Syntactic Boostrapping MA - action
Data V3!!
Anjie Cao and Molly Lewis
2020-08-07
options(knitr.duplicate.label = 'allow')DATA_PATH <- here("data/processed/syntactic_bootstrapping_tidy_data.csv") # make all variables (i.e. things that might change) as capital letters at the top of the scripts
ma_data <- read_csv(DATA_PATH) %>% filter(paradigm_type == "action_matching") %>% filter(inclusion_certainty == 2)Data Overview
effect size & paper
n_effect_sizes <- ma_data %>%
filter(!is.na(d_calc)) %>%
nrow()
n_papers <- ma_data %>%
distinct(unique_id) %>%
nrow()There are 42 effect sizes collected from 12 different papers.
Here are the papers in this analysis:
ma_data %>%
count(short_cite) %>%
arrange(-n) %>%
DT::datatable()ma_model <- rma(ma_data$d_calc, ma_data$d_var_calc)
ma_model##
## Random-Effects Model (k = 42; tau^2 estimator: REML)
##
## tau^2 (estimated amount of total heterogeneity): 0.5166 (SE = 0.1435)
## tau (square root of estimated tau^2 value): 0.7187
## I^2 (total heterogeneity / total variability): 83.83%
## H^2 (total variability / sampling variability): 6.19
##
## Test for Heterogeneity:
## Q(df = 41) = 181.5424, p-val < .0001
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## 0.5965 0.1253 4.7623 <.0001 0.3510 0.8420 ***
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1forest(ma_model,
header = T,
slab = ma_data$unique_id,
col = "red",
cex = .7)
## funnel plot {.tabset} ### vanilla
ma_data %>%
mutate(color = ifelse(sentence_structure == "transitive", "red", "blue"),
color_sample_size = ifelse(n_1 < 10, "red", ifelse(n_1 < 20, "orange", "yellow")),
color_confidence = ifelse(inclusion_certainty == 1, "red", "black"))-> ma_data_funnel
ma_data_funnel %>% filter (abs(d_calc) < 5) -> ma_data_funnel_no_outlier
ss_colors <- ma_data_funnel$color
ss_colors_no_outlier <- ma_data_funnel_no_outlier$color
ma_model_funnel <- rma(ma_data_funnel$d_calc, ma_data_funnel$d_var_calc)
ma_model_funnel_no_outlier <- rma(ma_data_funnel_no_outlier$d_calc, ma_data_funnel_no_outlier$d_var_calc)
f1<- funnel(ma_model_funnel, xlab = "Effect Size", col = ss_colors)
legend("topright",bg = "white",legend = c("transitive","intransitive"),pch=16,col=c("red", "blue"))
title(main = "All effect sizes break down by sentence structure")
f2<- funnel(ma_model_funnel_no_outlier, xlab = "Effect Size", col = ss_colors_no_outlier)
legend("topright",bg = "white",legend = c("transitive","intransitive"),pch=16,col=c("red", "blue"))
title(main = "effect sizes excluded outliers (abs <5) break down by sentence structure")
take into account sentence structure?
ma_model_sentence_structure <- rma(ma_data_funnel$d_calc~ma_data_funnel$sentence_structure, ma_data_funnel$d_var_calc)
ma_model_no_outlier_ss <- rma(ma_data_funnel_no_outlier$d_calc~ma_data_funnel_no_outlier$sentence_structure, ma_data_funnel_no_outlier$d_var_calc)
f3 <- funnel(ma_model_sentence_structure, xlab = "effect size", col = ss_colors)
f3_b <- funnel(ma_model_no_outlier_ss, xlab = "effect size", col = ss_colors_no_outlier)
take into account sentence structure and age?
ma_model_ss_age <- rma(ma_data_funnel$d_calc~ma_data_funnel$sentence_structure + ma_data_funnel $mean_age, ma_data_funnel$d_var_calc)
ma_model_funnel_no_outlier_ss_age <- rma(ma_data_funnel_no_outlier$d_calc~ma_data_funnel_no_outlier$sentence_structure+ma_data_funnel_no_outlier$mean_age, ma_data_funnel_no_outlier$d_var_calc)
f4 <- funnel(ma_model_ss_age, xlab = "effect size", col = ss_colors)
f4_b <- funnel(ma_model_funnel_no_outlier_ss_age, xlab = "effect size", col = ss_colors_no_outlier)
Variable Summary
Distribution with years:
ma_data %>%
ggplot(aes(x = publication_year, y = d_calc, size = n_1)) +
xlim(1990,2020) +
geom_point() +
geom_smooth(method = "lm") +
geom_smooth(color = "red") +
ylab("Effect Size") +
xlab("publication year") +
ggtitle("Syntactical Bootstrapping effect size with year") +
theme(legend.position = "none") 
m_year <- rma.mv(d_calc ~ publication_year, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_year)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -60.8998 121.7996 127.7996 132.7124 128.5055
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4539 0.6737 11 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 176.4803, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.0623, p-val = 0.8029
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 12.7595 48.7766 0.2616 0.7936 -82.8409 108.3600
## publication_year -0.0061 0.0243 -0.2496 0.8029 -0.0536 0.0415
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Continuous Variables
CONTINUOUS_VARS <- c("n_1", "x_1", "sd_1", "d_calc", "d_var_calc", "mean_age","n_train_test_pair","n_test_trial_per_pair","productive_vocab_mean", "productive_vocab_median")
long_continuous <- ma_data %>%
pivot_longer(cols = CONTINUOUS_VARS)
long_continuous %>%
ggplot(aes(x = value)) +
geom_histogram() +
facet_wrap(~ name, scale = "free_x") +
labs(title = "Distribution of continuous measures")
long_continuous %>%
group_by(name) %>%
summarize(mean = mean(value),
sd = sd(value)) %>%
kable()| name | mean | sd |
|---|---|---|
| d_calc | 0.7084939 | 1.1050916 |
| d_var_calc | 0.1757144 | 0.2039025 |
| mean_age | 719.6813238 | 190.9659270 |
| n_1 | 12.9761905 | 4.5289750 |
| n_test_trial_per_pair | 1.9761905 | 0.6043781 |
| n_train_test_pair | 1.8333333 | 1.2671587 |
| productive_vocab_mean | NA | NA |
| productive_vocab_median | NA | NA |
| sd_1 | 0.1282059 | 0.0525234 |
| x_1 | 0.5684455 | 0.0731893 |
Categorical Variables
CATEGORICAL_VARS <- c("sentence_structure", "agent_argument_type_clean", "patient_argument_type_clean", "stimuli_actor","agent_argument_number","transitive_event_type","intransitive_event_type",
"presentation_type", "character_identification",
"test_mass_or_distributed", "practice_phase", "test_method")
long_categorical <- ma_data %>%
pivot_longer(cols = CATEGORICAL_VARS) %>%
count(name, value) # this is a short cut for group_by() %>% summarize(count = n())
long_categorical %>%
ggplot(aes(x = value, y = n)) +
facet_wrap(~ name, scale = "free_x") +
geom_col(position = 'dodge',width=0.4) +
theme(text = element_text(size=8),
axis.text.x = element_text(angle = 90, hjust = 1)) # rotate x-axis text
Prep for Moderators
ma_data_young <- ma_data_young <- ma_data %>%
mutate(age_months = mean_age/30.44) %>%
filter(age_months < 36)
ma_data_old <- ma_data %>%
mutate(age_months = mean_age/30.44) %>%
filter(age_months > 36 | age_months == 36)
ma_data_vocab <- ma_data %>%
filter(!is.na(productive_vocab_median)) no moderators
all
m1 <- rma.mv(d_calc, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m1)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -62.5424 125.0848 129.0848 132.5119 129.4006
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3591 0.5992 12 no short_cite
##
## Test for Heterogeneity:
## Q(df = 41) = 181.5424, p-val < .0001
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## 0.5517 0.1852 2.9783 0.0029 0.1886 0.9147 **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1<36
m_young <- rma.mv(d_calc, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -60.1483 120.2966 124.2966 127.6237 124.6299
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3549 0.5958 12 no short_cite
##
## Test for Heterogeneity:
## Q(df = 39) = 174.7976, p-val < .0001
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## 0.5622 0.1843 3.0499 0.0023 0.2009 0.9235 **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1>= 36
m_old <- rma.mv(d_calc, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_old)
summary(m_old)##
## Multivariate Meta-Analysis Model (k = 2; method: REML)
##
## logLik Deviance AIC BIC AICc
## 0.4489 -0.8978 1.1022 -0.8978 5.1022
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0000 0.0000 1 yes short_cite
##
## Test for Heterogeneity:
## Q(df = 1) = 0.0011, p-val = 0.9736
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## -0.0776 0.1799 -0.4312 0.6663 -0.4301 0.2750
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Participants characteristics:
age only
colored by unique_id
ma_data %>%
ggplot(aes(x = mean_age/30.44, y = d_calc,color = unique_id)) +
geom_point() +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months)") 
all
ma_data %>%
ggplot(aes(x = mean_age/30.44, y = d_calc, size = n_1)) +
geom_point() +
geom_smooth(method = "lm") +
geom_smooth(color = "red") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months)") +
theme(legend.position = "none") 
m_simple <- rma.mv(d_calc ~ 1, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
m_age <- rma.mv(d_calc ~ mean_age, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
m_age_log <- rma.mv(d_calc ~ log(mean_age), V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_simple)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -62.5424 125.0848 129.0848 132.5119 129.4006
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3591 0.5992 12 no short_cite
##
## Test for Heterogeneity:
## Q(df = 41) = 181.5424, p-val < .0001
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## 0.5517 0.1852 2.9783 0.0029 0.1886 0.9147 **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1summary(m_age)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -59.3060 118.6119 124.6119 129.6785 125.2786
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3531 0.5942 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 172.5325, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 6.9726, p-val = 0.0083
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.0613 0.2667 3.9788 <.0001 0.5385 1.5842 ***
## mean_age -0.0007 0.0003 -2.6406 0.0083 -0.0012 -0.0002 **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1summary(m_age_log)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -59.4364 118.8729 124.8729 129.9395 125.5395
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3558 0.5965 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 173.6732, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 6.5950, p-val = 0.0102
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 3.9792 1.3475 2.9530 0.0031 1.3381 6.6202 **
## log(mean_age) -0.5181 0.2017 -2.5681 0.0102 -0.9134 -0.1227 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1forest(m_age,
header = T,
slab = ma_data$unique_id,
col = "red",
cex = .7
)
funnel(m_age)
young, < 36
ma_data_young %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1)) +
geom_point() +
geom_smooth(method = "lm") +
geom_smooth(color = "red") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months)") +
theme(legend.position = "none") 
m_age_young <- rma.mv(d_calc ~ mean_age, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -59.5766 119.1532 125.1532 130.0660 125.8591
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3553 0.5961 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 172.0856, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.8198, p-val = 0.3652
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.9055 0.4216 2.1479 0.0317 0.0792 1.7318 *
## mean_age -0.0005 0.0005 -0.9054 0.3652 -0.0015 0.0005
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1forest(m_age_young,
header = T,
slab = ma_data_young$unique_id,
col = "red",
cex = .7
)
old, >= 36
ma_data_old %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1)) +
geom_point() +
geom_smooth(method = "lm") +
geom_smooth(color = "red") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months)") +
theme(legend.position = "none") 
m_age_old <- rma.mv(d_calc ~ mean_age, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_old)
summary(m_age_old)##
## Multivariate Meta-Analysis Model (k = 2; method: REML)
##
## logLik Deviance AIC BIC AICc
## 0.0000 -0.0000 4.0000 -Inf 16.0000
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0000 0.0000 1 yes short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 0) = 0.0000, p-val = 1.0000
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.0011, p-val = 0.9736
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -1.7381 50.2056 -0.0346 0.9724 -100.1393 96.6631
## mean_age 0.0013 0.0394 0.0331 0.9736 -0.0759 0.0786
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1vocab
ma_data_vocab %>%
ggplot(aes(x = productive_vocab_median, y = d_calc, size = n_1)) +
geom_point() +
geom_smooth(method = "lm") +
geom_smooth(color = "red") +
ylab("Effect Size") +
xlab("Median Vocabulary Size") +
ggtitle("Syntactical Bootstrapping effect size vs. Median Vocabulary (months)") +
theme(legend.position = "none") 
m_vocab <- rma.mv(d_calc ~ productive_vocab_median, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -46.0146 92.0292 98.0292 101.9167 99.0727
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2797 0.5289 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 27) = 102.2718, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 5.5235, p-val = 0.0188
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.1366 0.2886 3.9383 <.0001 0.5710 1.7023
## productive_vocab_median -0.0065 0.0028 -2.3502 0.0188 -0.0120 -0.0011
##
## intrcpt ***
## productive_vocab_median *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_vocab_with_age <- rma.mv(d_calc ~ mean_age, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_vocab_with_age)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.7119 91.4238 97.4238 101.3113 98.4673
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2512 0.5012 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 27) = 99.3641, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 6.5365, p-val = 0.0106
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.3171 0.3105 4.2417 <.0001 0.7085 1.9256 ***
## mean_age -0.0007 0.0003 -2.5567 0.0106 -0.0012 -0.0002 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Linguistic Stimuli
Sentence structure
SS only
m_SS <- rma.mv(d_calc ~ sentence_structure, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_SS)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.9721 115.9442 121.9442 127.0108 122.6108
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4221 0.6497 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 181.0730, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 9.4979, p-val = 0.0021
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.3695 0.2082 1.7749 0.0759 -0.0385 0.7775
## sentence_structuretransitive 0.3330 0.1081 3.0819 0.0021 0.1212 0.5448
##
## intrcpt .
## sentence_structuretransitive **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1all with age
ma_data %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = sentence_structure)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months)") 
m_age_sentence <- rma.mv(d_calc ~ mean_age + sentence_structure, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_sentence)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -56.4046 112.8092 120.8092 127.4634 121.9856
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4073 0.6382 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 39) = 172.5204, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 13.0541, p-val = 0.0015
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.7803 0.2986 2.6131 0.0090 0.1950
## mean_age -0.0005 0.0003 -1.8967 0.0579 -0.0010
## sentence_structuretransitive 0.2765 0.1119 2.4700 0.0135 0.0571
## ci.ub
## intrcpt 1.3656 **
## mean_age 0.0000 .
## sentence_structuretransitive 0.4959 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_sentence <- rma.mv(d_calc ~ mean_age * sentence_structure, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_sentence)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -55.9772 111.9545 121.9545 130.1424 123.8295
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3924 0.6264 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 168.7293, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 13.9166, p-val = 0.0030
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.7256 0.3020 2.4029 0.0163
## mean_age -0.0004 0.0003 -1.4289 0.1530
## sentence_structuretransitive 0.7463 0.5099 1.4637 0.1433
## mean_age:sentence_structuretransitive -0.0007 0.0007 -0.9464 0.3439
## ci.lb ci.ub
## intrcpt 0.1338 1.3175 *
## mean_age -0.0010 0.0002
## sentence_structuretransitive -0.2530 1.7456
## mean_age:sentence_structuretransitive -0.0021 0.0007
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1< 36, young, age
ma_data_young %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, , color = sentence_structure)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), with sentence_structure") 
m_age_sentence_young <- rma.mv(d_calc ~ mean_age + sentence_structure, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_sentence_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -56.6513 113.3026 121.3026 127.7463 122.5526
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4108 0.6409 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 37) = 172.0851, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 6.9855, p-val = 0.0304
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.6009 0.4466 1.3455 0.1785 -0.2744
## mean_age -0.0003 0.0005 -0.5001 0.6170 -0.0013
## sentence_structuretransitive 0.2783 0.1120 2.4851 0.0130 0.0588
## ci.ub
## intrcpt 1.4762
## mean_age 0.0008
## sentence_structuretransitive 0.4978 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_sentence_young <- rma.mv(d_calc ~ mean_age * sentence_structure, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_sentence_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -56.2507 112.5014 122.5014 130.4190 124.5014
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3900 0.6245 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 164.7154, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 8.0723, p-val = 0.0445
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.4720 0.4602 1.0255 0.3051
## mean_age -0.0001 0.0006 -0.1089 0.9133
## sentence_structuretransitive 0.8173 0.5190 1.5750 0.1153
## mean_age:sentence_structuretransitive -0.0008 0.0007 -1.0664 0.2862
## ci.lb ci.ub
## intrcpt -0.4300 1.3740
## mean_age -0.0011 0.0010
## sentence_structuretransitive -0.1998 1.8345
## mean_age:sentence_structuretransitive -0.0022 0.0007
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1>= 36, old, age
ma_data_old %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = sentence_structure)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months)") 
all with vocab
ma_data_vocab %>%
ggplot(aes(x = productive_vocab_median, y = d_calc, size = n_1, color = sentence_structure)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Median Productive Vocab") +
ggtitle("Syntactical Bootstrapping effect size vs. Median Productive Vocab")
m_age_sentence_vocab <- rma.mv(d_calc ~ productive_vocab_median + sentence_structure, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_sentence_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -37.3877 74.7754 82.7754 87.8078 84.6802
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.1370 0.3701 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 79.0411, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 23.2106, p-val < .0001
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.6178 0.2527 2.4454 0.0145 0.1226
## productive_vocab_median -0.0021 0.0029 -0.7262 0.4677 -0.0079
## sentence_structuretransitive 0.6266 0.1482 4.2277 <.0001 0.3361
## ci.ub
## intrcpt 1.1130 *
## productive_vocab_median 0.0036
## sentence_structuretransitive 0.9171 ***
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_sentence_vocab_interaction <- rma.mv(d_calc ~ productive_vocab_median * sentence_structure, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_sentence_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -37.3877 74.7754 82.7754 87.8078 84.6802
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.1370 0.3701 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 79.0411, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 23.2106, p-val < .0001
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.6178 0.2527 2.4454 0.0145 0.1226
## productive_vocab_median -0.0021 0.0029 -0.7262 0.4677 -0.0079
## sentence_structuretransitive 0.6266 0.1482 4.2277 <.0001 0.3361
## ci.ub
## intrcpt 1.1130 *
## productive_vocab_median 0.0036
## sentence_structuretransitive 0.9171 ***
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Agent Arguement Type
m_AAT <- rma.mv(d_calc ~ agent_argument_type_clean, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
m_AAT_young <- rma.mv(d_calc ~ agent_argument_type_clean, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_AAT)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -59.7306 119.4613 129.4613 137.6492 131.3363
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2981 0.5460 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 161.5785, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 3.9182, p-val = 0.2704
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.2662 0.2290 1.1624 0.2451
## agent_argument_type_cleannoun_phrase 0.2212 0.1779 1.2434 0.2137
## agent_argument_type_cleanpronoun 0.5085 0.4069 1.2497 0.2114
## agent_argument_type_cleanvarying_agent 0.5913 0.3511 1.6840 0.0922
## ci.lb ci.ub
## intrcpt -0.1827 0.7152
## agent_argument_type_cleannoun_phrase -0.1275 0.5699
## agent_argument_type_cleanpronoun -0.2890 1.3059
## agent_argument_type_cleanvarying_agent -0.0969 1.2795 .
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1summary(m_AAT_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.2187 114.4374 124.4374 132.3549 126.4374
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2791 0.5283 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 148.3998, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 4.3429, p-val = 0.2267
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.2626 0.2230 1.1779 0.2388
## agent_argument_type_cleannoun_phrase 0.2218 0.1778 1.2477 0.2122
## agent_argument_type_cleanpronoun 0.5340 0.3986 1.3396 0.1804
## agent_argument_type_cleanvarying_agent 0.6211 0.3418 1.8172 0.0692
## ci.lb ci.ub
## intrcpt -0.1744 0.6997
## agent_argument_type_cleannoun_phrase -0.1266 0.5702
## agent_argument_type_cleanpronoun -0.2473 1.3152
## agent_argument_type_cleanvarying_agent -0.0488 1.2911 .
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1all
ma_data %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = agent_argument_type_clean)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by agent argument") 
m_age_aa <- rma.mv(d_calc ~ mean_age + agent_argument_type_clean, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_aa)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -56.6817 113.3634 125.3634 135.0289 128.1634
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3013 0.5489 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 37) = 155.8375, p-val < .0001
##
## Test of Moderators (coefficients 2:5):
## QM(df = 4) = 10.5496, p-val = 0.0321
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.7767 0.3033 2.5609 0.0104
## mean_age -0.0007 0.0003 -2.5800 0.0099
## agent_argument_type_cleannoun_phrase 0.2166 0.1780 1.2169 0.2236
## agent_argument_type_cleanpronoun 0.5155 0.4083 1.2626 0.2067
## agent_argument_type_cleanvarying_agent 0.5627 0.3529 1.5947 0.1108
## ci.lb ci.ub
## intrcpt 0.1823 1.3711 *
## mean_age -0.0012 -0.0002 **
## agent_argument_type_cleannoun_phrase -0.1322 0.5653
## agent_argument_type_cleanpronoun -0.2848 1.3158
## agent_argument_type_cleanvarying_agent -0.1289 1.2544
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_aa_interaction <- rma.mv(d_calc ~ mean_age * agent_argument_type_clean, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_aa_interaction)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -55.2818 110.5636 128.5636 142.3009 136.0636
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3222 0.5676 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 34) = 153.8321, p-val < .0001
##
## Test of Moderators (coefficients 2:8):
## QM(df = 7) = 12.3265, p-val = 0.0903
##
## Model Results:
##
## estimate se zval
## intrcpt 2.7599 1.9046 1.4491
## mean_age -0.0032 0.0024 -1.3190
## agent_argument_type_cleannoun_phrase 2.3830 3.6608 0.6509
## agent_argument_type_cleanpronoun -4.7721 12.0320 -0.3966
## agent_argument_type_cleanvarying_agent -1.4421 1.9335 -0.7459
## mean_age:agent_argument_type_cleannoun_phrase -0.0027 0.0045 -0.5986
## mean_age:agent_argument_type_cleanpronoun 0.0077 0.0187 0.4126
## mean_age:agent_argument_type_cleanvarying_agent 0.0026 0.0024 1.0565
## pval ci.lb ci.ub
## intrcpt 0.1473 -0.9730 6.4928
## mean_age 0.1872 -0.0080 0.0016
## agent_argument_type_cleannoun_phrase 0.5151 -4.7920 9.5579
## agent_argument_type_cleanpronoun 0.6917 -28.3544 18.8102
## agent_argument_type_cleanvarying_agent 0.4557 -5.2318 2.3475
## mean_age:agent_argument_type_cleannoun_phrase 0.5494 -0.0114 0.0061
## mean_age:agent_argument_type_cleanpronoun 0.6799 -0.0289 0.0443
## mean_age:agent_argument_type_cleanvarying_agent 0.2907 -0.0022 0.0074
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1<36, young
ma_data_young %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = agent_argument_type_clean)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by agent argument, young only") 
m_age_aa_young <- rma.mv(d_calc ~ mean_age + agent_argument_type_clean, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_aa_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -56.9120 113.8241 125.8241 135.1562 128.8241
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2921 0.5405 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 35) = 147.6122, p-val < .0001
##
## Test of Moderators (coefficients 2:5):
## QM(df = 4) = 4.6937, p-val = 0.3202
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.5383 0.4574 1.1766 0.2393
## mean_age -0.0004 0.0005 -0.6878 0.4916
## agent_argument_type_cleannoun_phrase 0.2190 0.1779 1.2311 0.2183
## agent_argument_type_cleanpronoun 0.5256 0.4045 1.2992 0.1939
## agent_argument_type_cleanvarying_agent 0.5915 0.3507 1.6866 0.0917
## ci.lb ci.ub
## intrcpt -0.3583 1.4348
## mean_age -0.0014 0.0007
## agent_argument_type_cleannoun_phrase -0.1297 0.5677
## agent_argument_type_cleanpronoun -0.2673 1.3185
## agent_argument_type_cleanvarying_agent -0.0959 1.2789 .
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_aa_young_interaction <- rma.mv(d_calc ~ mean_age * agent_argument_type_clean, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_aa_young_interaction)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -55.3365 110.6731 128.6731 141.8647 136.8549
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2960 0.5440 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 32) = 141.8028, p-val < .0001
##
## Test of Moderators (coefficients 2:8):
## QM(df = 7) = 7.0231, p-val = 0.4265
##
## Model Results:
##
## estimate se zval
## intrcpt 2.7205 1.8492 1.4712
## mean_age -0.0032 0.0024 -1.3393
## agent_argument_type_cleannoun_phrase 2.3820 3.6480 0.6529
## agent_argument_type_cleanpronoun -4.7815 12.0232 -0.3977
## agent_argument_type_cleanvarying_agent -1.7002 1.9036 -0.8931
## mean_age:agent_argument_type_cleannoun_phrase -0.0027 0.0044 -0.5995
## mean_age:agent_argument_type_cleanpronoun 0.0077 0.0187 0.4103
## mean_age:agent_argument_type_cleanvarying_agent 0.0030 0.0024 1.2272
## pval ci.lb ci.ub
## intrcpt 0.1412 -0.9039 6.3450
## mean_age 0.1805 -0.0078 0.0015
## agent_argument_type_cleannoun_phrase 0.5138 -4.7681 9.5320
## agent_argument_type_cleanpronoun 0.6909 -28.3466 18.7836
## agent_argument_type_cleanvarying_agent 0.3718 -5.4312 2.0309
## mean_age:agent_argument_type_cleannoun_phrase 0.5488 -0.0114 0.0060
## mean_age:agent_argument_type_cleanpronoun 0.6816 -0.0289 0.0442
## mean_age:agent_argument_type_cleanvarying_agent 0.2197 -0.0018 0.0077
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1>=36, old
ma_data_old %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = agent_argument_type_clean)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by agent argument, old only") 
Vocab
ma_data_vocab %>%
ggplot(aes(x = productive_vocab_median, y = d_calc, size = n_1, color = agent_argument_type_clean)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by agent argument, with vocab") 
m_age_aa_vocab <- rma.mv(d_calc ~ productive_vocab_median + agent_argument_type_clean, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_aa_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -43.1779 86.3557 96.3557 102.4501 99.5136
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.1152 0.3395 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 90.5335, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 9.9595, p-val = 0.0189
##
## Model Results:
##
## estimate se zval pval
## intrcpt 2.1576 0.5579 3.8677 0.0001
## productive_vocab_median -0.0060 0.0028 -2.1324 0.0330
## agent_argument_type_cleanpronoun -1.2519 0.6221 -2.0125 0.0442
## agent_argument_type_cleanvarying_agent -1.2755 0.5890 -2.1655 0.0303
## ci.lb ci.ub
## intrcpt 1.0642 3.2510 ***
## productive_vocab_median -0.0115 -0.0005 *
## agent_argument_type_cleanpronoun -2.4712 -0.0327 *
## agent_argument_type_cleanvarying_agent -2.4299 -0.1211 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_aa_vocab_interaction <- rma.mv(d_calc ~ productive_vocab_median * agent_argument_type_clean, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_aa_vocab_interaction)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -42.3340 84.6679 98.6679 106.6164 106.1346
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.1186 0.3444 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 23) = 89.7991, p-val < .0001
##
## Test of Moderators (coefficients 2:6):
## QM(df = 5) = 10.6887, p-val = 0.0579
##
## Model Results:
##
## estimate
## intrcpt -11.4787
## productive_vocab_median 0.4712
## agent_argument_type_cleanpronoun 12.2133
## agent_argument_type_cleanvarying_agent 12.3730
## productive_vocab_median:agent_argument_type_cleanpronoun -0.4723
## productive_vocab_median:agent_argument_type_cleanvarying_agent -0.4774
## se
## intrcpt 17.0503
## productive_vocab_median 0.5963
## agent_argument_type_cleanpronoun 17.0594
## agent_argument_type_cleanvarying_agent 17.0519
## productive_vocab_median:agent_argument_type_cleanpronoun 0.5964
## productive_vocab_median:agent_argument_type_cleanvarying_agent 0.5963
## zval pval
## intrcpt -0.6732 0.5008
## productive_vocab_median 0.7902 0.4294
## agent_argument_type_cleanpronoun 0.7159 0.4740
## agent_argument_type_cleanvarying_agent 0.7256 0.4681
## productive_vocab_median:agent_argument_type_cleanpronoun -0.7919 0.4284
## productive_vocab_median:agent_argument_type_cleanvarying_agent -0.8006 0.4233
## ci.lb
## intrcpt -44.8967
## productive_vocab_median -0.6975
## agent_argument_type_cleanpronoun -21.2224
## agent_argument_type_cleanvarying_agent -21.0482
## productive_vocab_median:agent_argument_type_cleanpronoun -1.6413
## productive_vocab_median:agent_argument_type_cleanvarying_agent -1.6461
## ci.ub
## intrcpt 21.9393
## productive_vocab_median 1.6399
## agent_argument_type_cleanpronoun 45.6490
## agent_argument_type_cleanvarying_agent 45.7941
## productive_vocab_median:agent_argument_type_cleanpronoun 0.6967
## productive_vocab_median:agent_argument_type_cleanvarying_agent 0.6913
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Patient Argument Type
m_PAT <- rma.mv(d_calc ~ patient_argument_type_clean, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_PAT)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -52.2215 104.4431 114.4431 122.6310 116.3181
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3252 0.5702 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 147.9291, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 19.8585, p-val = 0.0002
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.3808 0.1878 2.0277 0.0426
## patient_argument_type_cleannoun 0.2191 0.1137 1.9265 0.0540
## patient_argument_type_cleanpronoun 0.7752 0.2952 2.6256 0.0086
## patient_argument_type_cleanvarying_patient 1.8280 0.5648 3.2369 0.0012
## ci.lb ci.ub
## intrcpt 0.0127 0.7488 *
## patient_argument_type_cleannoun -0.0038 0.4421 .
## patient_argument_type_cleanpronoun 0.1965 1.3539 **
## patient_argument_type_cleanvarying_patient 0.7211 2.9350 **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1ma_data %>% group_by(patient_argument_type_clean) %>% count()## # A tibble: 4 x 2
## # Groups: patient_argument_type_clean [4]
## patient_argument_type_clean n
## <chr> <int>
## 1 intransitive 20
## 2 noun 18
## 3 pronoun 3
## 4 varying_patient 1Agent Argument Numbers
m_AAN <- rma.mv(d_calc ~ agent_argument_number, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_AAN)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -61.1652 122.3304 130.3304 136.9846 131.5068
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3180 0.5639 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 39) = 168.5915, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 2.2764, p-val = 0.3204
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.4325 0.1939 2.2305 0.0257 0.0525 0.8126
## agent_argument_number2 0.1949 0.1768 1.1023 0.2703 -0.1516 0.5415
## agent_argument_numbervarying 0.2268 0.2054 1.1040 0.2696 -0.1759 0.6295
##
## intrcpt *
## agent_argument_number2
## agent_argument_numbervarying
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1ma_data %>% group_by(agent_argument_number) %>% count()## # A tibble: 3 x 2
## # Groups: agent_argument_number [3]
## agent_argument_number n
## <chr> <int>
## 1 1 15
## 2 2 4
## 3 varying 23Number of sentence reptition
ma_data %>% ggplot(aes(x = n_repetitions_sentence)) +
geom_histogram() 
### all
ma_data %>%
ggplot(aes(x = n_repetitions_sentence, y = d_calc, size = n_1)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("number of reptition in sentence") +
ggtitle("Syntactical Bootstrapping effect size vs. number of sentence repetition") 
m_rep <- rma.mv(d_calc ~ n_repetitions_sentence, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_rep)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -60.9292 121.8584 127.8584 132.9250 128.5251
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3711 0.6092 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 179.6973, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 2.3875, p-val = 0.1223
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.3343 0.2352 1.4213 0.1552 -0.1267 0.7954
## n_repetitions_sentence 0.0226 0.0147 1.5452 0.1223 -0.0061 0.0514
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_rep_age <- rma.mv(d_calc ~ n_repetitions_sentence + mean_age, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_rep_age)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.7184 117.4369 125.4369 132.0911 126.6134
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3624 0.6020 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 39) = 171.9199, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 7.1728, p-val = 0.0277
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.9500 0.3657 2.5976 0.0094 0.2332 1.6669 **
## n_repetitions_sentence 0.0073 0.0162 0.4514 0.6517 -0.0244 0.0390
## mean_age -0.0006 0.0003 -2.1874 0.0287 -0.0012 -0.0001 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_rep_age_interaction <- rma.mv(d_calc ~ n_repetitions_sentence * mean_age, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_rep_age_interaction)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.3567 116.7135 126.7135 134.9014 128.5885
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3812 0.6174 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 171.8211, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 7.1642, p-val = 0.0668
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.8792 1.0816 0.8128 0.4163 -1.2407
## n_repetitions_sentence 0.0129 0.0802 0.1612 0.8719 -0.1443
## mean_age -0.0005 0.0015 -0.3450 0.7301 -0.0034
## n_repetitions_sentence:mean_age -0.0000 0.0001 -0.0724 0.9423 -0.0002
## ci.ub
## intrcpt 2.9991
## n_repetitions_sentence 0.1701
## mean_age 0.0024
## n_repetitions_sentence:mean_age 0.0002
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1< 36, young only
ma_data_young %>%
ggplot(aes(x = n_repetitions_sentence, y = d_calc, size = n_1)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("number of reptition in sentence") +
ggtitle("Syntactical Bootstrapping effect size vs. number of sentence repetition, young only") 
m_rep_young <- rma.mv(d_calc ~ n_repetitions_sentence, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_rep_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -59.3245 118.6490 124.6490 129.5617 125.3548
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3601 0.6001 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 172.9840, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.7193, p-val = 0.3964
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.4363 0.2380 1.8331 0.0668 -0.0302 0.9028 .
## n_repetitions_sentence 0.0130 0.0153 0.8481 0.3964 -0.0170 0.0430
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_rep_age_young <- rma.mv(d_calc ~ n_repetitions_sentence + mean_age, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_rep_age_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.9468 117.8936 125.8936 132.3373 127.1436
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3644 0.6037 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 37) = 171.2214, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 1.1158, p-val = 0.5724
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.7340 0.5278 1.3905 0.1644 -0.3006 1.7685
## n_repetitions_sentence 0.0090 0.0166 0.5457 0.5853 -0.0234 0.0415
## mean_age -0.0003 0.0005 -0.6316 0.5277 -0.0014 0.0007
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_rep_age_interaction_young <- rma.mv(d_calc ~ n_repetitions_sentence * mean_age, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_rep_age_interaction_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.4216 116.8433 126.8433 134.7609 128.8433
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3903 0.6248 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 170.8514, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 1.1968, p-val = 0.7538
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.3234 1.4349 0.2254 0.8217 -2.4890
## n_repetitions_sentence 0.0378 0.0942 0.4010 0.6884 -0.1468
## mean_age 0.0002 0.0020 0.1217 0.9031 -0.0036
## n_repetitions_sentence:mean_age -0.0000 0.0001 -0.3107 0.7561 -0.0003
## ci.ub
## intrcpt 3.1358
## n_repetitions_sentence 0.2224
## mean_age 0.0041
## n_repetitions_sentence:mean_age 0.0002
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1>= 36, old only
ma_data_old %>%
ggplot(aes(x = n_repetitions_sentence, y = d_calc, size = n_1)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("number of reptition in sentence") +
ggtitle("Syntactical Bootstrapping effect size vs. number of sentence repetition, old only") 
m_rep_old <- rma.mv(d_calc ~ n_repetitions_sentence, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_old)
summary(m_rep_old)##
## Multivariate Meta-Analysis Model (k = 2; method: REML)
##
## logLik Deviance AIC BIC AICc
## 0.0000 -0.0000 4.0000 -Inf 16.0000
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0000 0.0000 1 yes short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 0) = 0.0000, p-val = 1.0000
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.0011, p-val = 0.9736
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -0.1429 1.9821 -0.0721 0.9425 -4.0277 3.7420
## n_repetitions_sentence 0.0060 0.1800 0.0331 0.9736 -0.3468 0.3587
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_rep_age_old <- rma.mv(d_calc ~ n_repetitions_sentence + mean_age, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_old)
summary(m_rep_age_old)##
## Multivariate Meta-Analysis Model (k = 2; method: REML)
##
## logLik Deviance AIC BIC AICc
## 0.0000 -0.0000 4.0000 -Inf 16.0000
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0000 0.0000 1 yes short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 0) = 0.0000, p-val = 1.0000
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.0011, p-val = 0.9736
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -0.1429 1.9821 -0.0721 0.9425 -4.0277 3.7420
## n_repetitions_sentence 0.0060 0.1800 0.0331 0.9736 -0.3468 0.3587
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_rep_age_interaction_old <- rma.mv(d_calc ~ n_repetitions_sentence * mean_age, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_old)
summary(m_rep_age_interaction_old)##
## Multivariate Meta-Analysis Model (k = 2; method: REML)
##
## logLik Deviance AIC BIC AICc
## 0.0000 -0.0000 4.0000 -Inf 16.0000
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0000 0.0000 1 yes short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 0) = 0.0000, p-val = 1.0000
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.0011, p-val = 0.9736
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -0.1429 1.9821 -0.0721 0.9425 -4.0277 3.7420
## n_repetitions_sentence 0.0060 0.1800 0.0331 0.9736 -0.3468 0.3587
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1add vocab in
m_rep_v_r <- rma.mv(d_calc ~ n_repetitions_sentence + productive_vocab_median, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_rep_v_r)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.7865 91.5731 99.5731 104.6055 101.4778
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2843 0.5332 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 102.0079, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 5.5373, p-val = 0.0627
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.1021 0.4466 2.4679 0.0136 0.2268 1.9774 *
## n_repetitions_sentence 0.0022 0.0210 0.1055 0.9160 -0.0390 0.0435
## productive_vocab_median -0.0063 0.0034 -1.8549 0.0636 -0.0130 0.0004 .
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_rep_v_r_interaction <- rma.mv(d_calc ~ n_repetitions_sentence *productive_vocab_median, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_rep_v_r)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.7865 91.5731 99.5731 104.6055 101.4778
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2843 0.5332 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 102.0079, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 5.5373, p-val = 0.0627
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.1021 0.4466 2.4679 0.0136 0.2268 1.9774 *
## n_repetitions_sentence 0.0022 0.0210 0.1055 0.9160 -0.0390 0.0435
## productive_vocab_median -0.0063 0.0034 -1.8549 0.0636 -0.0130 0.0004 .
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Visual stimuli
stimuli_modality(video vs animation)
single moderator
m_SM <- rma.mv(d_calc ~ stimuli_modality, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_SM)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -59.9625 119.9249 125.9249 130.9916 126.5916
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3818 0.6179 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 180.6334, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 4.9187, p-val = 0.0266
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.9246 0.2533 3.6505 0.0003 0.4282 1.4211 ***
## stimuli_modalityvideo -0.3949 0.1781 -2.2178 0.0266 -0.7439 -0.0459 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1all
ma_data %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = stimuli_modality)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by stimuli modality") 
m_stim_mod<- rma.mv(d_calc ~ mean_age + stimuli_modality, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_stim_mod)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.0271 116.0543 124.0543 130.7085 125.2308
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3636 0.6030 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 39) = 171.2727, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 7.5577, p-val = 0.0228
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.0946 0.2717 4.0289 <.0001 0.5621 1.6271 ***
## mean_age -0.0012 0.0007 -1.6308 0.1029 -0.0026 0.0002
## stimuli_modalityvideo 0.3960 0.5160 0.7676 0.4427 -0.6152 1.4073
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_stim_mod_interaction<- rma.mv(d_calc ~ mean_age * stimuli_modality, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_stim_mod_interaction)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.2279 116.4559 126.4559 134.6438 128.3309
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3674 0.6062 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 171.2075, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 7.7419, p-val = 0.0517
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.6461 1.0735 0.6019 0.5473 -1.4578
## mean_age -0.0004 0.0020 -0.1999 0.8416 -0.0043
## stimuli_modalityvideo 0.9469 1.3743 0.6890 0.4908 -1.7468
## mean_age:stimuli_modalityvideo -0.0009 0.0022 -0.4323 0.6655 -0.0051
## ci.ub
## intrcpt 2.7500
## mean_age 0.0035
## stimuli_modalityvideo 3.6405
## mean_age:stimuli_modalityvideo 0.0033
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1< 36, young only
ma_data_young %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = stimuli_modality)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by stimuli modality, young") 
m_stim_mod_young<- rma.mv(d_calc ~ mean_age + stimuli_modality, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_stim_mod_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.3680 116.7359 124.7359 131.1796 125.9859
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3723 0.6102 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 37) = 171.2599, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 1.1870, p-val = 0.5524
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.0534 0.4873 2.1619 0.0306 0.0984 2.0085 *
## mean_age -0.0011 0.0012 -0.9393 0.3476 -0.0034 0.0012
## stimuli_modalityvideo 0.3655 0.6004 0.6087 0.5427 -0.8113 1.5422
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_stim_mod_interaction_young <- rma.mv(d_calc ~ mean_age * stimuli_modality, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_stim_mod_interaction_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.4338 116.8676 126.8676 134.7852 128.8676
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3879 0.6228 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 171.2019, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 1.3752, p-val = 0.7114
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.6281 1.0834 0.5798 0.5621 -1.4952
## mean_age -0.0004 0.0020 -0.1999 0.8416 -0.0043
## stimuli_modalityvideo 1.0987 1.7623 0.6234 0.5330 -2.3555
## mean_age:stimuli_modalityvideo -0.0011 0.0025 -0.4424 0.6582 -0.0060
## ci.ub
## intrcpt 2.7515
## mean_age 0.0035
## stimuli_modalityvideo 4.5528
## mean_age:stimuli_modalityvideo 0.0038
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1>= 36, old only
ma_data_old %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = stimuli_modality)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by stimuli modality, old") 
vocab
ma_data_vocab %>%
ggplot(aes(x = productive_vocab_median, y = d_calc, size = n_1, color = stimuli_modality)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("median vocab") +
ggtitle("Syntactical Bootstrapping effect size vs. median vocab, breakdown by stimuli modality, vocab") 
m_stim_mod_v <- rma.mv(d_calc ~ productive_vocab_median + stimuli_modality, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_stim_mod_v)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.3388 90.6776 98.6776 103.7100 100.5824
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3082 0.5551 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 99.1928, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 5.5391, p-val = 0.0627
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.1471 0.3151 3.6410 0.0003 0.5296 1.7646
## productive_vocab_median -0.0064 0.0059 -1.0972 0.2726 -0.0179 0.0050
## stimuli_modalityvideo -0.0103 0.3735 -0.0277 0.9779 -0.7423 0.7216
##
## intrcpt ***
## productive_vocab_median
## stimuli_modalityvideo
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_stim_mod_intereaction_v <- rma.mv(d_calc ~ productive_vocab_median * stimuli_modality, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_stim_mod_intereaction_v)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.3710 90.7419 100.7419 106.8363 103.8998
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3024 0.5499 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 98.6450, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 6.0046, p-val = 0.1114
##
## Model Results:
##
## estimate se zval
## intrcpt 0.6842 0.7426 0.9214
## productive_vocab_median 0.0164 0.0339 0.4842
## stimuli_modalityvideo 0.4879 0.8152 0.5985
## productive_vocab_median:stimuli_modalityvideo -0.0236 0.0344 -0.6845
## pval ci.lb ci.ub
## intrcpt 0.3569 -0.7712 2.1396
## productive_vocab_median 0.6282 -0.0500 0.0829
## stimuli_modalityvideo 0.5495 -1.1098 2.0855
## productive_vocab_median:stimuli_modalityvideo 0.4936 -0.0910 0.0439
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1vocab age comparison
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = stimuli_modality)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by stimuli modality, age compare vocab") 
m_stim_mod_v_a<- rma.mv(d_calc ~ mean_age + stimuli_modality, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_stim_mod_v_a)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -44.2751 88.5502 96.5502 101.5826 98.4550
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2122 0.4606 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 93.9088, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 7.6570, p-val = 0.0217
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.3622 0.3021 4.5088 <.0001 0.7701 1.9544 ***
## mean_age -0.0013 0.0007 -1.9192 0.0550 -0.0027 0.0000 .
## stimuli_modalityvideo 0.5099 0.4828 1.0561 0.2909 -0.4364 1.4563
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_stim_mod_interaction_v_a <- rma.mv(d_calc ~ mean_age * stimuli_modality, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_stim_mod_interaction_v_a)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -44.4528 88.9056 98.9056 105.0000 102.0635
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2145 0.4631 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 93.7830, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 7.8993, p-val = 0.0481
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.8481 1.0716 0.7914 0.4287 -1.2522
## mean_age -0.0004 0.0020 -0.1999 0.8416 -0.0043
## stimuli_modalityvideo 1.1301 1.3309 0.8492 0.3958 -1.4784
## mean_age:stimuli_modalityvideo -0.0011 0.0021 -0.5009 0.6165 -0.0053
## ci.ub
## intrcpt 2.9484
## mean_age 0.0035
## stimuli_modalityvideo 3.7387
## mean_age:stimuli_modalityvideo 0.0031
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1stimuli_actor (person vs non-person)
single moderator
m_SA <- rma.mv(d_calc ~ stimuli_actor, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_SA)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -60.9592 121.9184 127.9184 132.9851 128.5851
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4623 0.6799 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 181.4468, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 2.8628, p-val = 0.0906
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.7302 0.2303 3.1708 0.0015 0.2788 1.1815 **
## stimuli_actorperson -0.2834 0.1675 -1.6920 0.0906 -0.6117 0.0449 .
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1all
ma_data %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = stimuli_actor)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by stimuli actor, all") 
m_stim_actor_age <- rma.mv(d_calc ~ mean_age + stimuli_actor, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_stim_actor_age)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.4445 114.8891 122.8891 129.5433 124.0655
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3351 0.5788 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 39) = 169.9287, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 9.0313, p-val = 0.0109
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.2485 0.2950 4.2318 <.0001 0.6703 1.8267 ***
## mean_age -0.0013 0.0005 -2.5906 0.0096 -0.0022 -0.0003 **
## stimuli_actorperson 0.4541 0.3170 1.4323 0.1521 -0.1673 1.0754
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_stim_actor_age_interaction <- rma.mv(d_calc ~ mean_age * stimuli_actor, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_stim_actor_age_interaction)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -56.5707 113.1415 123.1415 131.3294 125.0165
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3394 0.5826 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 168.4545, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 9.1801, p-val = 0.0270
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.9240 0.8553 1.0804 0.2800 -0.7523
## mean_age -0.0008 0.0012 -0.6736 0.5006 -0.0032
## stimuli_actorperson 0.9494 1.2623 0.7521 0.4520 -1.5247
## mean_age:stimuli_actorperson -0.0007 0.0017 -0.4052 0.6853 -0.0040
## ci.ub
## intrcpt 2.6004
## mean_age 0.0016
## stimuli_actorperson 3.4235
## mean_age:stimuli_actorperson 0.0026
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1< 36, young only
ma_data_young %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = stimuli_actor)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by stimuli actor, young") 
m_stim_actor_age_young <- rma.mv(d_calc ~ mean_age + stimuli_actor, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_stim_actor_age_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.7929 115.5857 123.5857 130.0294 124.8357
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3405 0.5835 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 37) = 169.5477, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 2.6417, p-val = 0.2669
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.2074 0.4760 2.5363 0.0112 0.2744 2.1404 *
## mean_age -0.0012 0.0008 -1.6078 0.1079 -0.0027 0.0003
## stimuli_actorperson 0.4451 0.3301 1.3486 0.1775 -0.2018 1.0920
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_stim_actor_age_interaction_young <- rma.mv(d_calc ~ mean_age * stimuli_actor, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_stim_actor_age_interaction_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -56.7727 113.5454 123.5454 131.4630 125.5454
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3527 0.5939 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 168.3855, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 2.7939, p-val = 0.4245
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.8793 0.8953 0.9822 0.3260 -0.8754
## mean_age -0.0008 0.0013 -0.6151 0.5385 -0.0032
## stimuli_actorperson 1.2194 1.7964 0.6788 0.4973 -2.3015
## mean_age:stimuli_actorperson -0.0010 0.0023 -0.4380 0.6614 -0.0055
## ci.ub
## intrcpt 2.6341
## mean_age 0.0017
## stimuli_actorperson 4.7404
## mean_age:stimuli_actorperson 0.0035
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1>= 36, old only
ma_data_old %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = stimuli_actor)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by stimuli actor, old") 
vocab
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = productive_vocab_median, y = d_calc, size = n_1, color = stimuli_actor)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by stimuli actor, vocab") 
m_stim_actor_vocab <- rma.mv(d_calc ~ productive_vocab_median + stimuli_actor, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_stim_actor_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.3388 90.6776 98.6776 103.7100 100.5824
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3082 0.5551 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 99.1928, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 5.5391, p-val = 0.0627
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.1471 0.3151 3.6410 0.0003 0.5296 1.7646
## productive_vocab_median -0.0064 0.0059 -1.0972 0.2726 -0.0179 0.0050
## stimuli_actorperson -0.0103 0.3735 -0.0277 0.9779 -0.7423 0.7216
##
## intrcpt ***
## productive_vocab_median
## stimuli_actorperson
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_stim_actor_vocab_interaction <- rma.mv(d_calc ~ productive_vocab_median * stimuli_actor, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_stim_actor_vocab_interaction)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.3710 90.7419 100.7419 106.8363 103.8998
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3024 0.5499 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 98.6450, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 6.0046, p-val = 0.1114
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.6842 0.7426 0.9214 0.3569
## productive_vocab_median 0.0164 0.0339 0.4842 0.6282
## stimuli_actorperson 0.4879 0.8152 0.5985 0.5495
## productive_vocab_median:stimuli_actorperson -0.0236 0.0344 -0.6845 0.4936
## ci.lb ci.ub
## intrcpt -0.7712 2.1396
## productive_vocab_median -0.0500 0.0829
## stimuli_actorperson -1.1098 2.0855
## productive_vocab_median:stimuli_actorperson -0.0910 0.0439
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1vocab age comparison
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = stimuli_actor)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by stimuli actor, age vocab comparison") 
m_stim_actor_vocab_a <- rma.mv(d_calc ~ mean_age + stimuli_actor, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_stim_actor_vocab_a)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -44.2751 88.5502 96.5502 101.5826 98.4550
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2122 0.4606 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 93.9088, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 7.6570, p-val = 0.0217
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.3622 0.3021 4.5088 <.0001 0.7701 1.9544 ***
## mean_age -0.0013 0.0007 -1.9192 0.0550 -0.0027 0.0000 .
## stimuli_actorperson 0.5099 0.4828 1.0561 0.2909 -0.4364 1.4563
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_stim_actor_vocab_interaction_a <- rma.mv(d_calc ~ mean_age * stimuli_actor, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_stim_actor_vocab_interaction_a)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -44.4528 88.9056 98.9056 105.0000 102.0635
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2145 0.4631 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 93.7830, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 7.8993, p-val = 0.0481
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.8481 1.0716 0.7914 0.4287 -1.2522
## mean_age -0.0004 0.0020 -0.1999 0.8416 -0.0043
## stimuli_actorperson 1.1301 1.3309 0.8492 0.3958 -1.4784
## mean_age:stimuli_actorperson -0.0011 0.0021 -0.5009 0.6165 -0.0053
## ci.ub
## intrcpt 2.9484
## mean_age 0.0035
## stimuli_actorperson 3.7387
## mean_age:stimuli_actorperson 0.0031
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1transitive_event (indirect_caused_action vs direct caused action)
single moderator
m_TE <- rma.mv(d_calc ~ transitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_TE)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -61.3636 122.7272 130.7272 137.3814 131.9037
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4145 0.6438 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 39) = 178.9380, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 0.3690, p-val = 0.8315
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.5958 0.2082 2.8612 0.0042
## transitive_event_typeindirect_caused_action -0.0644 0.1756 -0.3668 0.7138
## transitive_event_typeminimal_contact -0.3471 0.6981 -0.4972 0.6191
## ci.lb ci.ub
## intrcpt 0.1877 1.0040 **
## transitive_event_typeindirect_caused_action -0.4085 0.2797
## transitive_event_typeminimal_contact -1.7154 1.0212
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1ma_data %>% count(transitive_event_type)## # A tibble: 3 x 2
## transitive_event_type n
## <chr> <int>
## 1 direct_caused_action 31
## 2 indirect_caused_action 9
## 3 minimal_contact 2all
ma_data %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = transitive_event_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by transitive_event_type") 
m_age_vs_tran <- rma.mv(d_calc ~ mean_age + transitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_vs_tran)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.9531 115.9063 125.9063 134.0942 127.7813
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3816 0.6177 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 169.6187, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 7.5822, p-val = 0.0555
##
## Model Results:
##
## estimate se zval pval
## intrcpt 1.1189 0.2807 3.9859 <.0001
## mean_age -0.0007 0.0003 -2.6864 0.0072
## transitive_event_typeindirect_caused_action 0.1248 0.1882 0.6632 0.5072
## transitive_event_typeminimal_contact -0.2626 0.6725 -0.3904 0.6962
## ci.lb ci.ub
## intrcpt 0.5687 1.6691 ***
## mean_age -0.0013 -0.0002 **
## transitive_event_typeindirect_caused_action -0.2441 0.4937
## transitive_event_typeminimal_contact -1.5807 1.0556
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_data_transtivity_interaction <- rma.mv(d_calc ~ mean_age * transitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_data_transtivity_interaction)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.3674 114.7349 128.7349 139.8195 132.7349
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3664 0.6053 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 165.4356, p-val < .0001
##
## Test of Moderators (coefficients 2:6):
## QM(df = 5) = 10.1522, p-val = 0.0710
##
## Model Results:
##
## estimate se zval
## intrcpt 0.9886 0.3066 3.2239
## mean_age -0.0006 0.0003 -1.8512
## transitive_event_typeindirect_caused_action 0.5856 0.5440 1.0766
## transitive_event_typeminimal_contact 6.0856 4.7943 1.2693
## mean_age:transitive_event_typeindirect_caused_action -0.0006 0.0006 -0.9079
## mean_age:transitive_event_typeminimal_contact -0.0077 0.0057 -1.3360
## pval ci.lb ci.ub
## intrcpt 0.0013 0.3876 1.5896
## mean_age 0.0641 -0.0012 0.0000
## transitive_event_typeindirect_caused_action 0.2817 -0.4806 1.6518
## transitive_event_typeminimal_contact 0.2043 -3.3110 15.4822
## mean_age:transitive_event_typeindirect_caused_action 0.3639 -0.0018 0.0006
## mean_age:transitive_event_typeminimal_contact 0.1816 -0.0189 0.0036
##
## intrcpt **
## mean_age .
## transitive_event_typeindirect_caused_action
## transitive_event_typeminimal_contact
## mean_age:transitive_event_typeindirect_caused_action
## mean_age:transitive_event_typeminimal_contact
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1young < 36
ma_data_young %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = transitive_event_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by transitive_event_type") 
m_age_vs_tran_young <- rma.mv(d_calc ~ mean_age + transitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_vs_tran_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.7902 115.5805 125.5805 133.4981 127.5805
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3661 0.6050 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 166.7696, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 2.0833, p-val = 0.5553
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.8762 0.4265 2.0543 0.0399
## mean_age -0.0004 0.0005 -0.8325 0.4052
## transitive_event_typeindirect_caused_action 0.2378 0.2318 1.0257 0.3050
## transitive_event_typeminimal_contact -0.2757 0.6609 -0.4171 0.6766
## ci.lb ci.ub
## intrcpt 0.0402 1.7121 *
## mean_age -0.0014 0.0006
## transitive_event_typeindirect_caused_action -0.2166 0.6922
## transitive_event_typeminimal_contact -1.5710 1.0197
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_data_transtivity_interaction_young <- rma.mv(d_calc ~ mean_age * transitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_data_transtivity_interaction_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.1752 114.3503 128.3503 139.0348 132.6580
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3726 0.6104 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 34) = 164.7456, p-val < .0001
##
## Test of Moderators (coefficients 2:6):
## QM(df = 5) = 4.3865, p-val = 0.4952
##
## Model Results:
##
## estimate se zval
## intrcpt 0.8075 0.4302 1.8771
## mean_age -0.0003 0.0005 -0.6345
## transitive_event_typeindirect_caused_action 2.1225 2.8480 0.7453
## transitive_event_typeminimal_contact 6.2666 4.8044 1.3044
## mean_age:transitive_event_typeindirect_caused_action -0.0031 0.0046 -0.6640
## mean_age:transitive_event_typeminimal_contact -0.0079 0.0058 -1.3764
## pval ci.lb ci.ub
## intrcpt 0.0605 -0.0357 1.6507
## mean_age 0.5258 -0.0013 0.0007
## transitive_event_typeindirect_caused_action 0.4561 -3.4595 7.7046
## transitive_event_typeminimal_contact 0.1921 -3.1498 15.6831
## mean_age:transitive_event_typeindirect_caused_action 0.5067 -0.0121 0.0060
## mean_age:transitive_event_typeminimal_contact 0.1687 -0.0192 0.0034
##
## intrcpt .
## mean_age
## transitive_event_typeindirect_caused_action
## transitive_event_typeminimal_contact
## mean_age:transitive_event_typeindirect_caused_action
## mean_age:transitive_event_typeminimal_contact
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1old > 36
ma_data_old %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = transitive_event_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by transitive_event_type") 
m_age_vs_tran_old <- rma.mv(d_calc ~ mean_age + transitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_old)
summary(m_age_vs_tran_old)##
## Multivariate Meta-Analysis Model (k = 2; method: REML)
##
## logLik Deviance AIC BIC AICc
## 0.0000 -0.0000 4.0000 -Inf 16.0000
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0000 0.0000 1 yes short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 0) = 0.0000, p-val = 1.0000
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.0011, p-val = 0.9736
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -1.7381 50.2056 -0.0346 0.9724 -100.1393 96.6631
## mean_age 0.0013 0.0394 0.0331 0.9736 -0.0759 0.0786
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_data_transtivity_interaction_old <- rma.mv(d_calc ~ mean_age * transitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_old)
summary(m_data_transtivity_interaction_old)##
## Multivariate Meta-Analysis Model (k = 2; method: REML)
##
## logLik Deviance AIC BIC AICc
## 0.0000 -0.0000 4.0000 -Inf 16.0000
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0000 0.0000 1 yes short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 0) = 0.0000, p-val = 1.0000
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.0011, p-val = 0.9736
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -1.7381 50.2056 -0.0346 0.9724 -100.1393 96.6631
## mean_age 0.0013 0.0394 0.0331 0.9736 -0.0759 0.0786
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1vocab
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = productive_vocab_median, y = d_calc, size = n_1, color = transitive_event_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("median vocab") +
ggtitle("Syntactical Bootstrapping effect size vs. median vocab, breakdown by transitive_event_type") 
m_age_vs_tran_vocab <- rma.mv(d_calc ~ productive_vocab_median + transitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_vs_tran_old)##
## Multivariate Meta-Analysis Model (k = 2; method: REML)
##
## logLik Deviance AIC BIC AICc
## 0.0000 -0.0000 4.0000 -Inf 16.0000
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0000 0.0000 1 yes short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 0) = 0.0000, p-val = 1.0000
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.0011, p-val = 0.9736
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -1.7381 50.2056 -0.0346 0.9724 -100.1393 96.6631
## mean_age 0.0013 0.0394 0.0331 0.9736 -0.0759 0.0786
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_data_transtivity_interaction_vocab <- rma.mv(d_calc ~ productive_vocab_median * transitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_data_transtivity_interaction_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.4496 90.8992 100.8992 106.9936 104.0571
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2477 0.4977 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 97.4381, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 6.5079, p-val = 0.0894
##
## Model Results:
##
## estimate
## intrcpt 1.0327
## productive_vocab_median -0.0056
## transitive_event_typeindirect_caused_action 0.3452
## productive_vocab_median:transitive_event_typeindirect_caused_action -0.0049
## se
## intrcpt 0.2982
## productive_vocab_median 0.0037
## transitive_event_typeindirect_caused_action 0.3445
## productive_vocab_median:transitive_event_typeindirect_caused_action 0.0061
## zval
## intrcpt 3.4638
## productive_vocab_median -1.5030
## transitive_event_typeindirect_caused_action 1.0021
## productive_vocab_median:transitive_event_typeindirect_caused_action -0.8035
## pval
## intrcpt 0.0005
## productive_vocab_median 0.1328
## transitive_event_typeindirect_caused_action 0.3163
## productive_vocab_median:transitive_event_typeindirect_caused_action 0.4217
## ci.lb
## intrcpt 0.4484
## productive_vocab_median -0.0129
## transitive_event_typeindirect_caused_action -0.3300
## productive_vocab_median:transitive_event_typeindirect_caused_action -0.0168
## ci.ub
## intrcpt 1.6171
## productive_vocab_median 0.0017
## transitive_event_typeindirect_caused_action 1.0204
## productive_vocab_median:transitive_event_typeindirect_caused_action 0.0071
##
## intrcpt ***
## productive_vocab_median
## transitive_event_typeindirect_caused_action
## productive_vocab_median:transitive_event_typeindirect_caused_action
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1vocab age comparsion
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = mean_age, y = d_calc, size = n_1, color = transitive_event_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by transitive_event_type") 
m_age_vs_tran_vocab <- rma.mv(d_calc ~ mean_age + transitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_vs_tran_old)##
## Multivariate Meta-Analysis Model (k = 2; method: REML)
##
## logLik Deviance AIC BIC AICc
## 0.0000 -0.0000 4.0000 -Inf 16.0000
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0000 0.0000 1 yes short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 0) = 0.0000, p-val = 1.0000
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.0011, p-val = 0.9736
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -1.7381 50.2056 -0.0346 0.9724 -100.1393 96.6631
## mean_age 0.0013 0.0394 0.0331 0.9736 -0.0759 0.0786
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_data_transtivity_interaction_vocab <- rma.mv(d_calc ~ mean_age * transitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_data_transtivity_interaction_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.2636 90.5271 100.5271 106.6215 103.6850
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2203 0.4694 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 95.7234, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 7.5997, p-val = 0.0551
##
## Model Results:
##
## estimate se zval
## intrcpt 1.1911 0.3302 3.6070
## mean_age -0.0006 0.0003 -1.8151
## transitive_event_typeindirect_caused_action 0.5459 0.5400 1.0109
## mean_age:transitive_event_typeindirect_caused_action -0.0005 0.0006 -0.8806
## pval ci.lb ci.ub
## intrcpt 0.0003 0.5439 1.8384
## mean_age 0.0695 -0.0012 0.0000
## transitive_event_typeindirect_caused_action 0.3120 -0.5124 1.6042
## mean_age:transitive_event_typeindirect_caused_action 0.3785 -0.0017 0.0007
##
## intrcpt ***
## mean_age .
## transitive_event_typeindirect_caused_action
## mean_age:transitive_event_typeindirect_caused_action
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1intransitive_event (one vs parallel)
single moderator
m_IE <- rma.mv(d_calc ~ intransitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_IE)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -62.3608 124.7215 130.7215 135.7882 131.3882
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3583 0.5986 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 179.3311, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.1596, p-val = 0.6895
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.6004 0.2218 2.7072 0.0068
## intransitive_event_typeparallel_actions -0.0578 0.1446 -0.3995 0.6895
## ci.lb ci.ub
## intrcpt 0.1657 1.0351 **
## intransitive_event_typeparallel_actions -0.3411 0.2256
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1ma_data %>% count(intransitive_event_type) ## # A tibble: 2 x 2
## intransitive_event_type n
## <chr> <int>
## 1 one_action 14
## 2 parallel_actions 28all
ma_data %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = intransitive_event_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by intransitive_event_type") 
m_age_vs_intran <- rma.mv(d_calc ~ mean_age + intransitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_vs_intran)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.9379 117.8757 125.8757 132.5300 127.0522
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3623 0.6019 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 39) = 172.4752, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 7.4195, p-val = 0.0245
##
## Model Results:
##
## estimate se zval pval
## intrcpt 1.0275 0.2732 3.7616 0.0002
## mean_age -0.0007 0.0003 -2.6945 0.0070
## intransitive_event_typeparallel_actions 0.1052 0.1567 0.6711 0.5021
## ci.lb ci.ub
## intrcpt 0.4921 1.5629 ***
## mean_age -0.0013 -0.0002 **
## intransitive_event_typeparallel_actions -0.2020 0.4124
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_data_intranstivity_interaction <- rma.mv(d_calc ~ mean_age * intransitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_data_intranstivity_interaction)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.0204 116.0409 126.0409 134.2288 127.9159
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3525 0.5937 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 167.3711, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 8.0617, p-val = 0.0448
##
## Model Results:
##
## estimate se zval
## intrcpt 0.4281 0.7948 0.5386
## mean_age 0.0003 0.0014 0.2366
## intransitive_event_typeparallel_actions 0.7376 0.8052 0.9161
## mean_age:intransitive_event_typeparallel_actions -0.0011 0.0014 -0.8014
## pval ci.lb ci.ub
## intrcpt 0.5901 -1.1297 1.9859
## mean_age 0.8130 -0.0023 0.0030
## intransitive_event_typeparallel_actions 0.3596 -0.8405 2.3158
## mean_age:intransitive_event_typeparallel_actions 0.4229 -0.0039 0.0017
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1young < 36
ma_data_young %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = intransitive_event_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by intransitive_event_type, young") 
m_age_vs_intran_young <- rma.mv(d_calc ~ mean_age + intransitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_vs_intran_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -59.2379 118.4758 126.4758 132.9194 127.7258
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3641 0.6034 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 37) = 171.9146, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 1.1888, p-val = 0.5519
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.9037 0.4228 2.1373 0.0326
## mean_age -0.0006 0.0005 -1.0518 0.2929
## intransitive_event_typeparallel_actions 0.0965 0.1584 0.6089 0.5426
## ci.lb ci.ub
## intrcpt 0.0750 1.7323 *
## mean_age -0.0016 0.0005
## intransitive_event_typeparallel_actions -0.2141 0.4070
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_data_intranstivity_interaction_young <- rma.mv(d_calc ~ mean_age * intransitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_data_intranstivity_interaction_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.4789 116.9579 126.9579 134.8755 128.9579
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3568 0.5974 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 167.3659, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 1.7133, p-val = 0.6340
##
## Model Results:
##
## estimate se zval
## intrcpt 0.4024 0.8091 0.4973
## mean_age 0.0003 0.0014 0.2525
## intransitive_event_typeparallel_actions 0.6920 0.8372 0.8266
## mean_age:intransitive_event_typeparallel_actions -0.0011 0.0015 -0.7249
## pval ci.lb ci.ub
## intrcpt 0.6190 -1.1834 1.9882
## mean_age 0.8007 -0.0023 0.0030
## intransitive_event_typeparallel_actions 0.4084 -0.9488 2.3328
## mean_age:intransitive_event_typeparallel_actions 0.4685 -0.0040 0.0018
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1old > 36
ma_data_old %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = intransitive_event_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by intransitive_event_type, old") 
vocab
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = productive_vocab_median, y = d_calc, size = n_1, color = intransitive_event_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("median vocab") +
ggtitle("Syntactical Bootstrapping effect size vs. median vocab, breakdown by intransitive_event_type") 
m_age_vs_intran_vocab <- rma.mv(d_calc ~ productive_vocab_median + intransitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_vs_intran_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.9339 91.8677 99.8677 104.9001 101.7725
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2819 0.5309 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 102.2686, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 5.6787, p-val = 0.0585
##
## Model Results:
##
## estimate se zval pval
## intrcpt 1.1123 0.2963 3.7540 0.0002
## productive_vocab_median -0.0068 0.0029 -2.3773 0.0174
## intransitive_event_typeparallel_actions 0.0588 0.1498 0.3923 0.6948
## ci.lb ci.ub
## intrcpt 0.5316 1.6930 ***
## productive_vocab_median -0.0124 -0.0012 *
## intransitive_event_typeparallel_actions -0.2349 0.3524
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_data_intranstivity_interaction_vocab <- rma.mv(d_calc ~ productive_vocab_median * intransitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_data_intranstivity_interaction_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -44.7974 89.5949 99.5949 105.6893 102.7528
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2463 0.4963 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 94.0010, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 7.3754, p-val = 0.0608
##
## Model Results:
##
## estimate
## intrcpt 0.7826
## productive_vocab_median 0.0021
## intransitive_event_typeparallel_actions 0.3646
## productive_vocab_median:intransitive_event_typeparallel_actions -0.0109
## se
## intrcpt 0.3731
## productive_vocab_median 0.0074
## intransitive_event_typeparallel_actions 0.2768
## productive_vocab_median:intransitive_event_typeparallel_actions 0.0083
## zval
## intrcpt 2.0974
## productive_vocab_median 0.2860
## intransitive_event_typeparallel_actions 1.3171
## productive_vocab_median:intransitive_event_typeparallel_actions -1.3106
## pval
## intrcpt 0.0360
## productive_vocab_median 0.7749
## intransitive_event_typeparallel_actions 0.1878
## productive_vocab_median:intransitive_event_typeparallel_actions 0.1900
## ci.lb
## intrcpt 0.0513
## productive_vocab_median -0.0123
## intransitive_event_typeparallel_actions -0.1780
## productive_vocab_median:intransitive_event_typeparallel_actions -0.0272
## ci.ub
## intrcpt 1.5138 *
## productive_vocab_median 0.0165
## intransitive_event_typeparallel_actions 0.9072
## productive_vocab_median:intransitive_event_typeparallel_actions 0.0054
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1vocab age comparsion
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = mean_age, y = d_calc, size = n_1, color = intransitive_event_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by intransitive_event_type") 
m_age_vs_intran_vocab <- rma.mv(d_calc ~ mean_age + intransitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_vs_intran_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.2784 90.5567 98.5567 103.5891 100.4615
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2430 0.4929 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 98.8755, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 7.3141, p-val = 0.0258
##
## Model Results:
##
## estimate se zval pval
## intrcpt 1.2952 0.3085 4.1982 <.0001
## mean_age -0.0007 0.0003 -2.7003 0.0069
## intransitive_event_typeparallel_actions 0.1384 0.1570 0.8812 0.3782
## ci.lb ci.ub
## intrcpt 0.6905 1.8999 ***
## mean_age -0.0013 -0.0002 **
## intransitive_event_typeparallel_actions -0.1694 0.4462
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_data_intranstivity_interaction_vocab <- rma.mv(d_calc ~ mean_age * intransitive_event_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_data_intranstivity_interaction_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -44.5744 89.1489 99.1489 105.2433 102.3068
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2503 0.5003 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 93.3137, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 7.5866, p-val = 0.0554
##
## Model Results:
##
## estimate se zval
## intrcpt 0.8844 0.8446 1.0471
## mean_age -0.0001 0.0013 -0.0474
## intransitive_event_typeparallel_actions 0.5467 0.7921 0.6901
## mean_age:intransitive_event_typeparallel_actions -0.0007 0.0014 -0.5262
## pval ci.lb ci.ub
## intrcpt 0.2950 -0.7709 2.5396
## mean_age 0.9622 -0.0027 0.0025
## intransitive_event_typeparallel_actions 0.4901 -1.0059 2.0992
## mean_age:intransitive_event_typeparallel_actions 0.5988 -0.0035 0.0020
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Visual Stimuli Pair
m_VSP <- rma.mv(d_calc ~ visual_stimuli_pair, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_VSP)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -61.1199 122.2397 134.2397 143.9052 137.0397
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4086 0.6392 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 37) = 176.2333, p-val < .0001
##
## Test of Moderators (coefficients 2:5):
## QM(df = 4) = 0.4912, p-val = 0.9744
##
## Model Results:
##
## estimate se
## intrcpt 0.6383 0.2454
## visual_stimuli_pairdirect_caused_action_parallel_actions -0.0572 0.1743
## visual_stimuli_pairindirect_caused_action_one_action -0.1105 0.3282
## visual_stimuli_pairindirect_caused_action_parallel_actions -0.0793 0.1941
## visual_stimuli_pairminimal_contact_parallel_actions -0.3895 0.7059
## zval pval
## intrcpt 2.6011 0.0093
## visual_stimuli_pairdirect_caused_action_parallel_actions -0.3280 0.7429
## visual_stimuli_pairindirect_caused_action_one_action -0.3367 0.7363
## visual_stimuli_pairindirect_caused_action_parallel_actions -0.4084 0.6830
## visual_stimuli_pairminimal_contact_parallel_actions -0.5519 0.5810
## ci.lb ci.ub
## intrcpt 0.1573 1.1192 **
## visual_stimuli_pairdirect_caused_action_parallel_actions -0.3988 0.2844
## visual_stimuli_pairindirect_caused_action_one_action -0.7537 0.5327
## visual_stimuli_pairindirect_caused_action_parallel_actions -0.4597 0.3011
## visual_stimuli_pairminimal_contact_parallel_actions -1.7730 0.9939
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1ma_data %>% count()## # A tibble: 1 x 1
## n
## <int>
## 1 42n_repetition_video (??)
single moderator
m_RV <- rma.mv(d_calc ~ n_repetitions_video, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_RV)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -61.7841 123.5683 129.5683 134.6349 130.2350
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3524 0.5936 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 180.4276, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.3773, p-val = 0.5390
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.7031 0.3084 2.2801 0.0226 0.0987 1.3074 *
## n_repetitions_video -0.0445 0.0724 -0.6143 0.5390 -0.1863 0.0974
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1ma_data %>% ggplot(aes(x = n_repetitions_video)) +
geom_histogram() 
Testing Procedure
test_method (point vs look)
single moderator
m_TM <- rma.mv(d_calc ~ test_method, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_TM)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -60.1775 120.3549 126.3549 131.4215 127.0216
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2610 0.5108 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 157.2610, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 3.6764, p-val = 0.0552
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.4472 0.1682 2.6582 0.0079 0.1175 0.7769 **
## test_methodpoint 1.1379 0.5935 1.9174 0.0552 -0.0253 2.3010 .
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1ma_data %>% count(test_method)## # A tibble: 2 x 2
## test_method n
## <chr> <int>
## 1 look 40
## 2 point 2presentation_type
single moderator
m_PT <- rma.mv(d_calc ~ presentation_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_PT)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -60.7455 121.4909 129.4909 136.1452 130.6674
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3708 0.6089 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 39) = 177.8636, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 1.3235, p-val = 0.5159
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.6694 0.2168 3.0874 0.0020 0.2445
## presentation_typeimmediate_after -0.0732 0.2118 -0.3456 0.7296 -0.4884
## presentation_typesimultaneous -0.5711 0.5039 -1.1334 0.2570 -1.5586
## ci.ub
## intrcpt 1.0944 **
## presentation_typeimmediate_after 0.3419
## presentation_typesimultaneous 0.4165
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1ma_data %>% count(presentation_type)## # A tibble: 3 x 2
## presentation_type n
## <chr> <int>
## 1 asynchronous 28
## 2 immediate_after 11
## 3 simultaneous 3all
ma_data %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = presentation_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by presentation type") 
m_age_pt <- rma.mv(d_calc ~ mean_age + presentation_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_pt)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.5458 115.0916 125.0916 133.2795 126.9666
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3624 0.6020 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 169.6358, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 8.2545, p-val = 0.0410
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 1.1687 0.2870 4.0720 <.0001 0.6062
## mean_age -0.0007 0.0003 -2.6291 0.0086 -0.0012
## presentation_typeimmediate_after -0.0435 0.2117 -0.2055 0.8372 -0.4585
## presentation_typesimultaneous -0.5648 0.4988 -1.1324 0.2574 -1.5424
## ci.ub
## intrcpt 1.7311 ***
## mean_age -0.0002 **
## presentation_typeimmediate_after 0.3714
## presentation_typesimultaneous 0.4128
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_pt_interaction <- rma.mv(d_calc ~ mean_age * presentation_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_pt_interaction)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -55.5329 111.0658 125.0658 136.1504 129.0658
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4239 0.6511 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 164.5657, p-val < .0001
##
## Test of Moderators (coefficients 2:6):
## QM(df = 5) = 9.6221, p-val = 0.0867
##
## Model Results:
##
## estimate se zval pval
## intrcpt 1.2567 0.3057 4.1104 <.0001
## mean_age -0.0007 0.0003 -2.6184 0.0088
## presentation_typeimmediate_after 2.1246 1.9965 1.0642 0.2873
## presentation_typesimultaneous -2.3283 2.9907 -0.7785 0.4363
## mean_age:presentation_typeimmediate_after -0.0033 0.0030 -1.0927 0.2745
## mean_age:presentation_typesimultaneous 0.0022 0.0039 0.5691 0.5693
## ci.lb ci.ub
## intrcpt 0.6575 1.8559 ***
## mean_age -0.0012 -0.0002 **
## presentation_typeimmediate_after -1.7885 6.0378
## presentation_typesimultaneous -8.1899 3.5334
## mean_age:presentation_typeimmediate_after -0.0092 0.0026
## mean_age:presentation_typesimultaneous -0.0054 0.0099
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1<36, young
ma_data_young %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = presentation_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by presentation type, young only") 
m_data_pt_young <- rma.mv(d_calc ~ mean_age + presentation_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_data_pt_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.8145 115.6289 125.6289 133.5465 127.6289
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3641 0.6034 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 168.9109, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 2.1418, p-val = 0.5435
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 1.0018 0.4308 2.3251 0.0201 0.1573
## mean_age -0.0004 0.0005 -0.8535 0.3934 -0.0014
## presentation_typeimmediate_after -0.0556 0.2131 -0.2608 0.7943 -0.4731
## presentation_typesimultaneous -0.5729 0.5001 -1.1456 0.2520 -1.5530
## ci.ub
## intrcpt 1.8462 *
## mean_age 0.0006
## presentation_typeimmediate_after 0.3620
## presentation_typesimultaneous 0.4072
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_pt_interaction_young <- rma.mv(d_calc ~ mean_age * presentation_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_pt_interaction_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -55.8108 111.6216 125.6216 136.3061 129.9293
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4287 0.6548 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 34) = 164.0161, p-val < .0001
##
## Test of Moderators (coefficients 2:6):
## QM(df = 5) = 3.5293, p-val = 0.6190
##
## Model Results:
##
## estimate se zval pval
## intrcpt 1.0844 0.4453 2.4350 0.0149
## mean_age -0.0004 0.0005 -0.8137 0.4158
## presentation_typeimmediate_after 2.1965 2.0062 1.0948 0.2736
## presentation_typesimultaneous -2.1559 3.0230 -0.7132 0.4757
## mean_age:presentation_typeimmediate_after -0.0034 0.0030 -1.1290 0.2589
## mean_age:presentation_typesimultaneous 0.0020 0.0039 0.5005 0.6167
## ci.lb ci.ub
## intrcpt 0.2115 1.9572 *
## mean_age -0.0014 0.0006
## presentation_typeimmediate_after -1.7356 6.1287
## presentation_typesimultaneous -8.0810 3.7691
## mean_age:presentation_typeimmediate_after -0.0093 0.0025
## mean_age:presentation_typesimultaneous -0.0058 0.0097
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1>= 36, old
ma_data_old %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = presentation_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by presentation type, young only") 
vocab
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = productive_vocab_median, y = d_calc, size = n_1, color = presentation_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("median vocab") +
ggtitle("Syntactical Bootstrapping effect size vs. median vocab, breakdown by presentation type") 
m_pt_vocab <- rma.mv(d_calc ~ productive_vocab_median + presentation_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_pt_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.6960 91.3921 99.3921 104.4244 101.2968
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2835 0.5325 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 101.3938, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 5.5859, p-val = 0.0612
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 1.1307 0.2914 3.8805 0.0001 0.5596
## productive_vocab_median -0.0067 0.0028 -2.3534 0.0186 -0.0122
## presentation_typeimmediate_after 0.0568 0.2315 0.2453 0.8062 -0.3969
## ci.ub
## intrcpt 1.7018 ***
## productive_vocab_median -0.0011 *
## presentation_typeimmediate_after 0.5104
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_pt_vocab_interaction <- rma.mv(d_calc ~ productive_vocab_median * presentation_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_pt_vocab_interaction)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.6015 91.2030 101.2030 107.2973 104.3608
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2955 0.5436 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 101.3912, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 5.7505, p-val = 0.1244
##
## Model Results:
##
## estimate se
## intrcpt 1.1477 0.2976
## productive_vocab_median -0.0070 0.0029
## presentation_typeimmediate_after -0.1292 0.5193
## productive_vocab_median:presentation_typeimmediate_after 0.0053 0.0132
## zval pval
## intrcpt 3.8568 0.0001
## productive_vocab_median -2.3826 0.0172
## presentation_typeimmediate_after -0.2488 0.8035
## productive_vocab_median:presentation_typeimmediate_after 0.3992 0.6898
## ci.lb ci.ub
## intrcpt 0.5645 1.7310 ***
## productive_vocab_median -0.0127 -0.0012 *
## presentation_typeimmediate_after -1.1469 0.8886
## productive_vocab_median:presentation_typeimmediate_after -0.0206 0.0312
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1vocab age comparsion
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = mean_age, y = d_calc, size = n_1, color = presentation_type)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("mean age ") +
ggtitle("Syntactical Bootstrapping effect size vs. mean age, breakdown by presentation type") 
m_pt_vocab_a <- rma.mv(d_calc ~ mean_age + presentation_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_pt_vocab_a)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.4452 90.8903 98.8903 103.9227 100.7951
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2591 0.5091 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 98.8716, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 6.5411, p-val = 0.0380
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 1.3225 0.3159 4.1869 <.0001 0.7034
## mean_age -0.0007 0.0003 -2.5492 0.0108 -0.0012
## presentation_typeimmediate_after -0.0171 0.2268 -0.0755 0.9398 -0.4617
## ci.ub
## intrcpt 1.9416 ***
## mean_age -0.0002 *
## presentation_typeimmediate_after 0.4275
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_pt_vocab_a_interaction <- rma.mv(d_calc ~ mean_age * presentation_type, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_pt_vocab_a_interaction)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.3403 90.6805 100.6805 106.7749 103.8384
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2689 0.5186 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 98.4911, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 6.5525, p-val = 0.0876
##
## Model Results:
##
## estimate se zval pval
## intrcpt 1.3279 0.3199 4.1512 <.0001
## mean_age -0.0007 0.0003 -2.5507 0.0108
## presentation_typeimmediate_after -1.2504 11.1334 -0.1123 0.9106
## mean_age:presentation_typeimmediate_after 0.0019 0.0174 0.1107 0.9118
## ci.lb ci.ub
## intrcpt 0.7010 1.9549 ***
## mean_age -0.0012 -0.0002 *
## presentation_typeimmediate_after -23.0715 20.5707
## mean_age:presentation_typeimmediate_after -0.0321 0.0360
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1character_identification
single moderator
m_CI <- rma.mv(d_calc ~ character_identification, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_CI)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -61.4733 122.9467 128.9467 134.0133 129.6133
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3999 0.6324 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 181.0760, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.2315, p-val = 0.6304
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.4910 0.2362 2.0788 0.0376 0.0281 0.9539
## character_identificationyes 0.1998 0.4153 0.4811 0.6304 -0.6142 1.0139
##
## intrcpt *
## character_identificationyes
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1ma_data %>% count(character_identification)## # A tibble: 2 x 2
## character_identification n
## <chr> <int>
## 1 no 34
## 2 yes 8all
ma_data %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = character_identification)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by character_identification") 
m_age_ci <- rma.mv(d_calc ~ mean_age + character_identification, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_ci)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.2226 116.4451 124.4451 131.0994 125.6216
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3898 0.6243 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 39) = 171.0139, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 7.2531, p-val = 0.0266
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.9951 0.3013 3.3025 0.0010 0.4045
## mean_age -0.0007 0.0003 -2.6493 0.0081 -0.0012
## character_identificationyes 0.2228 0.4108 0.5423 0.5876 -0.5823
## ci.ub
## intrcpt 1.5857 ***
## mean_age -0.0002 **
## character_identificationyes 1.0279
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_stimuli_ci_interaction <- rma.mv(d_calc ~ mean_age * character_identification, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_stimuli_ci_interaction)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -55.2000 110.4001 120.4001 128.5880 122.2751
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3124 0.5589 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 160.5083, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 12.2078, p-val = 0.0067
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.9553 0.2849 3.3531 0.0008
## mean_age -0.0006 0.0003 -2.4627 0.0138
## character_identificationyes 5.4901 2.4152 2.2731 0.0230
## mean_age:character_identificationyes -0.0067 0.0030 -2.2110 0.0270
## ci.lb ci.ub
## intrcpt 0.3969 1.5137 ***
## mean_age -0.0011 -0.0001 *
## character_identificationyes 0.7564 10.2239 *
## mean_age:character_identificationyes -0.0127 -0.0008 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1<36, young
ma_data_young %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = character_identification)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by character_identification, young only") 
m_age_ci_young <- rma.mv(d_calc ~ mean_age + character_identification, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_ci_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.5412 117.0824 125.0824 131.5260 126.3324
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3936 0.6274 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 37) = 170.9711, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 1.0718, p-val = 0.5852
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.8547 0.4398 1.9432 0.0520 -0.0074 1.7167
## mean_age -0.0005 0.0005 -0.9353 0.3497 -0.0015 0.0005
## character_identificationyes 0.2105 0.4135 0.5091 0.6107 -0.6000 1.0211
##
## intrcpt .
## mean_age
## character_identificationyes
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_ci_interaction_young <- rma.mv(d_calc ~ mean_age * character_identification, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_ci_interaction_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -55.3377 110.6753 120.6753 128.5929 122.6753
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3054 0.5527 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 159.9117, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 6.4630, p-val = 0.0911
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.7066 0.4283 1.6498 0.0990
## mean_age -0.0003 0.0005 -0.5496 0.5826
## character_identificationyes 5.7332 2.4227 2.3664 0.0180
## mean_age:character_identificationyes -0.0071 0.0031 -2.3109 0.0208
## ci.lb ci.ub
## intrcpt -0.1328 1.5461 .
## mean_age -0.0013 0.0007
## character_identificationyes 0.9847 10.4817 *
## mean_age:character_identificationyes -0.0130 -0.0011 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1>= 36, old
ma_data_old %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = character_identification)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by character_identification, old only") 
vocab
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = productive_vocab_median, y = d_calc, size = n_1, color = character_identification)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("median vocab") +
ggtitle("Syntactical Bootstrapping effect size vs. median voca, breakdown by character_identification, old only") 
m_age_ci_vocab <- rma.mv(d_calc ~ productive_vocab_median + character_identification, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_ci_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -43.4223 86.8447 94.8447 99.8771 96.7494
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.1053 0.3245 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 90.8078, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 10.1358, p-val = 0.0063
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.8765 0.2245 3.9045 <.0001 0.4365
## productive_vocab_median -0.0059 0.0027 -2.1387 0.0325 -0.0113
## character_identificationyes 1.2784 0.5772 2.2147 0.0268 0.1470
## ci.ub
## intrcpt 1.3165 ***
## productive_vocab_median -0.0005 *
## character_identificationyes 2.4098 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_ci_interaction_vocab <- rma.mv(d_calc ~ productive_vocab_median * character_identification, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_ci_interaction_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -42.6415 85.2830 95.2830 101.3774 98.4409
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.1054 0.3247 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 90.1698, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 10.7738, p-val = 0.0130
##
## Model Results:
##
## estimate se zval
## intrcpt 0.8770 0.2246 3.9056
## productive_vocab_median -0.0059 0.0027 -2.1425
## character_identificationyes -12.3557 17.0514 -0.7246
## productive_vocab_median:character_identificationyes 0.4771 0.5963 0.8000
## pval ci.lb ci.ub
## intrcpt <.0001 0.4369 1.3171
## productive_vocab_median 0.0322 -0.0113 -0.0005
## character_identificationyes 0.4687 -45.7758 21.0644
## productive_vocab_median:character_identificationyes 0.4237 -0.6917 1.6458
##
## intrcpt ***
## productive_vocab_median *
## character_identificationyes
## productive_vocab_median:character_identificationyes
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1age w/ vocab comparison
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = mean_age, y = d_calc, size = n_1, color = character_identification)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by character_identification, old only") 
m_age_ci_vocab_a <- rma.mv(d_calc ~ mean_age + character_identification, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_ci_vocab_a)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -42.7953 85.5907 93.5907 98.6231 95.4955
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0689 0.2626 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 87.9091, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 12.7424, p-val = 0.0017
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 1.0487 0.2436 4.3040 <.0001 0.5711
## mean_age -0.0006 0.0003 -2.4614 0.0138 -0.0011
## character_identificationyes 1.3323 0.5347 2.4918 0.0127 0.2843
## ci.ub
## intrcpt 1.5262 ***
## mean_age -0.0001 *
## character_identificationyes 2.3802 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_ci_interaction_vocab_a <- rma.mv(d_calc ~ mean_age * character_identification, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_ci_interaction_vocab_a)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -42.0190 84.0380 94.0380 100.1323 97.1959
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0689 0.2626 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 87.2777, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 13.3732, p-val = 0.0039
##
## Model Results:
##
## estimate se zval pval
## intrcpt 1.0489 0.2437 4.3050 <.0001
## mean_age -0.0006 0.0003 -2.4628 0.0138
## character_identificationyes -72.1316 92.4738 -0.7800 0.4354
## mean_age:character_identificationyes 0.1167 0.1469 0.7944 0.4269
## ci.lb ci.ub
## intrcpt 0.5714 1.5265 ***
## mean_age -0.0011 -0.0001 *
## character_identificationyes -253.3769 109.1137
## mean_age:character_identificationyes -0.1712 0.4047
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1practice_phase
single moderator
m_PP <- rma.mv(d_calc ~ practice_phase, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_PP)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -61.7978 123.5957 129.5957 134.6623 130.2623
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4533 0.6733 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 172.6382, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 1.4533, p-val = 0.2280
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.6382 0.2155 2.9613 0.0031 0.2158 1.0606 **
## practice_phaseyes -0.1665 0.1381 -1.2055 0.2280 -0.4371 0.1042
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1all
ma_data %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = practice_phase)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by practice_phase") 
m_age_pp <- rma.mv(d_calc ~ mean_age + character_identification, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_pp)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.2226 116.4451 124.4451 131.0994 125.6216
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3898 0.6243 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 39) = 171.0139, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 7.2531, p-val = 0.0266
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.9951 0.3013 3.3025 0.0010 0.4045
## mean_age -0.0007 0.0003 -2.6493 0.0081 -0.0012
## character_identificationyes 0.2228 0.4108 0.5423 0.5876 -0.5823
## ci.ub
## intrcpt 1.5857 ***
## mean_age -0.0002 **
## character_identificationyes 1.0279
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_pp_interaction <- rma.mv(d_calc ~ mean_age * character_identification, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_pp_interaction)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -55.2000 110.4001 120.4001 128.5880 122.2751
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3124 0.5589 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 160.5083, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 12.2078, p-val = 0.0067
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.9553 0.2849 3.3531 0.0008
## mean_age -0.0006 0.0003 -2.4627 0.0138
## character_identificationyes 5.4901 2.4152 2.2731 0.0230
## mean_age:character_identificationyes -0.0067 0.0030 -2.2110 0.0270
## ci.lb ci.ub
## intrcpt 0.3969 1.5137 ***
## mean_age -0.0011 -0.0001 *
## character_identificationyes 0.7564 10.2239 *
## mean_age:character_identificationyes -0.0127 -0.0008 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1<36, young
ma_data_young %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = practice_phase)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by practice_phase, young only") 
m_age_pp_young <- rma.mv(d_calc ~ mean_age + practice_phase, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_pp_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.8727 115.7454 123.7454 130.1891 124.9954
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2274 0.4769 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 37) = 147.0687, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 4.2355, p-val = 0.1203
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.2591 0.4495 2.8011 0.0051 0.3781 2.1401 **
## mean_age -0.0012 0.0006 -1.8612 0.0627 -0.0024 0.0001 .
## practice_phaseyes 0.3731 0.2029 1.8394 0.0659 -0.0245 0.7707 .
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_pp_interaction_young <- rma.mv(d_calc ~ mean_age * practice_phase, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_pp_interaction_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -56.8491 113.6982 123.6982 131.6158 125.6982
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2434 0.4934 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 146.8482, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 5.0994, p-val = 0.1647
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.8415 0.7161 2.5716 0.0101 0.4380 3.2451
## mean_age -0.0020 0.0010 -1.9543 0.0507 -0.0041 0.0000
## practice_phaseyes -0.7841 1.1023 -0.7113 0.4769 -2.9446 1.3764
## mean_age:practice_phaseyes 0.0017 0.0016 1.0592 0.2895 -0.0014 0.0048
##
## intrcpt *
## mean_age .
## practice_phaseyes
## mean_age:practice_phaseyes
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1>= 36, old
ma_data_old %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = practice_phase)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by practice_phase, old only") 
vocab
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = productive_vocab_median, y = d_calc, size = n_1, color = practice_phase)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("productive vocab") +
ggtitle("Syntactical Bootstrapping effect size vs. productive vocab, breakdown by practice_phase, old only") 
m_age_pp_vocab <- rma.mv(d_calc ~ productive_vocab_median + practice_phase, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_pp_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.5790 91.1579 99.1579 104.1903 101.0627
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2482 0.4982 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 96.9734, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 5.8414, p-val = 0.0539
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.1202 0.2771 4.0421 <.0001 0.5770 1.6634
## productive_vocab_median -0.0092 0.0054 -1.7139 0.0865 -0.0197 0.0013
## practice_phaseyes 0.1607 0.2760 0.5824 0.5603 -0.3802 0.7017
##
## intrcpt ***
## productive_vocab_median .
## practice_phaseyes
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_pp_interaction_vocab <- rma.mv(d_calc ~ productive_vocab_median * practice_phase, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_pp_interaction_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.4813 90.9626 100.9626 107.0570 104.1205
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2533 0.5033 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 96.6515, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 6.3738, p-val = 0.0948
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.6983 0.6426 1.0866 0.2772
## productive_vocab_median 0.0143 0.0325 0.4400 0.6600
## practice_phaseyes 0.5680 0.6232 0.9115 0.3620
## productive_vocab_median:practice_phaseyes -0.0232 0.0317 -0.7320 0.4642
## ci.lb ci.ub
## intrcpt -0.5612 1.9577
## productive_vocab_median -0.0494 0.0781
## practice_phaseyes -0.6534 1.7894
## productive_vocab_median:practice_phaseyes -0.0853 0.0389
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1age w/ vocab comparsion
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = mean_age, y = d_calc, size = n_1, color = practice_phase)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("mean age") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by practice_phase, old only") 
m_age_pp_vocab_a <- rma.mv(d_calc ~ mean_age + practice_phase, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_pp_vocab_a)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.3154 90.6309 98.6309 103.6633 100.5357
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.2027 0.4502 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 93.1252, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 6.9808, p-val = 0.0305
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.3456 0.3016 4.4613 <.0001 0.7544 1.9367 ***
## mean_age -0.0009 0.0004 -2.0607 0.0393 -0.0017 -0.0000 *
## practice_phaseyes 0.1579 0.2385 0.6618 0.5081 -0.3097 0.6254
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_pp_interaction_vocab_a <- rma.mv(d_calc ~ mean_age * practice_phase, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_pp_interaction_vocab_a)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -45.1723 90.3445 100.3445 106.4389 103.5024
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.1842 0.4292 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 92.2011, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 7.3474, p-val = 0.0616
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.7675 1.0653 0.7205 0.4712 -1.3205 2.8556
## mean_age 0.0002 0.0019 0.0799 0.9363 -0.0036 0.0039
## practice_phaseyes 0.8638 1.2688 0.6808 0.4960 -1.6229 3.3505
## mean_age:practice_phaseyes -0.0012 0.0021 -0.5596 0.5757 -0.0054 0.0030
##
## intrcpt
## mean_age
## practice_phaseyes
## mean_age:practice_phaseyes
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1test_or_mass_distributed
single moderator
m_TMD <- rma.mv(d_calc ~ test_mass_or_distributed, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_TMD)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -61.2996 122.5992 128.5992 133.6659 129.2659
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3940 0.6277 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 178.6641, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.2020, p-val = 0.6531
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.4935 0.2367 2.0845 0.0371 0.0295 0.9574
## test_mass_or_distributedmass 0.1837 0.4087 0.4494 0.6531 -0.6174 0.9848
##
## intrcpt *
## test_mass_or_distributedmass
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1all
ma_data %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = test_mass_or_distributed)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by test type") 
m_age_tt <- rma.mv(d_calc ~ mean_age + test_mass_or_distributed, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_tt)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.1412 116.2823 124.2823 130.9366 125.4588
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3920 0.6261 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 39) = 171.5749, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 7.0816, p-val = 0.0290
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 1.0143 0.3087 3.2862 0.0010 0.4094
## mean_age -0.0007 0.0003 -2.6227 0.0087 -0.0012
## test_mass_or_distributedmass 0.1433 0.4081 0.3511 0.7255 -0.6566
## ci.ub
## intrcpt 1.6193 **
## mean_age -0.0002 **
## test_mass_or_distributedmass 0.9432
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_tt_interaction <- rma.mv(d_calc ~ mean_age * test_mass_or_distributed, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_age_tt_interaction)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.0026 114.0051 124.0051 132.1930 125.8801
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4486 0.6697 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 38) = 171.5736, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 7.9704, p-val = 0.0466
##
## Model Results:
##
## estimate se zval pval
## intrcpt 2.9486 2.0323 1.4509 0.1468
## mean_age -0.0031 0.0026 -1.2147 0.2245
## test_mass_or_distributedmass -1.8049 2.0710 -0.8715 0.3835
## mean_age:test_mass_or_distributedmass 0.0025 0.0026 0.9611 0.3365
## ci.lb ci.ub
## intrcpt -1.0346 6.9317
## mean_age -0.0082 0.0019
## test_mass_or_distributedmass -5.8639 2.2541
## mean_age:test_mass_or_distributedmass -0.0026 0.0076
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1<36, young
ma_data_young %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = test_mass_or_distributed)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by test_mass_or_distributed, young only") 
m_age_tt_young <- rma.mv(d_calc ~ mean_age + test_mass_or_distributed, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_tt_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -58.4923 116.9846 124.9846 131.4283 126.2346
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3903 0.6247 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 37) = 168.6961, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 0.9872, p-val = 0.6104
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.8303 0.4659 1.7823 0.0747 -0.0828
## mean_age -0.0004 0.0005 -0.8393 0.4013 -0.0014
## test_mass_or_distributedmass 0.1714 0.4108 0.4172 0.6765 -0.6337
## ci.ub
## intrcpt 1.7434 .
## mean_age 0.0006
## test_mass_or_distributedmass 0.9765
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_tt_interaction_young <- rma.mv(d_calc ~ mean_age * test_mass_or_distributed, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_young)
summary(m_age_tt_interaction_young)##
## Multivariate Meta-Analysis Model (k = 40; method: REML)
##
## logLik Deviance AIC BIC AICc
## -57.2803 114.5607 124.5607 132.4782 126.5607
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.4407 0.6638 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 36) = 168.2898, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 2.0943, p-val = 0.5531
##
## Model Results:
##
## estimate se zval pval
## intrcpt 2.9321 2.0203 1.4513 0.1467
## mean_age -0.0031 0.0026 -1.2140 0.2247
## test_mass_or_distributedmass -2.0022 2.0813 -0.9620 0.3361
## mean_age:test_mass_or_distributedmass 0.0028 0.0026 1.0665 0.2862
## ci.lb ci.ub
## intrcpt -1.0276 6.8918
## mean_age -0.0082 0.0019
## test_mass_or_distributedmass -6.0815 2.0771
## mean_age:test_mass_or_distributedmass -0.0023 0.0079
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1>= 36, old
ma_data_old %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = age_months, y = d_calc, size = n_1, color = test_mass_or_distributed)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by test_mass_or_distributed, old only") 
vocab
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = productive_vocab_median, y = d_calc, size = n_1, color = test_mass_or_distributed)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("median vocab") +
ggtitle("Syntactical Bootstrapping effect size vs. median voca, breakdown by test_mass_or_distributed, old only") 
m_age_tt_vocab <- rma.mv(d_calc ~ productive_vocab_median + test_mass_or_distributed, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_tt_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -43.4223 86.8447 94.8447 99.8771 96.7494
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.1053 0.3245 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 90.8078, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 10.1358, p-val = 0.0063
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 2.1549 0.5487 3.9275 <.0001 1.0795
## productive_vocab_median -0.0059 0.0027 -2.1387 0.0325 -0.0113
## test_mass_or_distributedmass -1.2784 0.5772 -2.2147 0.0268 -2.4098
## ci.ub
## intrcpt 3.2303 ***
## productive_vocab_median -0.0005 *
## test_mass_or_distributedmass -0.1470 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_tt_interaction_vocab <- rma.mv(d_calc ~ productive_vocab_median * test_mass_or_distributed, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_tt_interaction_vocab)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -42.6415 85.2830 95.2830 101.3774 98.4409
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.1054 0.3247 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 90.1698, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 10.7738, p-val = 0.0130
##
## Model Results:
##
## estimate se
## intrcpt -11.4787 17.0499
## productive_vocab_median 0.4712 0.5963
## test_mass_or_distributedmass 12.3557 17.0514
## productive_vocab_median:test_mass_or_distributedmass -0.4771 0.5963
## zval pval ci.lb
## intrcpt -0.6732 0.5008 -44.8959
## productive_vocab_median 0.7902 0.4294 -0.6975
## test_mass_or_distributedmass 0.7246 0.4687 -21.0644
## productive_vocab_median:test_mass_or_distributedmass -0.8000 0.4237 -1.6458
## ci.ub
## intrcpt 21.9385
## productive_vocab_median 1.6399
## test_mass_or_distributedmass 45.7758
## productive_vocab_median:test_mass_or_distributedmass 0.6917
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1age w/ vocab comparison
ma_data_vocab %>%
mutate(age_months = mean_age/30.44) %>%
ggplot(aes(x = mean_age, y = d_calc, size = n_1, color = test_mass_or_distributed)) +
geom_point() +
geom_smooth(method = "lm") +
ylab("Effect Size") +
xlab("Age (months)") +
ggtitle("Syntactical Bootstrapping effect size vs. Age (months), breakdown by test_mass_or_distributed, old only") 
m_age_tt_vocab_a <- rma.mv(d_calc ~ mean_age + test_mass_or_distributed, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_tt_vocab_a)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -42.7953 85.5907 93.5907 98.6231 95.4955
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0689 0.2626 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 26) = 87.9091, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 12.7424, p-val = 0.0017
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 2.3810 0.5330 4.4668 <.0001 1.3362
## mean_age -0.0006 0.0003 -2.4614 0.0138 -0.0011
## test_mass_or_distributedmass -1.3323 0.5347 -2.4918 0.0127 -2.3802
## ci.ub
## intrcpt 3.4257 ***
## mean_age -0.0001 *
## test_mass_or_distributedmass -0.2843 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_age_tt_interaction_vocab_a <- rma.mv(d_calc ~ mean_age * test_mass_or_distributed, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data_vocab)
summary(m_age_tt_interaction_vocab_a)##
## Multivariate Meta-Analysis Model (k = 29; method: REML)
##
## logLik Deviance AIC BIC AICc
## -42.0190 84.0380 94.0380 100.1323 97.1959
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0689 0.2626 5 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 25) = 87.2777, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## QM(df = 3) = 13.3732, p-val = 0.0039
##
## Model Results:
##
## estimate se zval pval
## intrcpt -71.0827 92.4735 -0.7687 0.4421
## mean_age 0.1161 0.1469 0.7902 0.4294
## test_mass_or_distributedmass 72.1316 92.4738 0.7800 0.4354
## mean_age:test_mass_or_distributedmass -0.1167 0.1469 -0.7944 0.4269
## ci.lb ci.ub
## intrcpt -252.3274 110.1620
## mean_age -0.1719 0.4040
## test_mass_or_distributedmass -109.1137 253.3769
## mean_age:test_mass_or_distributedmass -0.4047 0.1712
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1n_train_test_pair
single moderator
m_TTP <- rma.mv(d_calc ~ n_train_test_pair, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_TTP)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -61.3100 122.6199 128.6199 133.6866 129.2866
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3927 0.6266 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 178.7744, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.3392, p-val = 0.5603
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.7673 0.4124 1.8605 0.0628 -0.0410 1.5757 .
## n_train_test_pair -0.0841 0.1444 -0.5824 0.5603 -0.3671 0.1989
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1ma_data %>% ggplot(aes(x = n_train_test_pair)) +
geom_histogram() 
n_test_trial
single moderator
m_TeT <- rma.mv(d_calc ~ n_test_trial_per_pair, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_TeT)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -62.4460 124.8919 130.8919 135.9586 131.5586
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3589 0.5990 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 178.8760, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.1984, p-val = 0.6560
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 0.4696 0.2612 1.7979 0.0722 -0.0423 0.9815 .
## n_test_trial_per_pair 0.0458 0.1028 0.4455 0.6560 -0.1557 0.2473
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1ma_data %>% ggplot(aes(x = n_test_trial_per_pair)) +
geom_histogram() 
Single Moderator
summary(m_age)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -59.3060 118.6119 124.6119 129.6785 125.2786
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3531 0.5942 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 40) = 172.5325, p-val < .0001
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 6.9726, p-val = 0.0083
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.0613 0.2667 3.9788 <.0001 0.5385 1.5842 ***
## mean_age -0.0007 0.0003 -2.6406 0.0083 -0.0012 -0.0002 **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1s_age <- coef(summary(m_age)) %>% rownames_to_column(var = "coefficient_name")
s_vocab <- coef(summary(m_vocab)) %>% rownames_to_column(var = "coefficient_name")
s_rep <- coef(summary(m_rep)) %>% rownames_to_column(var = "coefficient_name")
s_SS <- coef(summary(m_SS)) %>% rownames_to_column(var = "coefficient_name")
s_AAT <- coef(summary(m_AAT)) %>% rownames_to_column(var = "coefficient_name")
s_PAT <- coef(summary(m_PAT)) %>% rownames_to_column(var = "coefficient_name")
s_AAN <- coef(summary(m_AAN)) %>% rownames_to_column(var = "coefficient_name")
s_SM <- coef(summary(m_SM)) %>% rownames_to_column(var = "coefficient_name")
s_SA <- coef(summary(m_SA)) %>% rownames_to_column(var = "coefficient_name")
s_TE <- coef(summary(m_TE)) %>% rownames_to_column(var = "coefficient_name")
s_IE <- coef(summary(m_IE)) %>% rownames_to_column(var = "coefficient_name")
s_VSP <- coef(summary(m_VSP)) %>% rownames_to_column(var = "coefficient_name")
s_RV <- coef(summary(m_RV)) %>% rownames_to_column(var = "coefficient_name")
s_TM <- coef(summary(m_TM)) %>% rownames_to_column(var = "coefficient_name")
s_PT <- coef(summary(m_PT)) %>% rownames_to_column(var = "coefficient_name")
s_CI <- coef(summary(m_CI)) %>% rownames_to_column(var = "coefficient_name")
s_PP <- coef(summary(m_PP)) %>% rownames_to_column(var = "coefficient_name")
s_TMD <- coef(summary(m_TMD)) %>% rownames_to_column(var = "coefficient_name")
s_TTP <- coef(summary(m_TTP)) %>% rownames_to_column(var = "coefficient_name")
s_TeT <- coef(summary(m_TeT)) %>% rownames_to_column(var = "coefficient_name")
s_moderators <- bind_rows(s_age, s_vocab, s_SS,s_AAT,s_PAT,s_AAN,s_SM,s_SA,s_TE,s_IE,s_VSP,s_RV,s_TM,s_PT,s_CI,s_PP,s_TMD,s_TTP,s_TeT
) %>% filter(coefficient_name != "intrcpt") %>%
mutate(moderator_types = case_when(
coefficient_name == "mean_age" ~ "Participants_Characteristics",
coefficient_name == "productive_vocab_median" ~ "Participants_Characteristics",
coefficient_name == "sentence_structuretransitive" ~ "Linguistic_Stimuli",
coefficient_name == "agent_argument_type_cleannoun_phrase" ~ "Linguistic_Stimuli",
coefficient_name == "agent_argument_type_cleanpronoun" ~ "Linguistic_Stimuli",
coefficient_name == "agent_argument_type_cleanvarying_agent" ~ "Linguistic_Stimuli",
coefficient_name == "patient_argument_type_cleannoun"~ "Linguistic_Stimuli",
coefficient_name == "patient_argument_type_cleannoun_phrase"~ "Linguistic_Stimuli",
coefficient_name == "patient_argument_type_cleanpronoun"~ "Linguistic_Stimuli",
coefficient_name == "patient_argument_type_cleanvarying_patient"~ "Linguistic_Stimuli",
coefficient_name == "agent_argument_number2"~ "Linguistic_Stimuli",
coefficient_name == "agent_argument_numbervarying"~ "Linguistic_Stimuli",
coefficient_name == "n_repetitions_sentence"~ "Linguistic_Stimuli",
coefficient_name == "patient_argument_type_cleanvarying_patient"~ "Linguistic_Stimuli",
coefficient_name == "n_repetitions_video"~ "Visual_Stimuli",
coefficient_name == "stimuli_modalityvideo" ~ "Visual_Stimuli",
coefficient_name == "stimuli_actorperson" ~ "Visual_Stimuli",
coefficient_name == "transitive_event_typeindirect_caused_action" ~ "Visual_Stimuli",
coefficient_name == "transitive_event_typeminimal_contact" ~ "Visual_Stimuli",
coefficient_name == "intransitive_event_typeone_action" ~ "Visual_Stimuli",
coefficient_name == "intransitive_event_typeparallel_actions" ~ "Visual_Stimuli",
coefficient_name == "visual_stimuli_pairdirect_caused_action_one_action" ~ "Visual_Stimuli",
coefficient_name == "visual_stimuli_pairdirect_caused_action_parallel_actions" ~ "Visual_Stimuli",
coefficient_name == "visual_stimuli_pairindirect_caused_action_one_action" ~ "Visual_Stimuli",
coefficient_name == "visual_stimuli_pairindirect_caused_action_parallel_actions" ~ "Visual_Stimuli",
coefficient_name == "visual_stimuli_pairminimal_contact_one_action" ~ "Visual_Stimuli",
coefficient_name == "visual_stimuli_pairminimal_contact_parallel_actions" ~ "Visual_Stimuli",
coefficient_name == "test_methodpoint" ~ "Test_Procedure",
coefficient_name == "presentation_typeimmediate_after" ~ "Test_Procedure",
coefficient_name == "presentation_typesimultaneous" ~ "Test_Procedure",
coefficient_name == "character_identificationyes" ~ "Test_Procedure",
coefficient_name == "practice_phaseyes" ~ "Test_Procedure",
coefficient_name == "test_mass_or_distributedmass" ~ "Test_Procedure",
coefficient_name == "n_train_test_pair" ~ "Test_Procedure",
coefficient_name == "n_test_trial_per_pair" ~ "Test_Procedure",
)
)s_moderators %>% filter(moderator_types == "Participants_Characteristics") %>%
ggplot(aes(x = coefficient_name, y = estimate)) +
geom_point(size = 2) +
geom_linerange(aes(ymin = ci.lb, ymax = ci.ub)) +
geom_hline(aes(yintercept = 0), linetype = 2) +
theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))
s_moderators %>% filter(moderator_types == "Linguistic_Stimuli") %>%
ggplot(aes(x = coefficient_name, y = estimate)) +
geom_point(size = 2) +
geom_linerange(aes(ymin = ci.lb, ymax = ci.ub)) +
geom_hline(aes(yintercept = 0), linetype = 2) +
theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))
s_moderators %>% filter(moderator_types == "Visual_Stimuli") %>%
ggplot(aes(x = coefficient_name, y = estimate)) +
geom_point(size = 2) +
geom_linerange(aes(ymin = ci.lb, ymax = ci.ub)) +
geom_hline(aes(yintercept = 0), linetype = 2) +
theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))
s_moderators %>% filter(moderator_types == "Test_Procedure") %>%
ggplot(aes(x = coefficient_name, y = estimate)) +
geom_point(size = 2) +
geom_linerange(aes(ymin = ci.lb, ymax = ci.ub)) +
geom_hline(aes(yintercept = 0), linetype = 2) +
theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))
s_moderators %>%
ggplot(aes(x = coefficient_name, y = estimate, color = moderator_types)) +
geom_point(size = 2) +
geom_linerange(aes(ymin = ci.lb, ymax = ci.ub)) +
geom_hline(aes(yintercept = 0), linetype = 2) +
theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) 
Reasonalbe Moderators (??)
null <- rma.mv(d_calc~1, V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(null)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -62.5424 125.0848 129.0848 132.5119 129.4006
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.3591 0.5992 12 no short_cite
##
## Test for Heterogeneity:
## Q(df = 41) = 181.5424, p-val < .0001
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## 0.5517 0.1852 2.9783 0.0029 0.1886 0.9147 **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_sink <- rma.mv(d_calc ~ mean_age + sentence_structure + agent_argument_type_clean + agent_argument_number + n_repetitions_sentence + test_method + character_identification + practice_phase + test_mass_or_distributed , V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_sink)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -33.7536 67.5072 89.5072 105.6303 102.7072
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0000 0.0000 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 32) = 73.8636, p-val < .0001
##
## Test of Moderators (coefficients 2:10):
## QM(df = 9) = 107.6788, p-val < .0001
##
## Model Results:
##
## estimate se zval pval
## intrcpt 0.3738 0.4544 0.8226 0.4107
## mean_age -0.0018 0.0004 -4.2008 <.0001
## sentence_structuretransitive 0.6611 0.1457 4.5380 <.0001
## agent_argument_type_cleannoun_phrase 0.8190 0.2274 3.6009 0.0003
## agent_argument_type_cleanpronoun 0.0975 0.3440 0.2834 0.7769
## agent_argument_type_cleanvarying_agent 0.9092 0.1682 5.4065 <.0001
## n_repetitions_sentence -0.0102 0.0130 -0.7873 0.4311
## test_methodpoint 0.6972 0.3098 2.2505 0.0244
## character_identificationyes 0.7993 0.1949 4.1014 <.0001
## practice_phaseyes 1.0321 0.2211 4.6672 <.0001
## ci.lb ci.ub
## intrcpt -0.5168 1.2644
## mean_age -0.0027 -0.0010 ***
## sentence_structuretransitive 0.3756 0.9467 ***
## agent_argument_type_cleannoun_phrase 0.3732 1.2647 ***
## agent_argument_type_cleanpronoun -0.5767 0.7717
## agent_argument_type_cleanvarying_agent 0.5796 1.2388 ***
## n_repetitions_sentence -0.0357 0.0152
## test_methodpoint 0.0900 1.3045 *
## character_identificationyes 0.4173 1.1812 ***
## practice_phaseyes 0.5987 1.4655 ***
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1m_sink_2 <- rma.mv(d_calc ~ mean_age + sentence_structure + n_repetitions_sentence + test_method + character_identification + practice_phase + test_mass_or_distributed , V = d_var_calc,
random = ~ 1 | short_cite, data = ma_data)
summary(m_sink_2)##
## Multivariate Meta-Analysis Model (k = 42; method: REML)
##
## logLik Deviance AIC BIC AICc
## -42.3566 84.7132 102.7132 116.4504 110.2132
##
## Variance Components:
##
## estim sqrt nlvls fixed factor
## sigma^2 0.0000 0.0000 12 no short_cite
##
## Test for Residual Heterogeneity:
## QE(df = 34) = 91.5106, p-val < .0001
##
## Test of Moderators (coefficients 2:8):
## QM(df = 7) = 90.0317, p-val < .0001
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.0601 0.3322 0.1811 0.8563 -0.5909
## mean_age -0.0012 0.0003 -3.9291 <.0001 -0.0019
## sentence_structuretransitive 0.3267 0.1112 2.9386 0.0033 0.1088
## n_repetitions_sentence 0.0212 0.0088 2.4108 0.0159 0.0040
## test_methodpoint 1.6837 0.2973 5.6633 <.0001 1.1010
## character_identificationyes 1.0917 0.1820 5.9994 <.0001 0.7350
## practice_phaseyes 0.6209 0.1412 4.3986 <.0001 0.3442
## test_mass_or_distributedmass 0.5828 0.1415 4.1174 <.0001 0.3054
## ci.ub
## intrcpt 0.7112
## mean_age -0.0006 ***
## sentence_structuretransitive 0.5446 **
## n_repetitions_sentence 0.0385 *
## test_methodpoint 2.2664 ***
## character_identificationyes 1.4483 ***
## practice_phaseyes 0.8976 ***
## test_mass_or_distributedmass 0.8602 ***
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1