# Favorite Friend forced_choice_no_fam_friend_favorite = read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_friends_30_november.csv",header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) sum(forced_choice_no_fam_friend_favorite$Answer.choice == "\"target\"") forced_choice_no_fam_friend_favorite$target <- forced_choice_no_fam_friend_favorite$Answer.choice=="\"target\"" forced_choice_no_fam_friend_favorite$logical <- forced_choice_no_fam_friend_favorite$Answer.choice=="\"logical\"" forced_choice_no_fam_friend_favorite$foil <- forced_choice_no_fam_friend_favorite$Answer.choice=="\"foil\"" ms.favorite_friend <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=forced_choice_no_fam_friend_favorite, sum) # Favorite boat forced_choice_no_fam_boat_favorite = read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_boat_1_december_XYZQ_incomplete.csv",header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) sum(forced_choice_no_fam_boat_favorite$Answer.choice == "\"target\"") forced_choice_no_fam_boat_favorite$target <- forced_choice_no_fam_boat_favorite$Answer.choice=="\"target\"" forced_choice_no_fam_boat_favorite$logical <- forced_choice_no_fam_boat_favorite$Answer.choice=="\"logical\"" forced_choice_no_fam_boat_favorite$foil <- forced_choice_no_fam_boat_favorite$Answer.choice=="\"foil\"" ms.favorite_boat <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=forced_choice_no_fam_boat_favorite, sum) # Favorite pizza forced_choice_no_fam_pizza_favorite = read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_pizza_30_november.csv",header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) sum(forced_choice_no_fam_pizza_favorite$Answer.choice == "\"target\"") forced_choice_no_fam_pizza_favorite$target <- forced_choice_no_fam_pizza_favorite$Answer.choice=="\"target\"" forced_choice_no_fam_pizza_favorite$logical <- forced_choice_no_fam_pizza_favorite$Answer.choice=="\"logical\"" forced_choice_no_fam_pizza_favorite$foil <- forced_choice_no_fam_pizza_favorite$Answer.choice=="\"foil\"" ms.favorite_pizza <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=forced_choice_no_fam_pizza_favorite, sum) # Favorite snowman forced_choice_no_fam_snowman_favorite = read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_snowman_30_november.csv",header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) sum(forced_choice_no_fam_snowman_favorite$Answer.choice == "\"target\"") forced_choice_no_fam_snowman_favorite$target <- forced_choice_no_fam_snowman_favorite$Answer.choice=="\"target\"" forced_choice_no_fam_snowman_favorite$logical <- forced_choice_no_fam_snowman_favorite$Answer.choice=="\"logical\"" forced_choice_no_fam_snowman_favorite$foil <- forced_choice_no_fam_snowman_favorite$Answer.choice=="\"foil\"" ms.favorite_snowman <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=forced_choice_no_fam_snowman_favorite, sum) total_table_favorite = rbind(ms.favorite_boat, ms.favorite_friend, ms.favorite_pizza, ms.favorite_snowman) ms.favorite = sum(total_table_favorite$target) / sum(total_table_favorite[1:4,2:4]) favorite_separated_together = rbind(forced_choice_no_fam_friend_favorite, forced_choice_no_fam_snowman_favorite, forced_choice_no_fam_pizza_favorite, forced_choice_no_fam_boat_favorite) ################################################################################################################## # With "least favorite" least_favorite <- read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_4random_count_12_january_least_LEAS.csv", header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) count_compliant_least <- least_favorite$Answer.manip_check_target == "\"2\"" & least_favorite$Answer.manip_check_dist == "\"1\"" least_favorite_wrong <- subset(least_favorite,!count_compliant_least) least_favorite <- subset(least_favorite,count_compliant_least) least_favorite$target <- least_favorite$Answer.choice=="\"target\"" least_favorite$logical <- least_favorite$Answer.choice=="\"logical\"" least_favorite$foil <- least_favorite$Answer.choice=="\"foil\"" least_favorite_randomized <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=least_favorite, sum) ms.es_np_least_favorite <- mean(least_favorite$target) ################################################################################################################## # With new linguistic framing of one word -> The large and complete (but random kinds) one_word_lf <- read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_6random_count_ALLS.csv", header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) count_compliant <- one_word_lf$Answer.manip_check_target == "\"2\"" & one_word_lf$Answer.manip_check_dist == "\"1\"" one_word_lf_wrong <- subset(one_word_lf,!count_compliant) one_word_lf <- subset(one_word_lf,count_compliant) one_word_lf$target <- one_word_lf$Answer.choice=="\"target\"" one_word_lf$logical <- one_word_lf$Answer.choice=="\"logical\"" one_word_lf$foil <- one_word_lf$Answer.choice=="\"foil\"" total_one_word_randomized <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=one_word_lf, sum) ms.es_one_word <- mean(one_word_lf$target) summary(aov(target ~ as.factor(Answer.item) * as.factor(Answer.target_position),data = one_word_lf)) ################################################################################################################## # One word friend one_word_friend = read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_friend_oneword_1_december_AGSK.csv",header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) sum(one_word_friend$Answer.choice == "\"target\"") one_word_friend$target <- one_word_friend$Answer.choice=="\"target\"" one_word_friend$logical <- one_word_friend$Answer.choice=="\"logical\"" one_word_friend$foil <- one_word_friend$Answer.choice=="\"foil\"" ms.one_word_friend <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=one_word_friend, sum) # One word boat one_word_boat = read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_boat_oneword_7_december_LFBN_2.csv",header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) sum(one_word_boat$Answer.choice == "\"target\"") one_word_boat$target <- one_word_boat$Answer.choice=="\"target\"" one_word_boat$logical <- one_word_boat$Answer.choice=="\"logical\"" one_word_boat$foil <- one_word_boat$Answer.choice=="\"foil\"" ms.one_word_boat <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=one_word_boat, sum) # One word snowman one_word_snowman = read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_snowman_oneword_8_december_LSBN.csv",header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) sum(one_word_snowman$Answer.choice == "\"target\"") one_word_snowman$target <- one_word_snowman$Answer.choice=="\"target\"" one_word_snowman$logical <- one_word_snowman$Answer.choice=="\"logical\"" one_word_snowman$foil <- one_word_snowman$Answer.choice=="\"foil\"" ms.one_word_snowman <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=one_word_snowman, sum) # One word sundae one_word_sundae = read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_sundaes_oneword_8_december_LSBZ.csv",header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) sum(one_word_sundae$Answer.choice == "\"target\"") one_word_sundae$target <- one_word_sundae$Answer.choice=="\"target\"" one_word_sundae$logical <- one_word_sundae$Answer.choice=="\"logical\"" one_word_sundae$foil <- one_word_sundae$Answer.choice=="\"foil\"" ms.one_word_sundae <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=one_word_sundae, sum) total_table_one_word = rbind(ms.one_word_friend, ms.one_word_boat, ms.one_word_snowman, ms.one_word_sundae) ms.one_word_total = sum(total_table_one_word$target) / sum(total_table_one_word[1:4,2:4]) one_word_separated_together = rbind(one_word_sundae, one_word_snowman, one_word_boat, one_word_friend) ######################################################################################################## # Asking for what Bob thinks is the most beautiful most_beautiful <- read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_6random_count_beautiful_15_january_BEAU.csv", header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) count_compliant <- most_beautiful$Answer.manip_check_target == "\"2\"" & most_beautiful$Answer.manip_check_dist == "\"1\"" most_beautiful_wrong <- subset(most_beautiful,!count_compliant) most_beautiful <- subset(most_beautiful,count_compliant) most_beautiful$target <- most_beautiful$Answer.choice=="\"target\"" most_beautiful$logical <- most_beautiful$Answer.choice=="\"logical\"" most_beautiful$foil <- most_beautiful$Answer.choice=="\"foil\"" most_beautiful_randomized <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=most_beautiful, sum) ms.most_beautiful <- mean(most_beautiful$target) ################### This one is wrong ######### most_ugly_wrong <- read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_6random_count_ugly_15_january_UHGG.csv", header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) count_compliant <- most_ugly_wrong$Answer.manip_check_target == "\"2\"" & most_ugly_wrong$Answer.manip_check_dist == "\"1\"" most_ugly_wrong_www <- subset(most_ugly_wrong,!count_compliant) most_ugly_wrong <- subset(most_ugly_wrong,count_compliant) most_ugly_wrong$target <- most_ugly_wrong$Answer.choice=="\"target\"" most_ugly_wrong$logical <- most_ugly_wrong$Answer.choice=="\"logical\"" most_ugly_wrong$foil <- most_ugly_wrong$Answer.choice=="\"foil\"" most_ugly_wrong_randomized <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=most_ugly_wrong, sum) ms.most_ugly_wrong <- mean(most_ugly_wrong$target) ################### Ugly done right ############### most_ugly <- read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_6random_count_ugly_15_january_UGLY.csv", header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) count_compliant <- most_ugly$Answer.manip_check_target == "\"2\"" & most_ugly$Answer.manip_check_dist == "\"1\"" most_ugly_wrong_www <- subset(most_ugly,!count_compliant) most_ugly <- subset(most_ugly,count_compliant) most_ugly$target <- most_ugly$Answer.choice=="\"target\"" most_ugly$logical <- most_ugly$Answer.choice=="\"logical\"" most_ugly$foil <- most_ugly$Answer.choice=="\"foil\"" most_ugly_randomized <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=most_ugly, sum) ms.most_ugly <- mean(most_ugly$target) ################### Cheerful ############### most_cheerful <- read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_6random_count_cheerful_17_january_CHEE.csv", header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) count_compliant_cheerful <- most_cheerful$Answer.manip_check_target == "\"2\"" & most_cheerful$Answer.manip_check_dist == "\"1\"" most_cheerful_non_compliant <- subset(most_cheerful,!count_compliant_cheerful) most_cheerful <- subset(most_cheerful,count_compliant_cheerful) most_cheerful$target <- most_cheerful$Answer.choice=="\"target\"" most_cheerful$logical <- most_cheerful$Answer.choice=="\"logical\"" most_cheerful$foil <- most_cheerful$Answer.choice=="\"foil\"" most_cheerful_randomized <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=most_cheerful, sum) ms.most_cheerful <- mean(most_cheerful$target) ################### Depressing ############### most_depressing <- read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_6random_count_depressing_17_january_DEPR.csv", header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) count_compliant_depressing <- most_depressing$Answer.manip_check_target == "\"2\"" & most_depressing$Answer.manip_check_dist == "\"1\"" most_depressing_non_compliant <- subset(most_depressing,!count_compliant_depressing) most_depressing <- subset(most_depressing,count_compliant_depressing) most_depressing$target <- most_depressing$Answer.choice=="\"target\"" most_depressing$logical <- most_depressing$Answer.choice=="\"logical\"" most_depressing$foil <- most_depressing$Answer.choice=="\"foil\"" most_depressing_randomized <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=most_depressing, sum) ms.most_depressing <- mean(most_depressing$target) ################# Base Rate "favorite" Bob is silent ################# silent_favorite <- read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_6random_count_favorite_22_january_MUTE.csv", header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) count_compliant_silent_favorite <- silent_favorite$Answer.manip_check_target == "\"2\"" & silent_favorite$Answer.manip_check_dist == "\"1\"" silent_favorite_non_compliant <- subset(silent_favorite,!count_compliant_silent_favorite) silent_favorite <- subset(silent_favorite,count_compliant_silent_favorite) silent_favorite$target <- silent_favorite$Answer.choice=="\"target\"" silent_favorite$logical <- silent_favorite$Answer.choice=="\"logical\"" silent_favorite$foil <- silent_favorite$Answer.choice=="\"foil\"" silent_favorite_randomized <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=silent_favorite, sum) ms.silent_favorite <- mean(silent_favorite$target) ################# Base Rate "least favorite" Bob is silent ################# silent_least_favorite <- read.csv("/Users/andesgomez/Documents/Stanford/Autumn2013-Masters/PayedWork/andres_data/forced_choice_no_fam_6random_count_least_favorite_22_january_MUTL.csv", header=TRUE, sep="\t", row.names=NULL, stringsAsFactors = FALSE) count_compliant_silent_least_favorite <- silent_least_favorite$Answer.manip_check_target == "\"2\"" & silent_least_favorite$Answer.manip_check_dist == "\"1\"" silent_least_favorite_non_compliant <- subset(silent_least_favorite,!count_compliant_silent_least_favorite) silent_least_favorite <- subset(silent_least_favorite,count_compliant_silent_least_favorite) silent_least_favorite$target <- silent_least_favorite$Answer.choice=="\"target\"" silent_least_favorite$logical <- silent_least_favorite$Answer.choice=="\"logical\"" silent_least_favorite$foil <- silent_least_favorite$Answer.choice=="\"foil\"" silent_least_favorite_randomized <- aggregate(cbind(target, logical, foil) ~ Answer.item, data=silent_least_favorite, sum) ms.silent_least_favorite <- mean(silent_least_favorite$target) ################### ALL OF The RESUTS ######### # Favorite total_table_favorite ms.favorite # Least favorite least_favorite_randomized ms.es_np_least_favorite # One word - all at once total_one_word_randomized ms.es_one_word # One word - separated total_table_one_word ms.one_word_total # Beautiful most_beautiful_randomized ms.most_beautiful # Ugly Wrong most_ugly_wrong_randomized ms.most_ugly_wrong # Ugly - no mistake here most_ugly_randomized ms.most_ugly # Cheerful most_cheerful_randomized ms.most_cheerful # Depressing most_depressing_randomized ms.most_depressing # Bob's favorite without Bob saying anything silent_favorite_randomized ms.silent_favorite # Bob's least favorite without Bob saying anything silent_least_favorite_randomized ms.silent_least_favorite ####################################### SemFest writeup ####################################### # Brief comparisons of the different silent conditions. sum(total_table_favorite[1:4,3]) / sum(total_table_favorite[1:4,2:4]) sum(least_favorite_randomized[1:4,3]) / sum(least_favorite_randomized[1:4,2:4]) sum(silent_least_favorite_randomized[1:4,3]) / sum(silent_least_favorite_randomized[1:4,2:4]) sum(silent_favorite_randomized[1:4,4]) / sum(silent_favorite_randomized[1:4,2:4]) silent_least_favorite$response = silent_least_favorite$logical*2 + silent_least_favorite$target*1 silent_favorite$response = silent_favorite$logical*2 + silent_favorite$target*1 favorite_separated_together$response = favorite_separated_together$logical*2 + favorite_separated_together$target*1 least_favorite$response = least_favorite$logical*2 + least_favorite$target*1 summary(glm(response ~ as.factor(Answer.item) + as.factor(Answer.target_position), data = silent_least_favorite)) summary(glm(response ~ as.factor(Answer.item) + as.factor(Answer.target_position), data = silent_favorite)) summary(glm(response ~ as.factor(Answer.item) + as.factor(Answer.target_position), data = favorite_separated_together)) summary(glm(response ~ as.factor(Answer.item) + as.factor(Answer.target_position), data = least_favorite)) ############## Brief Michael's analysis #################### #To get the model predictions, you are trying to compute p(r | w, C), the probability of the referent given the word and the context, which by bayes rule is #p( w | r, C) * p(r | C) #divided by the sum of this across all referents. So for the level 1 implicature, for favorite for the target, this is something like #p("glasses" | target) * .205 / (p("glasses" | no-item) * .11 + p("glasses" | target) * .205 + p("glasses" | logical) * .685) #where p("glasses" | target) = (1/ |"glasses"|) / (1/ |"glasses"|) = 1 #and for logical it's (1/ |"glasses"|) / (1/ |"glasses"| + 1/ |"hat"|)) = (1/2) / ((1/2) + 1) = (1/2) / (3/2) = 1/3 #and for no-item it's p("glasses" | no-item) = 0 #So that gives us #p("glasses" | target) * .22 / (p("glasses" | no-item) * .13 + p("glasses" | target) * .22 + p("glasses" | logical) * .65) # .22 / (.22 + (1/3) * .65) .14 / (.14 + (1/3)*.21) #which is actually not bad as a prediction! (Compare to the actual result, .53, below). - I am writing fast and haven't checked this but it seems reasonable. # For least favorite: p( w | r, C) * p(r | C) = #P("glasses" | target ) * ######################### GLOBAL ANALYSIS OF VARIANCE AND PREDICTORS ################################ one_word_separated_together$Answer.age = 1:length(one_word_separated_together$target) * 0 one_word_separated_together$Answer.gender = 1:length(one_word_separated_together$target) * 0 favorite_separated_together$Answer.age = 1:length(favorite_separated_together$target) * 0 favorite_separated_together$Answer.gender = 1:length(favorite_separated_together$target) * 0 one_word_lf$Answer.age = 1:length(one_word_lf$target) * 0 one_word_lf$Answer.gender = 1:length(one_word_lf$target) * 0 global_pool = rbind(silent_favorite, most_depressing, most_cheerful, most_ugly, most_beautiful, least_favorite, one_word_separated_together, favorite_separated_together, one_word_lf) # Sanitize Gener and Age global_pool$males = global_pool$Answer.gender == "\"m\"" | global_pool$Answer.gender == "\"male\"" | global_pool$Answer.gender == "\"M\"" | global_pool$Answer.gender == "\"Male\"" | global_pool$Answer.gender == "\"MALE\"" global_pool$females = global_pool$Answer.gender == "\"f\"" | global_pool$Answer.gender == "\"female\"" | global_pool$Answer.gender == "\"F\"" | global_pool$Answer.gender == "\"Female\"" | global_pool$Answer.gender == "\"FEMALE\"" global_pool$twenties = global_pool$Answer.age == "\"20\"" | global_pool$Answer.age == "\"21\"" | global_pool$Answer.age == "\"22\"" | global_pool$Answer.age == "\"23\"" | global_pool$Answer.age == "\"24\"" | global_pool$Answer.age == "\"25\"" | global_pool$Answer.age == "\"26\"" | global_pool$Answer.age == "\"27\"" | global_pool$Answer.age == "\"28\"" | global_pool$Answer.age == "\"29\"" global_pool$thirties = global_pool$Answer.age == "\"30\"" | global_pool$Answer.age == "\"31\"" | global_pool$Answer.age == "\"32\"" | global_pool$Answer.age == "\"33\"" | global_pool$Answer.age == "\"34\"" | global_pool$Answer.age == "\"35\"" | global_pool$Answer.age == "\"36\"" | global_pool$Answer.age == "\"37\"" | global_pool$Answer.age == "\"38\"" | global_pool$Answer.age == "\"39\"" global_pool$fourties = global_pool$Answer.age == "\"40\"" | global_pool$Answer.age == "\"41\"" | global_pool$Answer.age == "\"42\"" | global_pool$Answer.age == "\"43\"" | global_pool$Answer.age == "\"44\"" | global_pool$Answer.age == "\"45\"" | global_pool$Answer.age == "\"46\"" | global_pool$Answer.age == "\"47\"" | global_pool$Answer.age == "\"48\"" | global_pool$Answer.age == "\"49\"" global_pool$fifties = global_pool$Answer.age == "\"50\"" | global_pool$Answer.age == "\"51\"" | global_pool$Answer.age == "\"52\"" | global_pool$Answer.age == "\"53\"" | global_pool$Answer.age == "\"54\"" | global_pool$Answer.age == "\"55\"" | global_pool$Answer.age == "\"56\"" | global_pool$Answer.age == "\"57\"" | global_pool$Answer.age == "\"58\"" | global_pool$Answer.age == "\"59\"" # ANOVA anova_effects = aov(target ~ as.factor(Answer.item) * as.factor(Answer.linguistic_framing_condition) + as.factor(Answer.target_position) + as.factor(Answer.name_check_correct) + males + females + twenties + thirties + fourties + fifties, data = global_pool) summary(anova_effects) anova_subtle = aov(target ~ as.factor(Answer.target_position) + as.factor(Answer.name_check_correct) + as.factor(Answer.target_property) + males + females, data = global_pool) summary(anova_subtle) # Regression regression_effects = glm(target ~ as.factor(Answer.item) + as.factor(Answer.linguistic_framing_condition) + as.factor(Answer.target_position) + males + females + twenties + thirties + fourties + fifties, data = global_pool) summary(regression_effects) subtle_regression = glm(target ~ Answer.age + Answer.gender, data = global_pool) summary(subtle_regression) regresion_lineal_simple = function (datos_como_matriz) { return ( lm( formula = as.formula(paste(colnames(datos_como_matriz), collapse='~') ), data = as.data.frame(datos_como_matriz))) }