Article
Mizumoto, A., & Chujo, K. (2015, in press). A meta-analysis of data-driven learning approach in the Japanese EFL classroom.
English Corpus Studies, 22 -> Download the paper
Mizumoto, A., & Chujo, K. (2015, in press). A meta-analysis of data-driven learning approach in the Japanese EFL classroom.
English Corpus Studies, 22 -> Download the paper
dat <- read.csv("http://www.mizumot.com/files/ecs2015.csv", header=T)
dat # Showing the data
## Study Measure n hours r
## 1 Chujo & Oghigian (2007) - 1 Level 2 (category) 20 15.0 0.6
## 2 Chujo & Oghigian (2007) - 2 Level 2 (category) 20 15.0 0.6
## 3 Chujo & Oghigian (2007) - 3 Level 3 (phrase) 20 15.0 0.6
## 4 Chujo & Oghigian (2007) - 4 Proficiency 20 15.0 0.6
## 5 Chujo (2008) - 1 Level 2 (category) 75 15.0 0.6
## 6 Chujo (2008) - 2 Level 3 (phrase) 75 15.0 0.6
## 7 Chujo et al. (2008) Level 3 (phrase) 21 7.5 0.6
## 8 Chujo, Anthony, & Oghigian (2009) - 1 Level 1 (lemma) 22 15.0 0.6
## 9 Chujo, Anthony, & Oghigian (2009) - 2 Level 2 (category) 22 15.0 0.6
## 10 Chujo, Anthony, & Oghigian (2009) - 3 Level 3 (phrase) 22 15.0 0.6
## 11 Chujo, Oghigian, & Nishigaki (2012) Level 3 (phrase) 15 7.5 0.6
## 12 Chujo et al. (2012) - 1 Level 3 (phrase) 22 7.5 0.6
## 13 Chujo et al. (2012) - 2 Level 3 (phrase) 22 7.5 0.6
## 14 Chujo et al. (2012) - 3 Level 3 (phrase) 22 7.5 0.6
## 15 Chujo & Oghigian (2012) - 1 Level 1 (lemma) 25 7.5 0.6
## 16 Chujo & Oghigian (2012) - 2 Level 3 (phrase) 25 7.5 0.6
## 17 Chujo & Oghigian (2012) - 3 Proficiency 25 15.0 0.6
## 18 Chujo & Oghigian (2012) - 4 Level 1 (lemma) 14 7.5 0.6
## 19 Chujo & Oghigian (2012) - 5 Level 3 (phrase) 14 7.5 0.6
## 20 Chujo & Oghigian (2012) - 6 Proficiency 14 15.0 0.6
## 21 Chujo, Oghigian, & Uchibori (2013) - 1 Level 3 (phrase) 25 7.5 0.6
## 22 Chujo, Oghigian, & Uchibori (2013) - 2 Proficiency 25 15.0 0.6
## 23 Chujo, Oghigian, & Uchibori (2013) - 3 Level 3 (phrase) 25 7.5 0.6
## 24 Chujo, Oghigian, & Uchibori (2013) - 4 Proficiency 25 15.0 0.6
## 25 Nishigaki, Chujo, & Kijima (2010) Level 1 (lemma) 12 1.0 0.6
## 26 Nishigaki, Minegishi, & Chujo (2012) Level 2 (category) 27 1.0 0.6
## 27 Nishigaki et al. (2013) - 1 Level 2 (category) 32 1.0 0.6
## 28 Nishigaki et al. (2013) - 2 Level 2 (category) 15 1.0 0.6
## 29 Anthony et al. (2014) - 1 Level 2 (category) 62 7.5 0.6
## 30 Anthony et al. (2014) - 2 Level 2 (category) 41 7.5 0.6
## 31 Chujo et al. (2014a) Level 3 (phrase) 145 7.5 0.6
## 32 Chujo et al. (2014b) Level 3 (phrase) 14 7.5 0.6
## yi vi
## 1 0.5268 0.0469
## 2 0.9844 0.0642
## 3 0.4102 0.0442
## 4 0.6454 0.0504
## 5 0.7642 0.0146
## 6 0.7551 0.0145
## 7 1.5154 0.0928
## 8 3.5950 0.3301
## 9 0.5365 0.0429
## 10 0.8148 0.0515
## 11 0.9454 0.0831
## 12 0.8986 0.0547
## 13 0.7046 0.0476
## 14 0.6612 0.0463
## 15 2.6498 0.1724
## 16 1.0241 0.0530
## 17 0.3387 0.0343
## 18 2.1781 0.2266
## 19 0.7886 0.0794
## 20 0.5971 0.0699
## 21 1.0237 0.0530
## 22 0.3813 0.0349
## 23 0.7665 0.0437
## 24 0.2316 0.0331
## 25 3.9703 0.7235
## 26 1.1476 0.0540
## 27 0.6504 0.0316
## 28 1.0941 0.0932
## 29 0.8943 0.0194
## 30 0.9306 0.0301
## 31 1.1238 0.0099
## 32 0.6601 0.0727
# Install package (library) if not installed
usePackage <- function(p) {
if (!is.element(p, installed.packages()[,1]))
install.packages(p, dep = TRUE)
require(p, character.only = TRUE)
}
usePackage("metafor") # Load "metafor" package
RE.res <- rma(yi, vi, data=dat, slab=paste(Study))
RE.res
##
## Random-Effects Model (k = 32; tau^2 estimator: REML)
##
## tau^2 (estimated amount of total heterogeneity): 0.1666 (SE = 0.0568)
## tau (square root of estimated tau^2 value): 0.4082
## I^2 (total heterogeneity / total variability): 80.16%
## H^2 (total variability / sampling variability): 5.04
##
## Test for Heterogeneity:
## Q(df = 31) = 119.1335, p-val < .0001
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## 0.9021 0.0845 10.6724 <.0001 0.7364 1.0678 ***
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
funnel(RE.res)
title(main="All")
regtest(RE.res, model="lm") # p > .05(non-significant)indicates the funnel plot is NOT asymmetry(No publication bias).
##
## Regression Test for Funnel Plot Asymmetry
##
## model: weighted regression with multiplicative dispersion
## predictor: standard error
##
## test for funnel plot asymmetry: t = 2.3661, df = 30, p = 0.0246
par(mar=c(4,4,1,2))
forest(RE.res, xlim=c(-3, 7.5), ylim=c(-1, 55), steps=9, rows=c(50:47, 41:33, 27:14, 8:4),
order=order(dat$Measure), ilab=dat$n, ilab.xpos=-0.5,
xlab="Standarized Mean Difference (95% CI)", mlab="RE Model for All Studies")
# set font expansion factor (as in forest() above) and use bold italic font and save original settings in object 'op'
op <- par(cex=1.15, font=4)
# add text for the subgroups
text(-3, c(51.5, 42.5, 28.5, 9.5), pos=4, c("Level 1 (lemma)",
"Level 2 (category)",
"Level 3 (phrase)",
"Proficiency"))
par(font=2) # switch to bold font
# add column headings to the plot
text(-3, 54, "Study", pos=4)
text(-0.65, 54, "n", pos=4)
text(7.5, 54, "Effect Size [95%CI]", pos=2)
# set par back to the original settings
par(op)
# fit random-effects model in the subgroups
res.1 <- rma(yi, vi, data=dat, subset=(Measure=="Level 1 (lemma)"))
res.2 <- rma(yi, vi, data=dat, subset=(Measure=="Level 2 (category)"))
res.3 <- rma(yi, vi, data=dat, subset=(Measure=="Level 3 (phrase)"))
res.4 <- rma(yi, vi, data=dat, subset=(Measure=="Proficiency"))
# add summary polygons for the subgroups
addpoly(res.1, row=45.5, cex=1, mlab="RE Model for Level 1")
addpoly(res.2, row=31.5, cex=1, mlab="RE Model for Level 2")
addpoly(res.3, row=12.5, cex=1, mlab="RE Model for Level 3")
addpoly(res.4, row=2.5, cex=1, mlab="RE Model for Proficiency")
dat2 <- split(dat, dat$Measure) # Deviding the studies into groups (procedures)
# Level 1 (lemma)
RE.res1 <- rma(yi, vi, data=dat2[[1]])
RE.res1
##
## Random-Effects Model (k = 4; tau^2 estimator: REML)
##
## tau^2 (estimated amount of total heterogeneity): 0.2621 (SE = 0.4611)
## tau (square root of estimated tau^2 value): 0.5120
## I^2 (total heterogeneity / total variability): 47.21%
## H^2 (total variability / sampling variability): 1.89
##
## Test for Heterogeneity:
## Q(df = 3) = 5.6338, p-val = 0.1309
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## 2.9293 0.3766 7.7781 <.0001 2.1912 3.6674 ***
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# Level 2 (category)
RE.res2 <- rma(yi, vi, data=dat2[[2]])
RE.res2
##
## Random-Effects Model (k = 9; tau^2 estimator: REML)
##
## tau^2 (estimated amount of total heterogeneity): 0.0000 (SE = 0.0151)
## tau (square root of estimated tau^2 value): 0.0017
## I^2 (total heterogeneity / total variability): 0.01%
## H^2 (total variability / sampling variability): 1.00
##
## Test for Heterogeneity:
## Q(df = 8) = 8.7032, p-val = 0.3680
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## 0.8092 0.0604 13.4020 <.0001 0.6908 0.9275 ***
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# Level 3 (phrase)
RE.res3 <- rma(yi, vi, data=dat2[[3]])
RE.res3
##
## Random-Effects Model (k = 14; tau^2 estimator: REML)
##
## tau^2 (estimated amount of total heterogeneity): 0.0207 (SE = 0.0241)
## tau (square root of estimated tau^2 value): 0.1440
## I^2 (total heterogeneity / total variability): 34.69%
## H^2 (total variability / sampling variability): 1.53
##
## Test for Heterogeneity:
## Q(df = 13) = 20.0767, p-val = 0.0933
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## 0.8604 0.0682 12.6126 <.0001 0.7267 0.9941 ***
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# Proficiency
RE.res4 <- rma(yi, vi, data=dat2[[4]])
RE.res4
##
## Random-Effects Model (k = 5; tau^2 estimator: REML)
##
## tau^2 (estimated amount of total heterogeneity): 0 (SE = 0.0286)
## tau (square root of estimated tau^2 value): 0
## I^2 (total heterogeneity / total variability): 0.00%
## H^2 (total variability / sampling variability): 1.00
##
## Test for Heterogeneity:
## Q(df = 4) = 2.7263, p-val = 0.6046
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## 0.4023 0.0905 4.4463 <.0001 0.2250 0.5796 ***
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# Level 1 (lemma)
funnel(RE.res1)
title(main="Level 1 (lemma)")
# Level 1 (lemma)
regtest(RE.res1, model="lm")
##
## Regression Test for Funnel Plot Asymmetry
##
## model: weighted regression with multiplicative dispersion
## predictor: standard error
##
## test for funnel plot asymmetry: t = 1.7381, df = 2, p = 0.2243
# Level 2 (category)
funnel(RE.res2)
title(main="Level 2 (category)")
# Level 2 (category)
regtest(RE.res2, model="lm")
##
## Regression Test for Funnel Plot Asymmetry
##
## model: weighted regression with multiplicative dispersion
## predictor: standard error
##
## test for funnel plot asymmetry: t = 0.5704, df = 7, p = 0.5863
# Level 3 (phrase)
funnel(RE.res3)
title(main="Level 3 (phrase)")
# Level 3 (phrase)
regtest(RE.res3, model="lm")
##
## Regression Test for Funnel Plot Asymmetry
##
## model: weighted regression with multiplicative dispersion
## predictor: standard error
##
## test for funnel plot asymmetry: t = -0.8382, df = 12, p = 0.4183
# Proficiency
funnel(RE.res4)
title(main="Proficiency")
# Proficiency
regtest(RE.res4, model="lm")
##
## Regression Test for Funnel Plot Asymmetry
##
## model: weighted regression with multiplicative dispersion
## predictor: standard error
##
## test for funnel plot asymmetry: t = 2.7825, df = 3, p = 0.0688
Atsushi MIZUMOTO, Ph.D.
Associate Professor of Applied Linguistics
Faculty of Foreign Language Studies
Graduate School of Foreign Language Education and Research
Kansai University, Osaka, Japan