Social Cross-Situational Word Learning
Kyle MacDonald & Allison Dods
July 14, 2014
Analysis code for SOC-XSIT project.
Load libraries.
library(bootstrap)
library(lme4)
library(ggplot2)
library(plyr)
library(dplyr)
source("/Users/kmacdonald/Documents/Projects/SOC_XSIT/XSIT-MIN/analysis/Ranalysis/useful.R")Read in data.
- d: manipulated gaze length, 4-referents, 0 delay
- d1: ran the short gaze in social and no-social condition
- d2: replicate expt 2 using a within-subjects design
- d3: manipulated the language cue during exposure (“this” vs. “one”), eye gaze in both
- d4: aggregated data from FYP experiment
#set path for reading data
read_path <- file.path("/Users", "kmacdonald", "Documents",
"Projects", "SOC_XSIT", "processed_data",
"adult-looks/")
read_path_fyp <- file.path("/Users", "kmacdonald", "Documents",
"Projects", "SOC_XSIT", "processed_data",
"adult-fyp/")
d <- tbl_df(read.csv(paste(read_path,
"soc_xsit_looks_pilot.csv", sep="")))
d1 <- tbl_df(read.csv(paste(read_path,
"soc_xsit_looks_pilot_2.csv", sep="")))
d2 <- tbl_df(read.csv(paste(read_path,
"soc_xsit_looks_pilot_3_withinsubs.csv", sep="")))
d3 <- tbl_df(read.csv(paste(read_path,
"soc_xsit_looks_pilot_4_thisone_btw.csv", sep="")))
d4 <- tbl_df(read.csv(paste(read_path_fyp,
"aggregate_soc_xsit.csv", sep="")))Merge data sets, tracking experiment version.
#remove samepos and random "X" column from d/d1
#rename example, exposure, and test columns in d/d1
d <- d %>%
select(-X, -samePos,
example_trial = exampleTrial,
exposure_trial = exposureTrial,
test_trial = testTrial,
matches(".")
)
d1 <- d1 %>%
select(-X, -samePos,
example_trial = exampleTrial,
exposure_trial = exposureTrial,
test_trial = testTrial,
matches(".")
)
#add experiment tracking variable to each data frame
d <- mutate(d, experiment = "look-length")
d1 <- mutate(d1, experiment = "soc_vs_no-soc_btw")
d2 <- mutate(d2, experiment = "soc_vs_no_soc_within")
d3 <- mutate(d3, experiment = "this_one")
# wrap in list
d.list <- list(d, d1, d2, d3)
#join all 4 data frames
d.all_df <- ldply(.data = d.list, .fun = rbind())
#check for people who participated more than once
temp <- d.all_df %>%
group_by(subid) %>%
summarise(
n = n(),
n_unique = n_distinct(subid)
)
#anonymize subids
d.all_df <- anonymize.sids(d.all_df, "subid")
d.all_df$experiment <- as.factor(d.all_df$experiment)Get the number of subjects in each experiment and condition.
## Source: local data frame [9 x 4]
## Groups: experiment, condition
##
## experiment condition gazeLength n_subs
## 1 look-length Social Long 48
## 2 look-length Social Medium 50
## 3 look-length Social Short 46
## 4 soc_vs_no_soc_within No-socialFirst Short 46
## 5 soc_vs_no_soc_within SocialFirst Short 47
## 6 soc_vs_no-soc_btw Social Short 47
## 7 soc_vs_no-soc_btw No-social Short 51
## 8 this_one One Short 49
## 9 this_one This Short 48Flag whether participants chose target of eye gaze on exposure.
#change values of faceIdx to correspond to values of chosenIdx
#(LU = 0, RU = 1, LD = 2, RD = 3, striaight = -1)
d.all_df$faceIdx <- revalue(d.all_df$face,
c("silentLUlong"= 0, "silentLUmedium" = 0,
"silentLUshort"= 0, "LUkidslonger" = 0,
"silentRUlong" = 1, "silentRUmedium"= 1,
"silentRUshort" = 1, "RUkidslonger" = 1,
"silentLDlong" = 2, "silentLDmedium" = 2,
"silentLDshort" = 2, "LDkidslonger" = 2,
"silentRDlong" = 3, "silentRDmedium" = 3,
"silentRDshort" = 3, "RDkidslonger" = 3,
"straightahead" = -1, "straightaheadlonger" = -1))
#flag trial, if subs chose target of gaze
d.expo_df <- d.all_df %>%
filter(exposure_trial == 1) %>%
mutate(correct_exposure = faceIdx == chosenIdx) %>%
select(subids, itemNum, correct_exposure)Get test trials and merge exposure trial information.
d.test_df <- d.all_df %>%
filter(test_trial == 1)
d.test_df <- join(d.expo_df, d.test_df, match = "all")## Joining by: subids, itemNumFlag subs who got <25% of test trials correct.
d.test_df <- d.test_df %>%
group_by(subids) %>%
summarise(mean_acc_exp = mean(correct_exposure)) %>%
mutate(include_expo = ifelse(mean_acc_exp > 0.25,1,0)) %>%
join(d.test_df, by = "subids")Flag trials with extremely slow or fast RTs (+/- 2SD).
d.test_df <- d.test_df %>%
mutate(include_good_rt = ifelse(log(rt) > mean(log(rt)) + 2 * sd(log(rt)) |
log(rt) < mean(log(rt)) - 2 * sd(log(rt)),
0,1)
)Summarize data set: means for acc on exposure and test
Experiment 1: Look-length manipulation.
Accuracy on exposure trials.
ms_exp_looks <- d.test_df %>%
filter(experiment == "look-length", include_good_rt == 1,
include_expo == 1) %>%
group_by(gazeLength) %>%
summarise(accuracy_exposure = mean(correct_exposure),
ci_low = ci.low(correct_exposure),
ci_high = ci.high(correct_exposure))
ms_exp_looks$gazeLength <- factor(ms_exp_looks$gazeLength,
levels = c("Short", "Medium", "Long"))
ggplot(data=ms_exp_looks,
aes(x=gazeLength, y=accuracy_exposure,
fill=gazeLength)) +
geom_bar(stat="identity") +
geom_errorbar(aes(ymin=accuracy_exposure-ci_low,
ymax=accuracy_exposure+ci_high), width = .1) +
geom_hline(yintercept=0.25, linetype = "dashed") +
scale_y_continuous(limits=c(0,1))
Accuracy on test trials: same/switch
ms_test_looks <- d.test_df %>%
filter(experiment == "look-length", include_good_rt == 1,
include_expo == 1, correct_exposure == TRUE) %>%
group_by(gazeLength, trialType) %>%
summarise(accuracy = mean(correct),
ci_low = ci.low(correct),
ci_high = ci.high(correct))
ms_test_looks$gazeLength <- factor(ms_test_looks$gazeLength,
levels = c("Short", "Medium", "Long"))
ggplot(ms_test_looks,
aes(x=gazeLength, y=accuracy,
group=trialType, colour=trialType)) +
geom_point(size=2) +
geom_line() +
geom_errorbar(aes(ymin=accuracy - ci_low,
ymax=accuracy + ci_high), width = .1) +
geom_hline(yintercept=0.25, linetype = "dashed") +
scale_y_continuous(limits=c(0,1))
Accuracy on exposure trials.
ms_exp_btw <- d.test_df %>%
filter(experiment == "soc_vs_no-soc_btw", condition == "Social",
include_good_rt == 1, include_expo == 1) %>%
summarise(accuracy_exposure = mean(correct_exposure),
ci_low = ci.low(correct_exposure),
ci_high = ci.high(correct_exposure))
ms_exp_btw## accuracy_exposure ci_low ci_high
## 1 0.8775 0.03974 0.03642Accuracy on test trials
ms_test_btw <- d.test_df %>%
filter(experiment == "soc_vs_no-soc_btw",
include_good_rt == 1,
include_expo == 1 | condition == "No-social") %>%
group_by(condition, trialType) %>%
summarise(accuracy_test = mean(correct),
ci_low = ci.low(correct),
ci_high = ci.high(correct))
#now plot
ggplot(ms_test_btw,
aes(x=condition, y=accuracy_test,
group=trialType, colour=trialType)) +
geom_point(size=2.5) +
geom_line() +
geom_errorbar(aes(ymin=accuracy_test - ci_low,
ymax=accuracy_test + ci_high), width = .1) +
geom_hline(yintercept=0.25, linetype = "dashed") +
scale_y_continuous(limits=c(0,1)) +
xlab("Condition") +
ylab("Proportion Correct") +
labs(colour = "Trial Type")
Accuracy on exposure trials
Replot fyp data
d4$condition <- factor(d4$condition,
levels = c("Social", "No-Social"))
ggplot(data=filter(d4, intervalNum==0, numPicN==4),
aes(x=condition, y=correct, colour=trialType, group=trialType)) +
geom_point(size = 2.5) +
geom_line() +
geom_errorbar(aes(ymin=correct - corr.cil,
ymax=correct + corr.cih), width = .1) +
geom_hline(yintercept=0.25, linetype = "dashed") +
scale_y_continuous(limits=c(0,1)) +
xlab("Condition") +
ylab("Proportion Correct") +
labs(colour = "Trial Type")