---
title: "007_LookAnalysis_eR_10_1_21"
author: "Katharine Aveni"
date: "October 1, 2021"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

#Load Libraries & Data
```{r load libraries and data}

#load Data viewer tab-separated sample report; set . and NA as missing value marker
#replace argument after "read.table" with your txt file
dat007Look <- read.table("C:\\Users\\katie\\OneDrive\\Desktop\\10_17_19_LookAt_SentenceDur_Extended.txt", header = T, sep = "\t", na.strings = c(".", "NA"))

#load required packages
library("Matrix")
library("plyr")
library("dplyr")
library("tidyr")
library("lme4")
library("ggplot2")
library("eyetrackingR")
```

# Create and clean required columns
##Subject & IA column relabeling
```{r Subject & IA column relabeling}
#rename column "RECORDING_SESSION_LABEL" to "Subject"
names(dat007Look)[names(dat007Look) == "RECORDING_SESSION_LABEL"] <- "Subject"

#label IAs consistently
dat007Look$RIGHT_INTEREST_AREA_ID[dat007Look$RIGHT_INTEREST_AREA_LABEL %in% c("DIS_ACTIONRELATED_IA")] <- 3

#label trials consistently
dat007Look$trial_name[dat007Look$trial_name=="ExpB_Item_A" | dat007Look$audiofile=="EB_helmet.wav"] <- "ExpB_Item 1"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_B")] <- "ExpB_Item 2"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_C")] <- "ExpB_Item 3"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_D")] <- "ExpB_Item 4"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_E")] <- "ExpB_Item 5"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_F")] <- "ExpB_Item 6"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_G")] <- "ExpB_Item 7"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_H")] <- "ExpB_Item 8"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_I")] <- "ExpB_Item 9"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_J")] <- "ExpB_Item 10"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_K")] <- "ExpB_Item 11"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_L")] <- "ExpB_Item 12"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_M")] <- "ExpB_Item 13"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_N")] <- "ExpB_Item 14"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_O")] <- "ExpB_Item 15"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_P")] <- "ExpB_Item 16"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_Q")] <- "ExpB_Item 17"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_R")] <- "ExpB_Item 18"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_S")] <- "ExpB_Item 19"
dat007Look$trial_name[dat007Look$trial_name  %in% c("ExpB_Item_T")] <- "ExpB_Item 20"

length(unique(dat007Look$trial_name))
list(unique(dat007Look$trial_name))
```

##MARK SAMPLES WHERE TRACKLOSS OCCURS, creating new TrackLoss column
Trackloss defined as missing values for right gaze x & y and left gaze x & y. Used is.na function bc previously set to read "." as missing data marker. If . still appears in data, use Juweiriya's version of this code
```{r Trackloss}
dat007Look %>% mutate(TrackLoss = ifelse(is.na(dat007Look$RIGHT_GAZE_X) & is.na(dat007Look$RIGHT_GAZE_Y) 
                                        & is.na(dat007Look$LEFT_GAZE_X) & is.na(dat007Look$RIGHT_GAZE_Y), TRUE, FALSE)) -> dat007Look
```


##MAKE AOI COLUMNS
```{r}
#check what IAS appear
unique(dat007Look$RIGHT_INTEREST_AREA_LABEL)

#Creates columns with T/F values for each IA for each sample (TRUE if that IA is being looked at & not missing)
dat007Look %>% mutate(Outside = ifelse((is.na(RIGHT_INTEREST_AREA_LABEL) & is.na(LEFT_INTEREST_AREA_LABEL)), TRUE, FALSE)) %>% 
mutate(Target = ifelse((RIGHT_INTEREST_AREA_LABEL =="TARGET_IA" & !is.na(RIGHT_INTEREST_AREA_LABEL)) | (LEFT_INTEREST_AREA_LABEL =="TARGET_IA" & !is.na(LEFT_INTEREST_AREA_LABEL)), TRUE, FALSE)) %>% 
mutate(Dis1 = ifelse((RIGHT_INTEREST_AREA_LABEL == "DIS_ACTIONRELATED_IA" & !is.na(RIGHT_INTEREST_AREA_LABEL)) |
                                   (LEFT_INTEREST_AREA_LABEL == "DIS_ACTIONRELATED_IA" & !is.na(LEFT_INTEREST_AREA_LABEL)), 
                                    TRUE, FALSE)) %>% 
mutate(Dis2 = ifelse((RIGHT_INTEREST_AREA_LABEL == "DIS_AGENTRELATED_IA" & !is.na(RIGHT_INTEREST_AREA_LABEL)) |
                                   (LEFT_INTEREST_AREA_LABEL == "DIS_AGENTRELATED_IA" & !is.na(LEFT_INTEREST_AREA_LABEL)), 
                                   TRUE, FALSE)) %>% 
mutate(Dis3 = ifelse((RIGHT_INTEREST_AREA_LABEL == "DIS_UNRELATED_IA" & !is.na(RIGHT_INTEREST_AREA_LABEL)) |
                                   (LEFT_INTEREST_AREA_LABEL == "DIS_UNRELATED_IA" & !is.na(LEFT_INTEREST_AREA_LABEL)),
                                   TRUE, FALSE)) -> dat007Look


#additionally recodes looks to cross IA or question block IA as outside
dat007Look$Outside[(dat007Look$RIGHT_INTEREST_AREA_LABEL == "CROSS"  & !is.na(dat007Look$RIGHT_INTEREST_AREA_LABEL)) |
(dat007Look$LEFT_INTEREST_AREA_LABEL == "CROSS" & !is.na(dat007Look$LEFT_INTEREST_AREA_LABEL)) ] <- TRUE

# dat007Look$Outside[(dat007Look$RIGHT_INTEREST_AREA_LABEL == "RECTANGLE_INTERESTAREA"  & !is.na(dat007Look$RIGHT_INTEREST_AREA_LABEL)) |
#                (dat007Look$LEFT_INTEREST_AREA_LABEL == "RECTANGLE_INTERESTAREA" & !is.na(dat007Look$LEFT_INTEREST_AREA_LABEL)) ] <- TRUE

#Create and fill out group column
dat007Look %>% mutate(Group = ifelse(Subject == "007_06" | Subject =="007_08"| 
                                 Subject =="007_10"| Subject =="007_13"| 
                                 Subject =="007_19"| Subject =="007_29"| 
                                 Subject =="007_34"| Subject =="007_35"| 
                                 Subject =="007_36"| Subject =="007_37"|
                                 Subject =="007_41"| Subject =="007_43"| 
                                 Subject =="007_44"| Subject =="007_45"| 
                                 Subject =="007_49"| Subject =="007_51"| 
                                 Subject =="007_52"| Subject =="007_54"| 
                                 Subject =="007_55"| Subject =="007_56"| 
                                 Subject =="007_60"| Subject =="007_61"|
                                 Subject =="007_63"| Subject =="007_69"|
                                 Subject =="007_71", "CONTROL", "PD")) -> dat007Look
```

# Select rows of interest
These parameters select AVT non-practice trials only & no question trials.
Excludes 007_32 due to closing eyes; 37 due to trackloss; 01 pilot
```{r}

#filter in desired trial types
dat007Look <- dat007Look[ which(dat007Look$blockgroup==3),]
dat007Look <- dat007Look[ which(dat007Look$attnquestion==0),]
dat007Look <- dat007Look[ which(dat007Look$Subject != "007_01"),]
dat007Look <- dat007Look[ which(dat007Look$Subject != "007_32"),]
dat007Look <- dat007Look[ which(dat007Look$Subject != "007_37"),]
```

# Check number of subjects and trials per subject
```{r}
# subjects
length(unique(dat007Look$Subject))
# trials
ddply(dat007Look, c("Subject"), summarize, N=length(unique(TRIAL_LABEL)))
length(unique(dat007Look$trial_name))

#uncomment this section to view individual trials present in data
#ddply(dat007Look, c("Subject"), summarize, (unique(TRIAL_LABEL)))
```

# Make Eyetracking R Data
```{r}
datLookER <- make_eyetrackingr_data(dat007Look,
                               participant_column = "Subject",
                               trial_column = "TRIAL_INDEX",
                               time_column = "TIMESTAMP",
                               trackloss_column = "TrackLoss",
                               aoi_columns = c('Target','Dis1', 'Dis2', 'Dis3'),
                               treat_non_aoi_looks_as_missing = TRUE
                               )

datLookER %>% mutate(UniqueTrial = paste(datLookER$Subject, datLookER$trial_name, sep = '_'))-> datLookER 

```

#Set the exported IP to timestamp of 0.
Begins right at the presentation of article sound, therefore includes initial ~200ms where looks are not driven by the played sound.

```{r}
look_window_plus <- subset_by_window(datLookER, rezero = TRUE, remove = TRUE,
                                    window_start_col = "IP_START_TIME", 
                                    window_end_col = "IP_END_TIME")

```


#Create time sequence for cleaning pre-anticipatory looks (includes initial 200ms)
```{r}
#create time sequence
look_time_seq_plus <-make_time_sequence_data(look_window_plus, time_bin_size = 50,
                                   predictor_columns = c("Group", "trial_name", "audiofile", "UniqueTrial"),
                                   aois = c("Target", "Dis1", "Dis2", "Dis3")
                                   )

#sent_time_seq[ which(sent_time_seq$TimeBin < 10),] -> sent_time_seq_mini
#sent_time_seq_mini <- sent_time_seq_mini[ , -which(names(sent_time_seq_mini) %in% c("ot1","ot2", "ot3","ot4", "ot5","ot6", "ot7"))]
#sent_time_seq_mini <- sent_time_seq_mini [ sent_time_seq_mini$Subject %in% c("007_02","007_03", "007_06"), ]


```

#Identify trials where Ss primarily looked to target in 50ms preceeding sentence onset
Do this before removing trials for trackloss (?)
```{r}
#trials to exclude/model: rows where time bin = 3 (149-200ms based on full sentence window)  &  AOI = target  &  Prop > .4999

#create a dataframe containing just time bin 3 for pre-anticipated trials in original data frame
trials_reference <- look_time_seq_plus[ which(look_time_seq_plus$TimeBin == 3),]

#label which of those rows were pre-anticipated (proportion looks to target >.49999)
trials_reference %>% mutate(pre_anticipated = ifelse(trials_reference$AOI=="Target" & trials_reference$TimeBin == 3 & trials_reference$Prop > .49999, TRUE, FALSE)) -> trials_reference

#remove trials in look_window_plus where UniqueTrial ID matches any UniqueTrial ID in pre_anticipated trials
# Then you use that subset of the data as a reference, and alter the original data by adding a new column (e.g., "drop2", but could be whatever) that, for every participant-trial id, looks up that id in the lookup data and assigns the value in "drop" to the new column.

look_window_plus <- look_window_plus %>%
mutate(to_remove = trials_reference$pre_anticipated[
    match(unlist(look_window_plus$UniqueTrial),
          trials_reference$UniqueTrial)])

#drop "to remove" items
filter(look_window_plus, look_window_plus$to_remove==FALSE) -> look_window_plus_clean
filter(look_window_plus, look_window_plus$to_remove==TRUE) -> removed_samples

```


```{r See pre-anticipated trials for early subjects, eval=FALSE, echo=FALSE}
# Next: OPTIONAL, get list of pre-anticipated trials for subset of subjects (used to find non-anticipated trials to use as the bases for interpolating trials later on)

pre_anticipated_trials <- look_time_seq_plus[which(look_time_seq_plus$AOI=="Target" & look_time_seq_plus$TimeBin == 3 & look_time_seq_plus$Prop > .49999),]

pre_anticipated_trials_earlySs <- pre_anticipated_trials[ which(pre_anticipated_trials$Subject == "007_02" | pre_anticipated_trials$Subject =="007_03"| pre_anticipated_trials$Subject =="007_04"| pre_anticipated_trials$Subject =="007_05"| pre_anticipated_trials$Subject =="007_06"),]
write.csv(pre_anticipated_trials_Look_earlySs, file = "earlySs_anticipated_look_trials.csv")

```
Finds 241 trials (out of 960). 25.1% of trials began with looks primarily to target

# Zoom in to sentence window duration (966ms) adjusted by 200. adds 50ms to allow extension of target analysis window
Rezeros as well, so all times are now saccade-adjusted. No need for further adjustments.
```{r}
look_window <- subset_by_window(look_window_plus, #OR look_window_plus_clean
                                window_start_time = 200, 
                                window_end_time = 1216,
                                rezero = TRUE, remove = TRUE)
```


# Assess and Clean Trackloss
```{r}
# analyze amount of trackloss by subjects and trials
(trackloss <- trackloss_analysis(data = look_window))

#remove trials with trackloss greater than threshold (here, 25%)
look_window_clean <- clean_by_trackloss(data = look_window, trial_prop_thresh = .25)

#re-assess trackloss following cleaning
trackloss_clean <- trackloss_analysis(data = look_window_clean)

#prints mean trackloss by participants following cleaning of trials >threshold loss
(trackloss_clean_subjects <- unique(trackloss_clean[, c('Subject','TracklossForParticipant')]))

# get mean%  samples contributed per trial, with SD
mean(1 - trackloss_clean_subjects$TracklossForParticipant)
sd(1- trackloss_clean_subjects$TracklossForParticipant)

# See number of (non-lossy) trials contributed by subject in the NumTrials column
(final_summary <- describe_data(look_window_clean, 'Target', 'Subject'))

#see mean/SD of trials remaining after cleaning
mean(final_summary$NumTrials)
sd(final_summary$NumTrials)
```
###Trackloss set to 25%, subjects 32 & 37 removed, pre-anticipated trials removed, based on look window:
25 trials removed for trackloss. Retains 98.11% of trials? saccades? (2.03%). Mean remaining trials is 13.21, sd 2.71.
Highest remaining % of trackloss is 8.3% (subject 51)
Lowest # of remaining trials is 4 (again subject 51, unclear whether driven mainly by trackloss or predictive looks).




#Save this file as base set of trials for subsequent window analyses
```{r Save eyetrackingR data cleaned for target-initial trials and trackloss}
#saveRDS(look_window_clean, file = "007_Look_data_clean.rds")
saveRDS(look_window_clean, file = "007_Look_data_clean_alltrials.rds")
```
saved 4/27/21. version without all trials saved 3/30/20

```{r load clean, interpolated verb time sequence}
#look_window_clean <- readRDS("007_Look_data_clean.rds", refhook = NULL)
look_window_clean <- readRDS("007_Look_data_clean_alltrials.rds", refhook = NULL)

#load required packages
library("Matrix")
library("plyr")
library("dplyr")
library("tidyr")
library("lme4")
library(lmerTest)
library("ggplot2")
library("eyetrackingR")
library(sjstats)
```

# Make and Plot Time Sequence
##Full sentence window
```{r}
#create time sequence
look_time_seq <-make_time_sequence_data(look_window_clean, time_bin_size = 50,
                                   predictor_columns = c("Group", "trial_name", "list")
                                   )
```


#Interpolate values for set 9 trials for subjects 2-6.
007_02: list 9
007_03: list 10
007_04: list 10
007_05: list 1
007_06: list 2

```{r Create vectors for interpolated trials (problem trials)}

# store vectors for: original trial info; props for each bin in surrounding 2 trials; averaged new trial.

#SUBJECT 007_05
#interpolate trial Q/17
`i_05_16` <- subset(look_time_seq, Subject == "007_05" & trial_name == "ExpB_Item 16", 
                    select = c("Prop"))
`i_05_15` <- subset(look_time_seq, Subject == "007_05" & trial_name == "ExpB_Item 15", 
                    select = c("Prop"))
look_time_seq[which(look_time_seq$Subject == "007_05" & look_time_seq$trial_name == "ExpB_Item 17"), "Prop"] <- ((`i_05_16` + `i_05_15`)/2)

#interpolate trial J/10
`i_05_8` <- subset(look_time_seq, Subject == "007_05" & trial_name == "ExpB_Item 8", 
                    select = c("Prop"))
`i_05_14` <- subset(look_time_seq, Subject == "007_05" & trial_name == "ExpB_Item 14", 
                    select = c("Prop"))
look_time_seq[which(look_time_seq$Subject == "007_05" & look_time_seq$trial_name == "ExpB_Item 10"), "Prop"] <- ((`i_05_8` + `i_05_14`)/2)


#SUBJECT 02
#interpolate trial J/10
`i_02_4` <- subset(look_time_seq, Subject == "007_02" & trial_name == "ExpB_Item 4", 
                    select = c("Prop"))
`i_02_16` <- subset(look_time_seq, Subject == "007_02" & trial_name == "ExpB_Item 16", 
                    select = c("Prop"))
look_time_seq[which(look_time_seq$Subject == "007_02" & look_time_seq$trial_name == "ExpB_Item 10"), "Prop"] <- ((`i_02_4` + `i_02_16`)/2)


#interpolate trial Q/17
`i_02_15` <- subset(look_time_seq, Subject == "007_02" & trial_name == "ExpB_Item 15", 
                    select = c("Prop"))
`i_02_5` <- subset(look_time_seq, Subject == "007_02" & trial_name == "ExpB_Item 5", 
                    select = c("Prop"))
look_time_seq[which(look_time_seq$Subject == "007_02" & look_time_seq$trial_name == "ExpB_Item 17"), "Prop"] <- ((`i_02_15` + `i_02_5`)/2)


#SUBJECT 06
#interpolate trial Q/17
`i_06_15` <- subset(look_time_seq, Subject == "007_06" & trial_name == "ExpB_Item 15", 
                    select = c("Prop"))
`i_06_11` <- subset(look_time_seq, Subject == "007_06" & trial_name == "ExpB_Item 11", 
                    select = c("Prop"))
look_time_seq[which(look_time_seq$Subject == "007_06" & look_time_seq$trial_name == "ExpB_Item 17"), "Prop"] <- ((`i_06_15` + `i_06_11`)/2)

#SUBJECT 04
#interpolate trial J/10
`i_04_12` <- subset(look_time_seq, Subject == "007_04" & trial_name == "ExpB_Item 12", 
                    select = c("Prop"))
`i_04_6` <- subset(look_time_seq, Subject == "007_04" & trial_name == "ExpB_Item 6", 
                    select = c("Prop"))
look_time_seq[which(look_time_seq$Subject == "007_04" & look_time_seq$trial_name == "ExpB_Item 10"), "Prop"] <- ((`i_04_12` + `i_04_6`)/2)
```


```{r Save interpolated data file}
#save the cleaned file to run in verb/object window analysis scripts. Or plot/analyse overall sentence below.
#saveRDS(look_time_seq, file = "007_Look_data_clean_interpolated.rds")
saveRDS(look_time_seq, file = "007_Look_data_clean_interpolated_alltrials.rds")

```

```{r Load previously cleaned data}
#look_time_seq <- readRDS("007_Look_data_clean_interpolated.rds", refhook = NULL)
look_time_seq <- readRDS("007_Look_data_clean_interpolated_alltrials.rds", refhook = NULL)

look_time_seq$Group <- as.factor(look_time_seq$Group)
look_time_seq$AOI <- as.factor(look_time_seq$AOI)
look_time_seq$list <- as.factor(look_time_seq$list)

#load required packages

library(Matrix)
library(plyr)
library(dplyr)
library(tidyr)
library(lme4)
library(lmerTest)
library(ggplot2)
library(eyetrackingR)
```

#final prep for plotting/modeling
```{r final prep for plotting/modeling}

#create column for natural quadratic time polynomial, if using; and to convert looks to binomial
look_time_seq %>% 
  mutate(Time2 = Time^2) %>% 
  mutate(BiProp = ifelse(Prop > .5, 1, 0)) -> look_time_seq
#center groupvariable/use sum contrasts
contrasts(look_time_seq$Group) <- c(-.5, .5)

#change order of levels in AOI factor in order to change plot default
levels(as.factor(look_time_seq$AOI))
look_time_seq$AOI <- factor(look_time_seq$AOI, levels = c("Target", "Dis1", "Dis2", "Dis3"))

#create dataframe with average proportions from each subject, splits AOIs into different columns
look_time_seq %>%
  filter(!is.na(Prop)) %>%
  group_by(Subject, AOI, Time, Group) %>% 
  summarise(MeanProp = mean(Prop), y=sum(Prop), N=length(Prop), Time=min(Time)) ->look_time_seq_avg

#add new column that converts probabilities to logits
look_time_seq_avg$Logit <- qlogis(look_time_seq_avg$MeanProp)

```
Added manually group-centered data (PD -> .5, Cx -> -.5), but outputs look the same. r automatically centers categoricals, so better to use in case of uneven n across?

# visualize time results
```{r}

tiff("Z:\\LCAN_Internal\\LCAN_007_LangPrediction_QRP\\Analyses\\ET_Data\\Results\\Look_plot_8_13_21.tiff", units="in", width=12, height=7.5, res=300)
print(ggplot(look_time_seq, aes(x = Time, y = BiProp, col= AOI)) +
  stat_summary(geom="line", size = 2) +  
  stat_summary(fun.data = mean_se, geom = "errorbar",  aes(color=paste("mean_se", AOI)))+
  theme_minimal() +
  theme(legend.position="NULL", legend.box = "horizontal") +
  theme(legend.text=element_text(size=rel(1.2))) +
  expand_limits(x = 1300) +
  facet_grid(. ~ Group) +
  geom_vline(xintercept=391) +
  annotate(geom="text", x=540, y=.97, label="[target]'",
              color="black", size=6.5) +
  annotate(geom="text", x=170, y=.97, label="'Look at the",
              color="gray58", size=6.5) +
  annotate(geom="text", x=1190, y=.04, label="Distractors",
              color="gray58", size=6.5) +
    annotate(geom="text", x=1150, y=.9, label="Target",
              color="black", size=6.5) +
  scale_color_manual(labels = c("Target", "Distractor 1", "Distractor 2", "Distractor 3"), 
               values=c('gray85','gray68','gray48', 'gray85', 'gray68','gray48', 'black', 'black' )) +
  labs(title="Non-Predictive Sentences",
        x ="Time (ms)", y = "Proportion of Trials") +
  theme(strip.text.x = element_text(size = 18, colour = "black", angle = 0)) +
  theme(text = element_text(size=18)))
dev.off()

```

```{r visualize baseline sentence results logit scale}
tiff("Z:\\LCAN_Internal\\LCAN_007_LangPrediction_QRP\\Analyses\\ET_Data\\Results\\Look_plot_logit_11_17.tiff", units="in", width=12, height=7.5, res=300)

print(ggplot(look_time_seq_avg, aes(x = Time, y = Logit, col= AOI)) +
  stat_summary(geom="line", size = 2) +  
  stat_summary(fun.data = mean_se, geom = "errorbar",  aes(color=paste("mean_se", AOI)))+
  theme_minimal() +
  theme(legend.position="NULL", legend.box = "horizontal") +
  theme(legend.text=element_text(size=rel(1.2))) +
  expand_limits(x = 1300) +
  facet_grid(. ~ Group) +
  geom_vline(xintercept=391) +
  annotate(geom="text", x=540, y=1.8, label="[target]'",
              color="black", size=6.5) +
  annotate(geom="text", x=170, y=1.8, label="'Look at the",
              color="gray58", size=6.5) +
  annotate(geom="text", x=1190, y=-3, label="Distractors",
              color="gray58", size=6.5) +
    annotate(geom="text", x=1150, y=2, label="Target",
              color="black", size=6.5) +
  scale_color_manual(labels = c("Target", "Distractor 1", "Distractor 2", "Distractor 3"), 
               values=c('gray85','gray68','gray48', 'gray85', 'gray68','gray48', 'black', 'black' )) +
  labs(title="Baseline Sentences",
        x ="Time (ms)", y = "Log odds of fixations on each AOI") +
  theme(strip.text.x = element_text(size = 18, colour = "black", angle = 0)) +
  theme(text = element_text(size=18)))

dev.off()

```

```{r Test whether lines differ from chance in beginning of look window}
# #subset and test dis1
# look_time_seq_early_dis1 <- look_time_seq[which(look_time_seq$TimeBin==0 & look_time_seq$AOI =="Dis1"),]
# 
# look_time_seq_early_dis1_PD <- look_time_seq[which(look_time_seq$TimeBin==0 & look_time_seq$AOI =="Dis1" & look_time_seq$Group == "PD"),]
# look_time_seq_early_dis1_Cx <- look_time_seq[which(look_time_seq$TimeBin==0 & look_time_seq$AOI =="Dis1" & look_time_seq$Group == "CONTROL"),]
# t.test(look_time_seq_early_dis1_PD$BiProp, mu=.333) #0.1135
# t.test(look_time_seq_early_dis1_Cx$BiProp, mu=.333) #0.6373
# 
# #subset and test dis2
# look_time_seq_early_dis2 <- look_time_seq[which(look_time_seq$TimeBin==0 & look_time_seq$AOI =="Dis2"),]
# t.test(look_time_seq_early_dis2$BiProp, mu=.333)
# 
# look_time_seq_early_dis2_PD <- look_time_seq[which(look_time_seq$TimeBin==0 & look_time_seq$AOI =="Dis2" & look_time_seq$Group == "PD"),]
# look_time_seq_early_dis2_Cx <- look_time_seq[which(look_time_seq$TimeBin==0 & look_time_seq$AOI =="Dis2" & look_time_seq$Group == "CONTROL"),]
# t.test(look_time_seq_early_dis2_PD$BiProp, mu=.333) #0.2894
# t.test(look_time_seq_early_dis2_Cx$BiProp, mu=.333) #0.08396
# mean(look_time_seq_early_dis2_Cx$BiProp) #0.2885
# 
# #subset and test dis3
# look_time_seq_early_dis3 <- look_time_seq[which(look_time_seq$TimeBin==0 & look_time_seq$AOI =="Dis3"),]
# t.test(look_time_seq_early_dis3$BiProp, mu=.333)
# 
# look_time_seq_early_dis3_PD <- look_time_seq[which(look_time_seq$TimeBin==0 & look_time_seq$AOI =="Dis3" & look_time_seq$Group == "PD"),]
# look_time_seq_early_dis3_Cx <- look_time_seq[which(look_time_seq$TimeBin==0 & look_time_seq$AOI =="Dis3" & look_time_seq$Group == "CONTROL"),]
# 
# t.test(look_time_seq_early_dis3_PD$BiProp, mu=.333)  #0.4778
# t.test(look_time_seq_early_dis3_Cx$BiProp, mu=.333)  #0.5456


```