Mass/Count Superordinate – Semantic Inferences from Syntax
2023-05-11
Demographics Stats
df.demog %>%
filter(demog_question == "gender") %>%
count(demog_response) %>%
knitr::kable()
df.demog %>%
group_by(subject_id, condition) %>%
summarise(rows = n()) %>%
select(-rows) %>%
group_by(condition) %>%
count(condition) %>%
knitr::kable()`summarise()` has grouped output by 'subject_id'. You can override using the
`.groups` argument.df.demog %>%
filter(demog_question == "age") %>%
summarise(mean_age = mean(as.numeric(demog_response)),
sd_age = sd(as.numeric(demog_response))) %>%
knitr::kable()| demog_response | n |
|---|---|
| Female | 49 |
| Male | 45 |
| Non-binary | 6 |
| condition | n |
|---|---|
| count | 45 |
| mass | 55 |
| mean_age | sd_age |
|---|---|
| 36.98 | 13.4066 |
Results
No difference between the mass and count syntax conditions (concurrent function dummy coded as 1).
ggplot(data = df.response,
mapping = aes(x = condition, y = selected_concurrent_func, color = condition)) +
geom_jitter(aes(group = subject_id),
height = 0,
alpha = 0.5) +
stat_summary(fun.data = "mean_cl_boot",
geom = "pointrange") +
theme(legend.position = "none") +
geom_hline(yintercept = 0.5, linetype = "dashed") +
ylab("Selected concurrent function (by trial)")
ggplot(data = df.response %>%
group_by(subject_id, condition) %>%
summarise(mean_concurrent_func = mean(selected_concurrent_func, na.rm = TRUE)),
mapping = aes(x = condition, y = mean_concurrent_func)) +
geom_violin(aes(fill = condition)) +
geom_jitter(height = 0,
alpha = 0.5) +
stat_summary(fun.data = "mean_cl_boot",
geom = "pointrange") +
theme(legend.position = "none") +
geom_hline(yintercept = 0.5, linetype = "dashed") +
ylab("Mean concurrent function selection (by participant)")`summarise()` has grouped output by 'subject_id'. You can override using the
`.groups` argument.
No differences between mass/count conditions even when grouped by stimuli category (by complexity, function, or complexity x function).
ggplot(data = df.response,
mapping = aes(x = category, y = selected_concurrent_func, color = condition)) +
geom_jitter(height = 0,
alpha = 0.5) +
stat_summary(fun.data = "mean_cl_boot",
geom = "pointrange",
position = position_dodge(0.7)) +
theme(axis.text.x = element_text(angle = 45, hjust = 1, size = 7)) +
geom_hline(yintercept = 0.5, linetype = "dashed") +
ylab("Mean concurrent function selection (by participant)")
ggplot(data = df.response,
mapping = aes(x = category_complexity, y = selected_concurrent_func, color = condition)) +
geom_jitter(height = 0,
alpha = 0.5) +
stat_summary(fun.data = "mean_cl_boot",
geom = "pointrange",
position = position_dodge(0.7)) +
theme(axis.text.x = element_text(angle = 45, hjust = 1, size = 7)) +
geom_hline(yintercept = 0.5, linetype = "dashed") +
ylab("Mean concurrent function selection (by participant)")
ggplot(data = df.response,
mapping = aes(x = category_function, y = selected_concurrent_func, color = condition)) +
geom_jitter(height = 0,
alpha = 0.5) +
stat_summary(fun.data = "mean_cl_boot",
geom = "pointrange",
position = position_dodge(0.7)) +
theme(axis.text.x = element_text(angle = 45, hjust = 1, size = 7)) +
geom_hline(yintercept = 0.5, linetype = "dashed") +
ylab("Mean concurrent function selection (by participant)")
Checking if the repeated trials prompt people to be more random with their selections (if this was the case, we might see a significant difference in responding between mass/count conditions in the 1st trial, as opposed to the later trials.) No distinction between conditions in any trial. But participants chose significantly more individual function groups (for both mass/count conditions) in the 1st trial, and were at chance for the other trials.
ggplot(data = df.response,
mapping = aes(x = trial_index, y = selected_concurrent_func, color = condition)) +
geom_jitter(height = 0,
alpha = 0.5) +
stat_summary(fun.data = "mean_cl_boot",
geom = "pointrange",
position = position_dodge(0.7)) +
geom_hline(yintercept = 0.5, linetype = "dashed") +
ylab("Mean concurrent function selection (by participant)")
Running the regressions show no effect for any predictors.
fit.base <- glmer(selected_concurrent_func ~ 1 + (1|subject_id) + (1|category), data = df.response, family="binomial")boundary (singular) fit: see help('isSingular')summary(fit.base)
Anova(fit.base, type = 3)
fit.cond <- glmer(selected_concurrent_func ~ condition + (1|subject_id) + (1|category), data = df.response, family="binomial")boundary (singular) fit: see help('isSingular')summary(fit.cond)
Anova(fit.cond, type = 3)
anova(fit.base, fit.cond)
fit.category <- glmer(selected_concurrent_func ~ condition + category + (1|subject_id), data = df.response, family="binomial")
summary(fit.category)
Anova(fit.category, type = 3)
fit.category_complexity <- glmer(selected_concurrent_func ~ condition + category_complexity + (1|subject_id), data = df.response, family="binomial")
summary(fit.category_complexity)
Anova(fit.category_complexity, type = 3)
fit.category_function <- glmer(selected_concurrent_func ~ condition + category_function + (1|subject_id), data = df.response, family="binomial")
summary(fit.category_function)
Anova(fit.category_function, type = 3)
fit.category_int <- glmer(selected_concurrent_func ~ condition + category_function * category_complexity + (1|subject_id), data = df.response, family="binomial")fixed-effect model matrix is rank deficient so dropping 4 columns / coefficientssummary(fit.category_int)
Anova(fit.category_int, type = 3)Generalized linear mixed model fit by maximum likelihood (Laplace
Approximation) [glmerMod]
Family: binomial ( logit )
Formula: selected_concurrent_func ~ 1 + (1 | subject_id) + (1 | category)
Data: df.response
AIC BIC logLik deviance df.resid
679.5 692.1 -336.7 673.5 497
Scaled residuals:
Min 1Q Median 3Q Max
-1.1290 -0.8351 -0.6103 1.0296 1.3969
Random effects:
Groups Name Variance Std.Dev.
subject_id (Intercept) 0.4815 0.6939
category (Intercept) 0.0000 0.0000
Number of obs: 500, groups: subject_id, 100; category, 5
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.3342 0.1190 -2.808 0.00499 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
optimizer (Nelder_Mead) convergence code: 0 (OK)
boundary (singular) fit: see help('isSingular')
Analysis of Deviance Table (Type III Wald chisquare tests)
Response: selected_concurrent_func
Chisq Df Pr(>Chisq)
(Intercept) 7.8834 1 0.004989 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Generalized linear mixed model fit by maximum likelihood (Laplace
Approximation) [glmerMod]
Family: binomial ( logit )
Formula: selected_concurrent_func ~ condition + (1 | subject_id) + (1 |
category)
Data: df.response
AIC BIC logLik deviance df.resid
681.5 698.3 -336.7 673.5 496
Scaled residuals:
Min 1Q Median 3Q Max
-1.1307 -0.8340 -0.6095 1.0280 1.3987
Random effects:
Groups Name Variance Std.Dev.
subject_id (Intercept) 0.4815 0.6939
category (Intercept) 0.0000 0.0000
Number of obs: 500, groups: subject_id, 100; category, 5
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.329339 0.176143 -1.870 0.0615 .
conditionmass -0.008924 0.237228 -0.038 0.9700
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr)
conditinmss -0.737
optimizer (Nelder_Mead) convergence code: 0 (OK)
boundary (singular) fit: see help('isSingular')
Analysis of Deviance Table (Type III Wald chisquare tests)
Response: selected_concurrent_func
Chisq Df Pr(>Chisq)
(Intercept) 3.4959 1 0.06152 .
condition 0.0014 1 0.96999
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Data: df.response
Models:
fit.base: selected_concurrent_func ~ 1 + (1 | subject_id) + (1 | category)
fit.cond: selected_concurrent_func ~ condition + (1 | subject_id) + (1 | category)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
fit.base 3 679.49 692.13 -336.74 673.49
fit.cond 4 681.48 698.34 -336.74 673.48 0.0014 1 0.97
Generalized linear mixed model fit by maximum likelihood (Laplace
Approximation) [glmerMod]
Family: binomial ( logit )
Formula: selected_concurrent_func ~ condition + category + (1 | subject_id)
Data: df.response
AIC BIC logLik deviance df.resid
686.6 716.1 -336.3 672.6 493
Scaled residuals:
Min 1Q Median 3Q Max
-1.2234 -0.8225 -0.6016 1.0179 1.4868
Random effects:
Groups Name Variance Std.Dev.
subject_id (Intercept) 0.4859 0.697
Number of obs: 500, groups: subject_id, 100
Fixed effects:
Estimate Std. Error z value
(Intercept) -3.571e-01 2.603e-01 -1.372
conditionmass -8.992e-03 2.378e-01 -0.038
categoryhigh complexity, low function 1.055e-05 3.022e-01 0.000
categorylow complexity, high function 4.570e-02 3.018e-01 0.151
categorylow complexity, low function -9.220e-02 3.032e-01 -0.304
categorymedium complexity, medium function 1.815e-01 3.011e-01 0.603
Pr(>|z|)
(Intercept) 0.170
conditionmass 0.970
categoryhigh complexity, low function 1.000
categorylow complexity, high function 0.880
categorylow complexity, low function 0.761
categorymedium complexity, medium function 0.547
Correlation of Fixed Effects:
(Intr) cndtnm ctgryhcmplxty,lf ccm,hf ctgrylcmplxty,lf
conditinmss -0.500
ctgryhcmplxty,lf -0.581 0.000
ccmplxty,hf -0.582 0.000 0.501
ctgrylcmplxty,lf -0.578 0.000 0.498 0.499
ccmplxty,mf -0.584 0.000 0.502 0.503 0.500
Analysis of Deviance Table (Type III Wald chisquare tests)
Response: selected_concurrent_func
Chisq Df Pr(>Chisq)
(Intercept) 1.8816 1 0.1702
condition 0.0014 1 0.9698
category 0.8766 4 0.9279
Generalized linear mixed model fit by maximum likelihood (Laplace
Approximation) [glmerMod]
Family: binomial ( logit )
Formula: selected_concurrent_func ~ condition + category_complexity +
(1 | subject_id)
Data: df.response
AIC BIC logLik deviance df.resid
682.8 703.9 -336.4 672.8 495
Scaled residuals:
Min 1Q Median 3Q Max
-1.2229 -0.8155 -0.6001 1.0416 1.4357
Random effects:
Groups Name Variance Std.Dev.
subject_id (Intercept) 0.4848 0.6963
Number of obs: 500, groups: subject_id, 100
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.356992 0.211875 -1.685 0.092 .
conditionmass -0.008972 0.237646 -0.038 0.970
category_complexitylow -0.022931 0.213848 -0.107 0.915
category_complexitymedium 0.181475 0.260410 0.697 0.486
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) cndtnm ctgry_cmplxtyl
conditinmss -0.614
ctgry_cmplxtyl -0.504 0.000
ctgry_cmplxtym -0.416 0.000 0.410
Analysis of Deviance Table (Type III Wald chisquare tests)
Response: selected_concurrent_func
Chisq Df Pr(>Chisq)
(Intercept) 2.8389 1 0.0920 .
condition 0.0014 1 0.9699
category_complexity 0.6711 2 0.7149
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Generalized linear mixed model fit by maximum likelihood (Laplace
Approximation) [glmerMod]
Family: binomial ( logit )
Formula: selected_concurrent_func ~ condition + category_function + (1 |
subject_id)
Data: df.response
AIC BIC logLik deviance df.resid
682.7 703.8 -336.4 672.7 495
Scaled residuals:
Min 1Q Median 3Q Max
-1.2231 -0.8061 -0.6069 1.0297 1.4525
Random effects:
Groups Name Variance Std.Dev.
subject_id (Intercept) 0.4853 0.6966
Number of obs: 500, groups: subject_id, 100
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.33417 0.21169 -1.579 0.114
conditionmass -0.00895 0.23770 -0.038 0.970
category_functionlow -0.06883 0.21390 -0.322 0.748
category_functionmedium 0.15860 0.26027 0.609 0.542
Correlation of Fixed Effects:
(Intr) cndtnm ctgry_fnctnl
conditinmss -0.614
ctgry_fnctnl -0.502 0.000
ctgry_fnctnm -0.415 0.000 0.409
Analysis of Deviance Table (Type III Wald chisquare tests)
Response: selected_concurrent_func
Chisq Df Pr(>Chisq)
(Intercept) 2.4919 1 0.1144
condition 0.0014 1 0.9700
category_function 0.7624 2 0.6830
Generalized linear mixed model fit by maximum likelihood (Laplace
Approximation) [glmerMod]
Family: binomial ( logit )
Formula:
selected_concurrent_func ~ condition + category_function * category_complexity +
(1 | subject_id)
Data: df.response
AIC BIC logLik deviance df.resid
686.6 716.1 -336.3 672.6 493
Scaled residuals:
Min 1Q Median 3Q Max
-1.2234 -0.8225 -0.6016 1.0179 1.4867
Random effects:
Groups Name Variance Std.Dev.
subject_id (Intercept) 0.4859 0.697
Number of obs: 500, groups: subject_id, 100
Fixed effects:
Estimate Std. Error z value
(Intercept) -3.571e-01 2.603e-01 -1.372
conditionmass -8.954e-03 2.378e-01 -0.038
category_functionlow 6.572e-05 3.022e-01 0.000
category_functionmedium 1.816e-01 3.011e-01 0.603
category_complexitylow 4.576e-02 3.018e-01 0.152
category_functionlow:category_complexitylow -1.379e-01 4.279e-01 -0.322
Pr(>|z|)
(Intercept) 0.170
conditionmass 0.970
category_functionlow 1.000
category_functionmedium 0.546
category_complexitylow 0.880
category_functionlow:category_complexitylow 0.747
Correlation of Fixed Effects:
(Intr) cndtnm ctgry_fnctnl ctgry_fnctnm ctgry_c
conditinmss -0.500
ctgry_fnctnl -0.581 0.000
ctgry_fnctnm -0.584 0.000 0.502
ctgry_cmplx -0.582 0.000 0.501 0.503
ctgry_fnc:_ 0.411 0.000 -0.706 -0.355 -0.706
fit warnings:
fixed-effect model matrix is rank deficient so dropping 4 columns / coefficients
Analysis of Deviance Table (Type III Wald chisquare tests)
Response: selected_concurrent_func
Chisq Df Pr(>Chisq)
(Intercept) 1.8823 1 0.1701
condition 0.0014 1 0.9700
category_function 0.4859 2 0.7843
category_complexity 0.0230 1 0.8795
category_function:category_complexity 0.1039 1 0.7471