)] if(length(model)==0){model="1"} cond.m.mat=model.matrix(object=as.formula(paste(c("~", paste(model, collapse="+")), collapse="")), data=new.data) if(set.circ.var.to.zero){ cond.m.mat[,paste(c("sin(", circ.var.name, ")"), collapse="")]=0 cond.m.mat[,paste(c("cos(", circ.var.name, ")"), collapse="")]=0 } | Statistical Modeling | https://osf.io/vjeb3/ | boot_glmm.r |
470 | Function that draws any bellcurve in the form and shape as the ones used in the paper | docurve<-function( sd=2, dist=1, xcent=10, ycent=0.6, paroff=0.15, offoff=0.1, percoff=0.06, percoff2=0.03, thresh.vis=0.01, bellims=15, botlims=8, pardown=0.01, xoff=0, par.col="#0097BD", off.col="#FF0066", off.in.col="#FF0066", off.out.col="#FF6600", fading="33", col.bot="#555555", lwd.par=1.2, lwd.bot=1.2){ off.in.col2<-paste(off.in.col,fading,sep="") off.out.col2<-paste(off.out.col,fading,sep="") p1<-xcent-dist/2 p2<-xcent+dist/2 c1<-gen.curve(sd,average=0,bellims=bellims) polygon(replast(c1[[1]][c1[[1]]<(-dist/2)])+xcent,c(c1[[2]][c1[[1]]<(-dist/2)],0)+ycent,col=off.out.col2,border=NA) polygon(repfirst(c1[[1]][c1[[1]]>(dist/2)])+xcent,c(0,c1[[2]][c1[[1]]>(dist/2)])+ycent,col=off.out.col2,border=NA) polygon(repboth(c1[[1]][c1[[1]]>=(-dist/2)&c1[[1]]<=(dist/2)])+xcent,c(0,c1[[2]][c1[[1]]>=(-dist/2)&c1[[1]]<=(dist/2)],0)+ycent,col=off.in.col2,border=NA) lines(c(xcent-botlims,xcent+botlims),c(ycent,ycent),col=col.bot,lwd=lwd.bot) lines(c(p1,p1),c(ycent-pardown,ycent+paroff),col=par.col,lwd=lwd.par) lines(c(p2,p2),c(ycent-pardown,ycent+paroff),col=par.col,lwd=lwd.par) outprob<-sum(c1[[2]][c1[[1]]>dist/2])/sum(c1[[2]]) inprob<-sum(c1[[2]][c1[[1]]<=dist/2&c1[[1]]>=-dist/2])/sum(c1[[2]]) offtail<-mean(c(c1[[1]][c1[[1]]>dist/2][1], c1[[1]][c1[[1]]>dist/2][!(c1[[2]][c1[[1]]>dist/2]>thresh.vis)][1])) text(xcent,ycent+percoff,paste(round(inprob*100),"%",sep=""),col=off.in.col,cex=0.8) text(c(xcent-offtail,xcent+offtail)+c(-1,1)*xoff,ycent+percoff2,paste(round(outprob*100),"%",sep=""),col=off.out.col,cex=0.8) text(c(p1,p2),ycent+paroff,c(expression(t[p[1]]),expression(t[p[2]])),pos=3,col=par.col,offset=0.15) } | Visualization | https://osf.io/pvyhe/ | Figure1.R |
471 | loglikelihood for each subject using their mean parameter vector | mean_pars_ll <- numeric(ncol(mean_pars)) data <- transform(sampled$data, subject = match(subject, unique(subject))) for (j in 1:nsubj) { mean_pars_ll[j] <- sampled$ll_func(mean_pars[j, ], data = data[data$subject == j,], sample = FALSE) } | Statistical Modeling | https://osf.io/tbczv/ | pmwgDIC.r |
472 | Create the plots and populate them. The yaxt "n" serves to remove axis labels from the yaxis while the axis function serves to add axis labels in line with APS standards for graph publication. The lines and points functions add the connecting lines and symbols to the graphs. We repeat the process three times to create the three graphs corresponding to different values of tau. | plot(k, end.data.cov[ ,1], type = "l", ylab = "Coverage", main = "A: Tau = 0.2", yaxt = "n", ylim = c(0.9, 1),xlab = expression("Number of Studies " (italic(k)))) axis(side = 2, at = c(0.92,0.94,0.96,0.98,1),labels = c(".92",".94",".96",".98","1.0")) points(k, end.data.cov[ ,1], pch = 0) lines(k, end.data.cov[ ,2]) points(k, end.data.cov[ ,2], pch = 3) lines(k, end.data.cov[ ,7]) points(k, end.data.cov[ ,7], pch = 5) lines(k, end.data.cov[ ,14]) points(k, end.data.cov[ ,14], pch = 4) legend("bottomright", c("bWT-DL","bWT-REML", "bWT-HE", "Sub-Q"), pch = c(0,5,4,3), cex = 0.9, ncol = 2) plot(k, end.data.cov[ ,3], type = "l", ylab = "Coverage", main = "B: Tau = 0.4", yaxt = "n", ylim = c(0.9, 1),xlab = expression("Number of Studies " (italic(k)))) axis(side = 2, at = c(0.92,0.94,0.96,0.98,1),labels = c(".92",".94",".96",".98","1.0")) points(k, end.data.cov[ ,3], pch = 0) lines(k, end.data.cov[ ,4]) points(k, end.data.cov[ ,4], pch = 3) lines(k, end.data.cov[ ,9]) points(k, end.data.cov[ ,9], pch = 5) lines(k, end.data.cov[ ,17]) points(k, end.data.cov[ ,17], pch = 4) legend("bottomright", c("bWT-DL","bWT-REML", "bWT-HE", "Sub-Q"), pch = c(0,5,4,3), cex = 0.9, ncol = 2) plot(k, end.data.cov[ ,5], type = "l", ylab = "Coverage", main = "C: Tau = 0.6", yaxt = "n", ylim = c(0.9, 1),xlab = expression("Number of Studies " (italic(k)))) axis(side = 2, at = c(0.92,0.94,0.96,0.98,1),labels = c(".92",".94",".96",".98","1.0")) points(k, end.data.cov[ ,5], pch = 0) lines(k, end.data.cov[ ,6]) points(k, end.data.cov[ ,6], pch = 3) lines(k, end.data.cov[ ,11]) points(k, end.data.cov[ ,11], pch = 5) lines(k, end.data.cov[ ,20]) points(k, end.data.cov[ ,20], pch = 4) legend("topright", c("bWT-DL","bWT-REML", "bWT-HE", "Sub-Q"), pch = c(0,5,4,3), cex = 0.9, ncol = 2) | Visualization | https://osf.io/gwn4y/ | Line_plots.R |
473 | Define the initial column name for x/y coordinates you want to use | xy_columns = list( x = "GazePointX (ADCSpx)", y = "GazePointY (ADCSpx)" ), | Data Variable | https://osf.io/mp9td/ | interface.R |
474 | compute cutoff for guessing by taking the 99% quantile from the binomial distribution (given by guessing probability and number of tests) | mutate(guessing_probability = 1/19, cut_off = qbinom(p = .99, size = number_tests, prob = guessing_probability)/number_tests, guessing_check = ifelse(mean_acc > cut_off, TRUE, FALSE)) %>% group_by(participant_id, condition) %>% summarize(guessing_check = as.logical(min(guessing_check))) %>% ungroup() | Statistical Modeling | https://osf.io/dpkyb/ | data-processing.R |
475 | split the data into different data frames depending on the intended analysis create dataframe with data on the hebb effect for the main analysis of the learning data | data_hebb_task = data_filtered %>% filter(phase == "WM") %>% select(participant_id, condition, phase, block, trial_number, hebb_trial, presentation_order:duration) %>% ungroup() | Data Variable | https://osf.io/dpkyb/ | data-processing.R |
476 | check cluster agreement using hierarchical clustering on raw items instead of factors | HCheck.HClusterD <- dist(EFA.mainData[, c(2:56)], method = "euclidean") HCheck.HClusterFit <- hclust(HCheck.HClusterD, method="ward.D2") HCheck.HCluster <- cutree(HCheck.HClusterFit, k = 3) HCheckCluster <- data.frame(Factor = HCluster, Item = HCheck.HCluster) HCheckCluster <- cbind(HCheckCluster, ifelse(HCheckCluster[1] == HCheckCluster[2], 1, 0)) colnames(HCheckCluster)[3] <- 'Agreement' HCheckClusterAgreement <- HCheckCluster[, c(2:3)] %>% group_by(Item) %>% summarise(total = n(), n_agree = sum(Agreement)) %>% rename(Cluster = Item) | Statistical Modeling | https://osf.io/2j47e/ | Cluster analysis refined factor scores.R |
477 | create dataframe with factor scores and cluster membership | KClusterScores <- cbind(EFA.mainData[, c(111:116)], KClusterFit$cluster) colnames(KClusterScores)[7] <- "KCluster" | Data Variable | https://osf.io/2j47e/ | Cluster analysis refined factor scores.R |
478 | testing homogeneity of variance in factors | KClusterScores %>% pivot_longer(c(Sensory, CognitiveDemand, ThreatToSelf, CrossSettings, Safety, States), names_to = c("Factor")) %>% group_by(Factor) %>% levene_test(value ~ as.factor(KCluster)) | Statistical Test | https://osf.io/2j47e/ | Cluster analysis refined factor scores.R |
479 | univariate Welch's ANOVA test with Bonferroni correction | table1 <- KClusterScores %>% pivot_longer(c(Sensory, CognitiveDemand, ThreatToSelf, CrossSettings, Safety, States), names_to = c("Factor")) %>% mutate(Factor = factor(Factor, levels = c("Sensory", "CognitiveDemand", "ThreatToSelf", "CrossSettings", "Safety", "States"))) %>% group_by(Factor) %>% welch_anova_test(value ~ as.factor(KCluster)) %>% adjust_pvalue(method = "bonferroni") | Statistical Test | https://osf.io/2j47e/ | Cluster analysis refined factor scores.R |
480 | manually calculating omega squared and adjusted confidence intervals for each test | table1$omegasq <- apply(table1, 1, function(x) omega.F(as.numeric(x[5]), as.numeric(x[6]), as.numeric(x[4]), as.numeric(x[3]), 0.05/6)$omega) table1$omegalow <- apply(table1, 1, function(x) omega.F(as.numeric(x[5]), as.numeric(x[6]), as.numeric(x[4]), as.numeric(x[3]), 0.05/6)$omegalow) table1$omegahigh <- apply(table1, 1, function(x) omega.F(as.numeric(x[5]), as.numeric(x[6]), as.numeric(x[4]), as.numeric(x[3]), 0.05/6)$omegahigh) | Statistical Test | https://osf.io/2j47e/ | Cluster analysis refined factor scores.R |
481 | Linear mixed effect model (accuracy) | LDTnonword_ACC_LME = glmer(accuracy ~ DISHARc + (1+DISHARc|item) + (1|subject), data = byTrial, family=binomial, control=glmerControl(optimizer="bobyqa", optCtrl=list(maxfun=2e5))) summary(LDTnonword_ACC_LME) | Statistical Modeling | https://osf.io/gztxa/ | Vowel_Harmony_LDT_Exp3.R |
482 | Estimate d from t and df | dest <-function (t,df) { r2 <- (t^2) / (t^2 + df) d.est <- (2*(r2^.5)) / (1-r2) } | Data Variable | https://osf.io/he8mu/ | Custom_Functions.R |
483 | Stimuli categorized according to a multinomial distribution | y[i,] ~ dmulti(r[i,], t[i]) predy[i,1:ncat] ~ dmulti(r[i,], t[i]) } | Statistical Modeling | https://osf.io/hrf5t/ | Prototype.R |
484 | Denominator is just the sum of the numerator values | denominator[i] = sum(numerator[i,]) } | Data Variable | https://osf.io/hrf5t/ | Prototype.R |
485 | annotate lines so that each has a unique id for ggplot overplotting (else two lines from the same draw but different replicates can get confused with each other) | func_samples_surface <- func_samples_surface %>% mutate(line_id = as.numeric(rownames(func_samples_surface))) func_samples_aer <- func_samples_aer %>% mutate(line_id = as.numeric(rownames(func_samples_aer))) | Visualization | https://osf.io/fb5tw/ | figure_main.R |
486 | cross product decay_rates with x (time) values and calculate y (titer) values | cat('setting up x values...\n') to_plot_surface <- func_samples_surface %>% crossing(surface_plot_times) to_plot_aer <- func_samples_aer %>% crossing(aer_plot_times) to_plot_surface <- to_plot_surface %>% mutate(predicted_titer = 10^(intercept - decay_rate * time)) to_plot_aer <- to_plot_aer %>% mutate(predicted_titer = convert_mL_media_to_L_air * 10^(intercept - decay_rate * time)) max_nonzero_time <- to_plot_surface %>% filter(log10(predicted_titer) > lowest_log_titer) %>% select(time) %>% max() surface_xlim <- c(0, max_nonzero_time) aer_xlim <- c(0, aer_max_x) print(aer_xlim) aer_jitwid <- 3/100 fit_panel_surface <- to_plot_surface %>% ggplot(aes( x = time, y = predicted_titer, color = virus, group = line_id)) + geom_line(alpha = line_alpha, size = line_size) + scale_colour_manual(values = unlist(virus_colors)) + geom_point(aes(x = time, y = 10^(log10_titer), group = trial_unique_id), data = surface_dat, color = pointborder, fill = pointfill, alpha = pointalpha, size = pointsize, stroke = pointstroke, position = position_jitter( width = jitwid, height = jith, seed = 5)) + geom_hline( data = experiment_dat_virus_surface, aes(yintercept = detection_limit), linetype = detection_linestyle, size = detection_linesize) + scale_y_continuous(trans = ytrans, breaks = ybreaks, labels = yformat) + coord_cartesian(ylim = surface_ylim, xlim = surface_xlim) + facet_grid(vars(virus), vars(material), drop = TRUE) | Data Variable | https://osf.io/fb5tw/ | figure_main.R |
487 | group adjustment for sigma with prior 0,0.1 for the beta | ACT.BF.sigma.prior01 <- stan(data=ACT.data,file="./BayesFactor/models/ACT/ACT.BF.sigma.prior01.stan", iter=40000, warmup =1000, chains=3,control = list(adapt_delta = 0.9)) ACT.BF.sigma.prior01.bridge <- bridge_sampler(ACT.BF.sigma.prior01) saveRDS(ACT.BF.sigma.prior01.bridge, "./BayesFactor/marginal_lik/ACT/ACT.BF.sigma.prior01.rds") DA.BF.sigma.prior01 <- stan(data=DA.standata,file="./BayesFactor/models/DA/DA.BF.sigma01.stan", iter=40000, warmup =1000, chains=3) DA.BF.sigma.prior01.bridge <- bridge_sampler(DA.BF.sigma.prior01) saveRDS(DA.BF.sigma.prior01.bridge, "./BayesFactor/marginal_lik/DA/DA.BF.sigma.prior01.rds") | Statistical Modeling | https://osf.io/kdjqz/ | BF_LissonEtAl2020.R |
488 | increase in rsq when using elastic net with items instead of linear regression with sum scores (without covariates) | tab_delta_rsq <- merge(tab[learner.id == "elastic net" & predictors == "items" & control == "excluded", .(elasticnet_items = rsq.test.mean), by = c("target", "design")], tab[learner.id == "linear regr" & predictors == "sum scores" & control == "excluded", .(linearregr_sumscores = rsq.test.mean), by = c("target", "design")]) tab_delta_rsq[, .(mean_delta_rsq = mean(elasticnet_items - linearregr_sumscores)), by = "design"] | Statistical Modeling | https://osf.io/t7a28/ | collect_results.R |
489 | Calculate Cohen's d_z | return_Cohen_d_z <- function(variable_1, variable_2){ mean_differences <- mean(variable_1) - mean(variable_2) sd_var1 <- sd(variable_1) sd_var2 <- sd(variable_2) cor_var1_var2 <- cor(variable_1, variable_2) Cohen_d_z <- abs(mean_differences) / sqrt( (sd_var1 ^ 2) + (sd_var2 ^ 2) - (2 * sd_var1 * sd_var2 * cor_var1_var2) ) return(Cohen_d_z) } | Statistical Test | https://osf.io/5te7n/ | return_effect_sizes.R |
490 | visualizing the data add +scale_x_discrete(guide guide_axis(angle 45)) to the plot to change the angle of the lables on the X axis | ggplot(mydata_p6_la, aes(x = Condition, y = Voltage)) + geom_boxplot() + facet_grid(.~ Specificity,)+ scale_x_discrete(guide = guide_axis(angle = 45)) ggplot(mydata_p6_ra, aes(x = Condition, y = Voltage)) + geom_boxplot() + facet_grid(.~ Specificity,)+ scale_x_discrete(guide = guide_axis(angle = 45)) ggplot(mydata_p6_lp, aes(x = Condition, y = Voltage)) + geom_boxplot() + facet_grid(.~ Specificity,)+ scale_x_discrete(guide = guide_axis(angle = 45)) ggplot(mydata_p6_rp, aes(x = Condition, y = Voltage)) + geom_boxplot() + facet_grid(.~ Specificity,)+ scale_x_discrete(guide = guide_axis(angle = 45)) | Visualization | https://osf.io/p7zwr/ | P600.R |
491 | fill output matrix with the calculated social measure | socialmeasures[,3] <- measure1 socialmeasures[,4] <- measure2 socialmeasures[,5] <- measure3 socialmeasures[,6] <- measure4 socialmeasures[,7] <- measure5 socialmeasures[,8] <- measure6 socialmeasures[,10] <- measure8 socialmeasures[,11] <- measure9 socialmeasures[,12] <- measure10 socialmeasures[,3] <- measure1 socialmeasures[,4] <- measure2 socialmeasures[,5] <- measure3 socialmeasures[,6] <- measure4 socialmeasures[,7] <- measure5 socialmeasures[,8] <- measure6 socialmeasures[,10] <- measure8 socialmeasures[,11] <- measure9 socialmeasures[,12] <- measure10 | Data Variable | https://osf.io/wc3nq/ | 2) soc_measures_code.R |
492 | 7 calculate the AVERAGE SHORTEST PATH of each node of the matrix | measure7 <- mean_distance(graphN, directed = TRUE, unconnected = TRUE) print(measure7) measure7 <- mean_distance(graphN, directed = TRUE, unconnected = TRUE) print(measure7) | Statistical Modeling | https://osf.io/wc3nq/ | 2) soc_measures_code.R |
493 | Add Age Bins This acts to help preserve privacy and allows easy plotting: e.g., ggplot(Data, aes(x Agebins) ) + geom_bar() | Agebin = c('< 20', '20-29', '30-39','40-49','50-59','60-69','70-79','80+') Data %<>% mutate(Agebins = case_when( Age < 20 ~ Agebin[1], Age >= 20 & Age < 30 ~ Agebin[2], Age >= 30 & Age < 40 ~ Agebin[3], Age >= 40 & Age < 50 ~ Agebin[4], Age >= 50 & Age < 60 ~ Agebin[5], Age >= 60 & Age < 70 ~ Agebin[6], Age >= 70 & Age < 80 ~ Agebin[7], Age >= 80 ~ Agebin[8] ) ) %>% apply_labels(Agebins = c('< 20' = 1, '20-29' = 2, '30-39' = 3, '40-49' = 4,'50-59' = 5,'60-69' = 6, '70-79' = 7,'80+' = 8)) Agebin = c('< 20', '20-29', '30-39','40-49','50-59','60-69','70-79','80+') Data %<>% mutate(Agebins = case_when( Age < 20 ~ Agebin[1], Age >= 20 & Age < 30 ~ Agebin[2], Age >= 30 & Age < 40 ~ Agebin[3], Age >= 40 & Age < 50 ~ Agebin[4], Age >= 50 & Age < 60 ~ Agebin[5], Age >= 60 & Age < 70 ~ Agebin[6], Age >= 70 & Age < 80 ~ Agebin[7], Age >= 80 ~ Agebin[8] ) ) %>% apply_labels(Agebins = c('< 20' = 1, '20-29' = 2, '30-39' = 3, '40-49' = 4,'50-59' = 5,'60-69' = 6, '70-79' = 7,'80+' = 8)) | Visualization | https://osf.io/sw7rq/ | Functions.R |
494 | ChiSquared Gender create dataframe of gender and cluster membership, then exclude participant with 'Other' as response to analyse with Chisquared test | data.OtherExcluded <- demographics %>% select(Gender, KCluster) %>% filter(Gender != 'Other') data.OtherExcluded$Gender <- droplevels(data.OtherExcluded$Gender) | Statistical Test | https://osf.io/2j47e/ | Demographics.R |
495 | posthoc pairwise Chisquared with Bonferroni correction | pairwiseNominalIndependence(ASD.contable, compare = "row", fisher = FALSE, gtest = FALSE, chisq = TRUE, method = "bonferroni", digits = 3) | Statistical Test | https://osf.io/2j47e/ | Demographics.R |
496 | compile parameter estimates and fit of population ODE/SDE models | modt <- readRDS("PSM_transits_ODE.RDS") mods <- readRDS("PSM_population_ODE.RDS") i <- length(mods) pars <- c(round(modt[[4]]$THETA,1), "OMEGA_stress"=1.5, "OMEGA_ke"=.05, "OMEGA_kt"=.15, "OMEGA_init"=.15) pars[c("init","sigma")] <- c(.001,.001) parA <- list(LB=pars*.2, Init=pars, UB=pars*2.5) #bounds + inits npars <- Vectorize(function(x) sum(parA$Init != round(mods[[x]]$THETA,3)))(1:i) res <- Vectorize(function(x) round(mods[[x]]$THETA,3))(1:i) #parameter estimates for each fitted model res <- rbind(res, "Mtt"= Vectorize(function(x) round(4/mods[[x]]$THETA["kt"],3))(1:i)) #add mean transit time for each model res <- rbind(res, "LL"= -Vectorize(function(x) mods[[x]]$NegLogL)(1:i)) res <- rbind(res, "AIC"= 2*npars +2*Vectorize(function(x) mods[[x]]$NegLogL)(1:i)) res <- rbind(res, "R2"= Vectorize(function(x) 1 - mods[[x]]$THETA["S"]/var(PKdata$DV))(1:i)) colnames(res) <- letters[1:i] print(res) | Statistical Modeling | https://osf.io/ecjy6/ | CortStressResponse.R |
497 | define a fixed grid of cutoff (threshold) values from [1, 0] with length `resolution` | resolution <- 500 cutoff_out <- seq(1, 0, length.out = resolution) resolution <- 500 cutoff_out <- seq(1, 0, length.out = resolution) | Data Variable | https://osf.io/w7pjy/ | bootstrapConfusionMatrix.R |
498 | Compute F1 skill score... ' ...for a given cutoffthreshold. ' @param sim vector of numeric values between [0, 1] (e.g., proportion of unstable grid points) ' @param obs vector of logicals (TRUE/FALSE) stating whether the layer was observed (of concern) or not. ' @param cutoff in percentage within (0, 1] ' @return numeric value of skill score ' @export | calculateF1Score <- function(sim, obs, cutoff) { tp <- length(sim[sim >= cutoff & obs]) fn <- length(sim[sim < cutoff & obs]) fp <- length(sim[sim >= cutoff & !obs]) f1 <- 2*tp / (2*tp + fp + fn) return(f1) } | Data Variable | https://osf.io/w7pjy/ | bootstrapConfusionMatrix.R |
499 | Underscore split: split up the "variable" column in order to get the different factors (verbtype, input, moment) | participantsdata.long <- cbind(participantsdata.long, colsplit(participantsdata.long$variable, "_", names = c("verbtype", "input", "testmoment"))) | Data Variable | https://osf.io/938ye/ | Descriptive_statistics.R |
500 | LDA training the model | model_lda = train(class ~ ., data=trainSet, method='lda', trControl = trainControl(method = "cv")) fitted <- predict(model_lda) | Statistical Modeling | https://osf.io/xuz8d/ | ThesisMLRCode.R |
Subsets and Splits