TAB_COMP
Tablet comparison script.
Set up some functions.
rm(list=ls())
library(xtable)
library(psych)
library(tidyr)
library(stringr)
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(ggplot2)##
## Attaching package: 'ggplot2'## The following objects are masked from 'package:psych':
##
## %+%, alphalibrary(bootstrap)
library(lme4)## Loading required package: Matrix##
## Attaching package: 'Matrix'## The following object is masked from 'package:tidyr':
##
## expand## for bootstrapping 95% confidence intervals
theta <- function(x,xdata,na.rm=T) {mean(xdata[x],na.rm=na.rm)}
ci.low <- function(x,na.rm=T) {
mean(x,na.rm=na.rm) - quantile(bootstrap(1:length(x),1000,theta,x,na.rm=na.rm)$thetastar,.025,na.rm=na.rm)}
ci.high <- function(x,na.rm=T) {
quantile(bootstrap(1:length(x),1000,theta,x,na.rm=na.rm)$thetastar,.975,na.rm=na.rm) - mean(x,na.rm=na.rm)}Reading in experimental data
Demographics.
tab.demo <- read.csv("../data/tablet_demographics.csv")
tab.demo <- tab.demo %>% rename(subid = SID)
story.demo <- read.csv("../data/story_demographics.csv")
story.demo <- story.demo %>% rename(subid = SID)
et.demo <- read.csv("../data/et_demographics.csv")
et.demo <- et.demo %>% rename(subid = SID)First read tablet.
tab <- read.csv("../data/tabletstudyresults.csv")
tab$reaction.time <- as.numeric(as.character(tab$reaction.time))
tab$trial.type <- factor(tab$trial.type,
levels = c("rec","MEcontrol","MEexperimental","filler"),
labels = c("FAM-fam","FAM-nov","NOV-fam","filler"))Then read storybook.
story <- read.csv("../data/storystudyresults.csv")
story <- story %>%
gather(trial, correct, ends_with(".Correct"), na.rm=FALSE) %>%
select(-starts_with("X")) %>%
mutate(trial = as.numeric(str_replace_all(str_replace(trial, "X", ""), ".Correct", "")))
lists <- read.csv("../data/lists.csv")
story <- left_join(story, lists, by=c("List","trial"))
story$trial.type <- factor(story$trial.type,
levels = c("familiar-familiar","FAMILIAR-novel",
"familiar-NOVEL","filler"),
labels = c("FAM-fam","FAM-nov","NOV-fam", "filler"))Then read eye-tracking.
et <- read.csv("../eye_tracking/R/eye.tracking.csv", row.names="X")
et$trial.type <- factor(et$word.type,
levels = c("Familiar-Familiar","Familiar-Novel",
"Novel-Familiar"),
labels = c("FAM-fam","FAM-nov","NOV-fam"))Merge these three together. This is quite ugly because of differing naming conventions across data files.
et <- et %>%
select(sid, whole.trial.num, trial.type, rt, prop) %>%
rename(subid = sid,
trial.num = whole.trial.num,
correct = prop)
et <- left_join(et, et.demo, by="subid")## Warning in left_join_impl(x, y, by$x, by$y, suffix$x, suffix$y): joining
## factors with different levels, coercing to character vectoret$condition <- "eye-tracker"
et$rt <- et$rt * 1000 # convert to ms
tab <- tab %>%
mutate(correct = response=="Y") %>%
select(subject.id, trial.number, trial.type, reaction.time, correct) %>%
rename(subid = subject.id,
trial.num = trial.number,
rt = reaction.time)
tab <- left_join(tab, tab.demo)## Joining, by = "subid"## Warning in left_join_impl(x, y, by$x, by$y, suffix$x, suffix$y): joining
## factors with different levels, coercing to character vectortab$condition <- "tablet"
story <- story %>%
select(SID, trial, trial.type, correct) %>%
rename(subid = SID,
trial.num = trial)
story <- left_join(story, story.demo)## Joining, by = "subid"## Warning in left_join_impl(x, y, by$x, by$y, suffix$x, suffix$y): joining
## factors with different levels, coercing to character vectorstory$condition <- "storybook"
d <- rbind_list(et, tab, story) %>%
mutate(age.grp = floor(age))## Warning: `rbind_list()` is deprecated. Please use `bind_rows()` instead.## Warning in rbind_list__impl(environment()): Unequal factor levels: coercing
## to character
## Warning in rbind_list__impl(environment()): Unequal factor levels: coercing
## to character
## Warning in rbind_list__impl(environment()): Unequal factor levels: coercing
## to character
## Warning in rbind_list__impl(environment()): Unequal factor levels: coercing
## to character
## Warning in rbind_list__impl(environment()): Unequal factor levels: coercing
## to character
## Warning in rbind_list__impl(environment()): Unequal factor levels: coercing
## to character
## Warning in rbind_list__impl(environment()): Unequal factor levels: coercing
## to characterDemographics and filtering
Report function.
report <- function(x) {
x %>%
group_by(subid, condition, age.grp) %>%
summarise(age = age[1],
sex = sex[1]) %>%
group_by(condition, age.grp, add=FALSE) %>%
summarise(n=n(),
m.age = mean(age),
m.male = mean(sex=="male"))
}Filter for ages and filler trials.
d <- d %>%
filter(age.grp > 0, age.grp < 5, !is.na(age.grp))
d <- d %>% filter(trial.type != "filler")Non-inclusion criteria that don’t relate to methods.
d %>% group_by(subid, condition, exclusion.crit) %>%
summarise(n = 1) %>%
group_by(condition, exclusion.crit, add=FALSE) %>%
summarise(n=sum(n)) ## Source: local data frame [14 x 3]
## Groups: condition [?]
##
## condition exclusion.crit n
## <chr> <chr> <dbl>
## 1 eye-tracker 71
## 2 eye-tracker interference 4
## 3 eye-tracker lang 25
## 4 storybook 73
## 5 storybook dd 1
## 6 storybook interference 4
## 7 storybook lang 16
## 8 storybook non-compliant 5
## 9 tablet 86
## 10 tablet dd 1
## 11 tablet error 2
## 12 tablet interference 4
## 13 tablet lang 7
## 14 tablet non-compliant 1n <- d %>%
filter(exclusion.crit != "lang",
exclusion.crit != "dd")Now, other exclusion criteria.
d %>%
group_by(subid, condition, exclusion.crit) %>%
summarise(n = 1) %>%
group_by(condition, exclusion.crit, add=FALSE) %>%
summarise(n= sum(n)) %>%
group_by(condition, add=FALSE) %>%
mutate(prop = n/ sum(n))## Source: local data frame [14 x 4]
## Groups: condition [3]
##
## condition exclusion.crit n prop
## <chr> <chr> <dbl> <dbl>
## 1 eye-tracker 71 0.71000000
## 2 eye-tracker interference 4 0.04000000
## 3 eye-tracker lang 25 0.25000000
## 4 storybook 73 0.73737374
## 5 storybook dd 1 0.01010101
## 6 storybook interference 4 0.04040404
## 7 storybook lang 16 0.16161616
## 8 storybook non-compliant 5 0.05050505
## 9 tablet 86 0.85148515
## 10 tablet dd 1 0.00990099
## 11 tablet error 2 0.01980198
## 12 tablet interference 4 0.03960396
## 13 tablet lang 7 0.06930693
## 14 tablet non-compliant 1 0.00990099d <- d %>%
filter(exclude == 0)
xtable(report(d))## % latex table generated in R 3.3.0 by xtable 1.8-2 package
## % Tue Sep 6 16:07:06 2016
## \begin{table}[ht]
## \centering
## \begin{tabular}{rlrrrr}
## \hline
## & condition & age.grp & n & m.age & m.male \\
## \hline
## 1 & eye-tracker & 1.00 & 17 & 1.43 & 0.47 \\
## 2 & eye-tracker & 2.00 & 19 & 2.48 & 0.58 \\
## 3 & eye-tracker & 3.00 & 17 & 3.55 & 0.76 \\
## 4 & eye-tracker & 4.00 & 16 & 4.52 & 0.56 \\
## 5 & storybook & 1.00 & 15 & 1.62 & 0.60 \\
## 6 & storybook & 2.00 & 19 & 2.49 & 0.32 \\
## 7 & storybook & 3.00 & 19 & 3.48 & 0.53 \\
## 8 & storybook & 4.00 & 20 & 4.48 & 0.45 \\
## 9 & tablet & 1.00 & 18 & 1.71 & 0.44 \\
## 10 & tablet & 2.00 & 22 & 2.46 & 0.68 \\
## 11 & tablet & 3.00 & 24 & 3.56 & 0.67 \\
## 12 & tablet & 4.00 & 22 & 4.51 & 0.59 \\
## \hline
## \end{tabular}
## \end{table}Data-based exclusions
Did they finish?
qplot(trial.num, facets=~condition, data=d)## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
mst <- d %>% group_by(subid, condition, age.grp) %>%
summarise(max.trial = max(trial.num)) %>%
group_by(condition, age.grp, add=FALSE) %>%
summarise(finished = ifelse(condition[1]=="storybook", mean(max.trial==23),
mean(max.trial==28)),
num.trials = mean(max.trial))
qplot(age.grp, finished, col=condition,
geom="line", stat="identity",
position="dodge",
data=mst)## Warning: `stat` is deprecated## Warning: `position` is deprecated
Reaction time pruning
Reaction time distribution. We also remove incorrect trials for the tablet, and target-initial trials for the eye-tracker.
d$rt[d$condition == "tablet" & !d$correct] <- NA # start by removing false trials
d$rt[d$condition == "eye-tracker" & d$rt < 317] <- NA # remove T initial trials
d %>%
group_by(condition) %>%
summarise(m = mean(log(rt), na.rm=TRUE),
s = sd(log(rt), na.rm=TRUE))## # A tibble: 3 x 3
## condition m s
## <chr> <dbl> <dbl>
## 1 eye-tracker 6.855382 0.5848310
## 2 storybook NaN NA
## 3 tablet 7.561642 0.6179562qplot(rt/1000,
data=filter(d, condition != "storybook")) +
facet_grid(.~condition) +
# geom_vline(xintercept=exp(m - 2*s)/1000, col="red",lty=2) +
# geom_vline(xintercept=exp(m + 2*s)/1000, col="red",lty=2) +
scale_x_log10(breaks=c(1,2,5,10,20,50))## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.## Warning: Removed 722 rows containing non-finite values (stat_bin).
Age effects:
ggplot(d) +
geom_bar(aes(x = rt/1000, y=..count..)) +
scale_x_log10(breaks=c(1,5,10,50)) +
facet_grid(condition~age.grp)## Warning: Removed 2182 rows containing non-finite values (stat_count).
Prune excessive reaction times. (Arbitrary threshold).
d$rt[d$rt > 4000 |
d$rt < 500] <- NAActual analysis starts here
Aggregation.
mss <- d %>%
group_by(subid, trial.type, age.grp, condition) %>%
summarise(age=age[1],
rt=mean(rt, na.rm=TRUE),
correct=mean(correct))
mss$trial.type <- factor(mss$trial.type,
levels = c("FAM-fam","FAM-nov","NOV-fam"),
labels=c("Familiar Word","ME Control","ME Inference"))
ms <- mss %>%
group_by(trial.type, age.grp, condition, add=FALSE) %>%
summarise(rt.cih=ci.high(rt),
rt.cil=ci.low(rt),
rt=mean(rt,na.rm=TRUE),
acc.cih=ci.high(correct),
acc.cil=ci.low(correct),
acc=mean(correct,na.rm=TRUE)) Accuracy as a barplot.
ggplot(ms, aes(x = age.grp, y = acc, fill=trial.type)) +
geom_bar(stat = "identity",
position="dodge") +
geom_linerange(aes(ymin=acc-acc.cil, ymax=acc+acc.cih),
position=position_dodge(width=.9)) +
facet_grid(condition ~ .) +
xlab("Age (Years)") +
ylab("Accuracy") +
geom_hline(yintercept=.5,lty=2) +
scale_fill_discrete(name="Trial Type") 
Accuracy (continuous)
qplot(age, correct, col=trial.type, data=mss) +
facet_grid(. ~ condition) +
geom_smooth(method="lm", formula=y ~ boot::inv.logit(x)) +
geom_hline(yintercept=.5,lty=2) +
ylim(c(0,1.2))
and RT:
ggplot(filter(ms, condition != "storybook"),
aes(x = age.grp, y =rt/1000, fill=trial.type)) +
geom_bar(position=position_dodge(width=.9),
stat="identity") +
geom_linerange(aes(ymin=rt/1000-rt.cil/1000,
ymax=rt/1000+rt.cih/1000),
position=position_dodge(width=.9)) +
xlab("Age (Years)") +
ylab("Reaction Time (s)") +
scale_fill_discrete(name="Trial Type") +
facet_grid(condition ~ . , scales="free_y") 
Number of trials
Nearly everyone sticks it out to the end!
mst <- d %>%
group_by(subid, age.grp, condition) %>%
summarise(max.trial = max(trial.num))
qplot(max.trial, facets=condition~age.grp, data=mst)## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
and means across groups.
mstm <- mst %>%
group_by(age.grp, condition, add=FALSE) %>%
summarise(trials = mean(max.trial),
trials.cih = ci.high(max.trial),
trials.cil = ci.low(max.trial))
ggplot(mstm, aes(x = age.grp, y =trials)) +
geom_bar(aes(fill=factor(age.grp)),
stat="identity",
position=position_dodge(width=.9)) +
geom_linerange(aes(ymin=trials-trials.cil,
ymax=trials+trials.cih),
position=position_dodge(width=.9)) +
facet_grid(.~condition) +
geom_hline(yintercept=28, lty=2)
Statistics
t-tests vs. chance.
Accuracies for eye-tracker.
t.test(mss$correct[mss$condition == "eye-tracker" &
mss$trial.type == "Familiar Word" &
mss$age.grp == 1] - .5)##
## One Sample t-test
##
## data: mss$correct[mss$condition == "eye-tracker" & mss$trial.type == "Familiar Word" & mss$age.grp == 1] - 0.5
## t = 1.2337, df = 16, p-value = 0.2351
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## -0.03976846 0.15048986
## sample estimates:
## mean of x
## 0.0553607t.test(mss$correct[mss$condition == "eye-tracker" &
mss$trial.type == "ME Control" &
mss$age.grp == 1] - .5)##
## One Sample t-test
##
## data: mss$correct[mss$condition == "eye-tracker" & mss$trial.type == "ME Control" & mss$age.grp == 1] - 0.5
## t = 4.4867, df = 12, p-value = 0.0007437
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## 0.08842591 0.25538958
## sample estimates:
## mean of x
## 0.1719077t.test(mss$correct[mss$condition == "eye-tracker" &
mss$trial.type == "ME Inference" &
mss$age.grp == 1] - .5)##
## One Sample t-test
##
## data: mss$correct[mss$condition == "eye-tracker" & mss$trial.type == "ME Inference" & mss$age.grp == 1] - 0.5
## t = 0.51624, df = 11, p-value = 0.6159
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## -0.06401885 0.10325235
## sample estimates:
## mean of x
## 0.01961675t.test(mss$correct[mss$condition == "eye-tracker" &
mss$trial.type == "ME Inference" &
mss$age.grp == 2] - .5)##
## One Sample t-test
##
## data: mss$correct[mss$condition == "eye-tracker" & mss$trial.type == "ME Inference" & mss$age.grp == 2] - 0.5
## t = 1.9766, df = 17, p-value = 0.06454
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## -0.005588977 0.171484150
## sample estimates:
## mean of x
## 0.08294759Other accuracies.
t.test(mss$correct[mss$condition == "storybook" &
mss$trial.type == "Familiar Word" &
mss$age.grp == 1] - .5)##
## One Sample t-test
##
## data: mss$correct[mss$condition == "storybook" & mss$trial.type == "Familiar Word" & mss$age.grp == 1] - 0.5
## t = 3.9561, df = 14, p-value = 0.001434
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## 0.1024722 0.3451468
## sample estimates:
## mean of x
## 0.2238095t.test(mss$correct[mss$condition == "tablet" &
mss$trial.type == "Familiar Word" &
mss$age.grp == 1] - .5)##
## One Sample t-test
##
## data: mss$correct[mss$condition == "tablet" & mss$trial.type == "Familiar Word" & mss$age.grp == 1] - 0.5
## t = 2.5764, df = 17, p-value = 0.01961
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## 0.03227744 0.32420404
## sample estimates:
## mean of x
## 0.1782407t.test(mss$correct[mss$condition == "tablet" &
mss$trial.type == "ME Control" &
mss$age.grp == 1] - .5)##
## One Sample t-test
##
## data: mss$correct[mss$condition == "tablet" & mss$trial.type == "ME Control" & mss$age.grp == 1] - 0.5
## t = 2.2316, df = 17, p-value = 0.03939
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## 0.008170309 0.291168315
## sample estimates:
## mean of x
## 0.1496693Accuracy model.
acc.mod <- lmer(correct ~ age * trial.type * condition +
(trial.type | subid),
family="binomial",
data=filter(d, condition != "eye-tracker"))## Warning in lmer(correct ~ age * trial.type * condition + (trial.type |
## subid), : calling lmer with 'family' is deprecated; please use glmer()
## instead## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control
## $checkConv, : Model failed to converge with max|grad| = 0.243542 (tol =
## 0.001, component 1)summary(acc.mod)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: correct ~ age * trial.type * condition + (trial.type | subid)
## Data: filter(d, condition != "eye-tracker")
## Control:
## structure(list(optimizer = c("bobyqa", "Nelder_Mead"), calc.derivs = TRUE,
## use.last.params = FALSE, restart_edge = FALSE, boundary.tol = 1e-05,
## tolPwrss = 1e-07, compDev = TRUE, nAGQ0initStep = TRUE, checkControl = structure(list(
## check.nobs.vs.rankZ = "ignore", check.nobs.vs.nlev = "stop",
## check.nlev.gtreq.5 = "ignore", check.nlev.gtr.1 = "stop",
## check.nobs.vs.nRE = "stop", check.rankX = "message+drop.cols",
## check.scaleX = "warning", check.formula.LHS = "stop",
## check.response.not.const = "stop"), .Names = c("check.nobs.vs.rankZ",
## "check.nobs.vs.nlev", "check.nlev.gtreq.5", "check.nlev.gtr.1",
## "check.nobs.vs.nRE", "check.rankX", "check.scaleX", "check.formula.LHS",
## "check.response.not.const")), checkConv = structure(list(
## check.conv.grad = structure(list(action = "warning",
## tol = 0.001, relTol = NULL), .Names = c("action",
## "tol", "relTol")), check.conv.singular = structure(list(
## action = "ignore", tol = 1e-04), .Names = c("action",
## "tol")), check.conv.hess = structure(list(action = "warning",
## tol = 1e-06), .Names = c("action", "tol"))), .Names = c("check.conv.grad",
## "check.conv.singular", "check.conv.hess")), optCtrl = list()), .Names = c("optimizer",
## "calc.derivs", "use.last.params", "restart_edge", "boundary.tol",
## "tolPwrss", "compDev", "nAGQ0initStep", "checkControl", "checkConv",
## "optCtrl"), class = c("glmerControl", "merControl"))
##
## AIC BIC logLik deviance df.resid
## 1882.7 1993.1 -923.4 1846.7 3376
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -11.8794 0.0817 0.1561 0.2931 2.1647
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## subid (Intercept) 1.2167 1.1031
## trial.typeFAM-nov 0.5661 0.7524 0.18
## trial.typeNOV-fam 0.1771 0.4208 -0.99 -0.35
## Number of obs: 3394, groups: subid, 159
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.1918 0.6837 -0.280 0.77911
## age 1.1733 0.2580 4.547 5.44e-06
## trial.typeFAM-nov -3.8573 0.9593 -4.021 5.80e-05
## trial.typeNOV-fam -1.3564 0.6808 -1.992 0.04632
## conditiontablet -0.4475 0.9523 -0.470 0.63840
## age:trial.typeFAM-nov 1.3893 0.4242 3.275 0.00106
## age:trial.typeNOV-fam 0.1374 0.2758 0.498 0.61823
## age:conditiontablet 0.1326 0.3572 0.371 0.71045
## trial.typeFAM-nov:conditiontablet 3.3941 1.2377 2.742 0.00610
## trial.typeNOV-fam:conditiontablet 0.2555 0.9457 0.270 0.78700
## age:trial.typeFAM-nov:conditiontablet -1.2359 0.5141 -2.404 0.01622
## age:trial.typeNOV-fam:conditiontablet -0.3073 0.3736 -0.823 0.41079
##
## (Intercept)
## age ***
## trial.typeFAM-nov ***
## trial.typeNOV-fam *
## conditiontablet
## age:trial.typeFAM-nov **
## age:trial.typeNOV-fam
## age:conditiontablet
## trial.typeFAM-nov:conditiontablet **
## trial.typeNOV-fam:conditiontablet
## age:trial.typeFAM-nov:conditiontablet *
## age:trial.typeNOV-fam:conditiontablet
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) age tr.FAM- tr.NOV- cndtnt ag:.FAM- ag:.NOV- ag:cnd
## age -0.906
## trl.typFAM- -0.317 0.310
## trl.typNOV- -0.722 0.675 0.342
## conditntblt -0.714 0.654 0.225 0.513
## ag:trl.FAM- 0.248 -0.324 -0.931 -0.274 -0.182
## ag:trl.NOV- 0.626 -0.733 -0.310 -0.903 -0.454 0.327
## ag:cndtntbl 0.664 -0.711 -0.230 -0.497 -0.934 0.228 0.519
## trl.tyFAM-: 0.242 -0.244 -0.770 -0.261 -0.392 0.719 0.244 0.404
## trl.tyNOV-: 0.515 -0.490 -0.244 -0.716 -0.752 0.202 0.653 0.734
## ag:tr.FAM-: -0.212 0.261 0.755 0.234 0.356 -0.789 -0.264 -0.422
## ag:tr.NOV-: -0.471 0.530 0.236 0.679 0.693 -0.235 -0.724 -0.775
## t.FAM-: t.NOV-: a:.FAM-:
## age
## trl.typFAM-
## trl.typNOV-
## conditntblt
## ag:trl.FAM-
## ag:trl.NOV-
## ag:cndtntbl
## trl.tyFAM-:
## trl.tyNOV-: 0.412
## ag:tr.FAM-: -0.944 -0.377
## ag:tr.NOV-: -0.401 -0.936 0.420
## convergence code: 0
## Model failed to converge with max|grad| = 0.243542 (tol = 0.001, component 1)RT models.
rt.mod <- lmer(log(rt) ~ age * trial.type * condition +
(trial.type | subid),
data=filter(d, condition != "storybook"))
summary(rt.mod)## Linear mixed model fit by REML ['lmerMod']
## Formula: log(rt) ~ age * trial.type * condition + (trial.type | subid)
## Data: filter(d, condition != "storybook")
##
## REML criterion at convergence: 1521.2
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -3.8516 -0.6074 -0.0980 0.5213 3.8837
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## subid (Intercept) 0.0189439 0.13764
## trial.typeFAM-nov 0.0005773 0.02403 0.73
## trial.typeNOV-fam 0.0165280 0.12856 0.17 0.80
## Residual 0.1064916 0.32633
## Number of obs: 2048, groups: subid, 153
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) 7.385535 0.087705 84.21
## age -0.135068 0.026131 -5.17
## trial.typeFAM-nov 0.008596 0.107699 0.08
## trial.typeNOV-fam 0.069957 0.122063 0.57
## conditiontablet 0.323953 0.113886 2.84
## age:trial.typeFAM-nov -0.003914 0.032139 -0.12
## age:trial.typeNOV-fam -0.008950 0.036132 -0.25
## age:conditiontablet 0.030964 0.033510 0.92
## trial.typeFAM-nov:conditiontablet 0.075405 0.130686 0.58
## trial.typeNOV-fam:conditiontablet -0.133566 0.154553 -0.86
## age:trial.typeFAM-nov:conditiontablet -0.028968 0.038376 -0.75
## age:trial.typeNOV-fam:conditiontablet 0.089003 0.044948 1.98
##
## Correlation of Fixed Effects:
## (Intr) age tr.FAM- tr.NOV- cndtnt ag:.FAM- ag:.NOV- ag:cnd
## age -0.939
## trl.typFAM- -0.534 0.493
## trl.typNOV- -0.468 0.431 0.454
## conditntblt -0.770 0.723 0.411 0.361
## ag:trl.FAM- 0.493 -0.514 -0.941 -0.421 -0.380
## ag:trl.NOV- 0.435 -0.452 -0.425 -0.946 -0.335 0.444
## ag:cndtntbl 0.732 -0.780 -0.385 -0.336 -0.946 0.401 0.353
## trl.tyFAM-: 0.440 -0.406 -0.824 -0.374 -0.500 0.775 0.350 0.466
## trl.tyNOV-: 0.370 -0.340 -0.358 -0.790 -0.422 0.333 0.747 0.391
## ag:tr.FAM-: -0.413 0.430 0.788 0.353 0.467 -0.837 -0.372 -0.484
## ag:tr.NOV-: -0.350 0.364 0.341 0.760 0.397 -0.357 -0.804 -0.409
## t.FAM-: t.NOV-: a:.FAM-:
## age
## trl.typFAM-
## trl.typNOV-
## conditntblt
## ag:trl.FAM-
## ag:trl.NOV-
## ag:cndtntbl
## trl.tyFAM-:
## trl.tyNOV-: 0.448
## ag:tr.FAM-: -0.947 -0.421
## ag:tr.NOV-: -0.425 -0.952 0.443et.rt.mod <- lmer(log(rt) ~ age * trial.type +
(trial.type | subid),
data=filter(d, condition == "eye-tracker"))
summary(et.rt.mod)## Linear mixed model fit by REML ['lmerMod']
## Formula: log(rt) ~ age * trial.type + (trial.type | subid)
## Data: filter(d, condition == "eye-tracker")
##
## REML criterion at convergence: 668.8
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.3674 -0.6692 -0.1325 0.5546 3.0401
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## subid (Intercept) 0.02185 0.1478
## trial.typeFAM-nov 0.01515 0.1231 0.20
## trial.typeNOV-fam 0.03613 0.1901 -0.23 0.91
## Residual 0.17611 0.4197
## Number of obs: 524, groups: subid, 68
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) 7.384524 0.107729 68.55
## age -0.134633 0.031966 -4.21
## trial.typeFAM-nov 0.003722 0.145471 0.03
## trial.typeNOV-fam 0.044870 0.157934 0.28
## age:trial.typeFAM-nov -0.003417 0.043503 -0.08
## age:trial.typeNOV-fam -0.001021 0.046922 -0.02
##
## Correlation of Fixed Effects:
## (Intr) age t.FAM- t.NOV- a:.FAM
## age -0.939
## trl.typFAM- -0.538 0.499
## trl.typNOV- -0.571 0.532 0.526
## ag:trl.FAM- 0.495 -0.518 -0.940 -0.491
## ag:trl.NOV- 0.532 -0.561 -0.494 -0.945 0.520tab.rt.mod <- lmer(log(rt) ~ age * trial.type +
(trial.type | subid),
data=filter(d, condition == "tablet"))
summary(tab.rt.mod)## Linear mixed model fit by REML ['lmerMod']
## Formula: log(rt) ~ age * trial.type + (trial.type | subid)
## Data: filter(d, condition == "tablet")
##
## REML criterion at convergence: 731
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -4.3576 -0.6111 -0.0899 0.5615 3.3786
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## subid (Intercept) 0.0154384 0.12425
## trial.typeFAM-nov 0.0006457 0.02541 -0.01
## trial.typeNOV-fam 0.0099593 0.09980 0.42 -0.91
## Residual 0.0832721 0.28857
## Number of obs: 1524, groups: subid, 85
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) 7.70014 0.06473 118.95
## age -0.10170 0.01871 -5.44
## trial.typeFAM-nov 0.10037 0.06566 1.53
## trial.typeNOV-fam -0.05791 0.08108 -0.71
## age:trial.typeFAM-nov -0.03712 0.01861 -2.00
## age:trial.typeNOV-fam 0.07887 0.02282 3.46
##
## Correlation of Fixed Effects:
## (Intr) age t.FAM- t.NOV- a:.FAM
## age -0.957
## trl.typFAM- -0.498 0.470
## trl.typNOV- -0.280 0.260 0.335
## ag:trl.FAM- 0.479 -0.489 -0.962 -0.321
## ag:trl.NOV- 0.267 -0.265 -0.323 -0.963 0.334–> –>
–>
–> –> –> –> –> –> –> –> –> –>
–> –>
–> –> –> –> –> –>
–> –> –> –> –> –> –>
–>
–>
–> –> –>
–> –> –> –> –> –> –> –> –> –> –> –>
–> –> –> –>
–> –>
–> –> –> –> –>
–> –> –> –> –> –> –> –>
–> –> –> –> –> –> –> –>
–> –> –> –> –> –> –> –> –> –>
–> –> –> –> –> –> –> –> –> –> –> –> –>
–> –>
–> –> –> –> –> –> –> –> –> –> –> –> –> –> –> –> –> –>
–>
–> –> –> –> –> –> –> –> –> –> –> –>
–> –>
–> –> –> –> –> –> –> –> –> –> –> –> –> –>
–> –> –> –> –> –> –> –> –> –> –>