Appendix B — Statistical Summaries
B.0.1 1. Task1 - Phrase Judgement
Logistic regression model:
formula6_b1 <- responses ~ conditions + trials + familiarity + conditions * trials + conditions * familiarity + (1 + trials|subject) + (1+ conditions |stimulus)
model6_b1 <- glmer(formula6_b1, data = data_s1, control=glmerControl(optimizer = "optimx", calc.derivs = FALSE, optCtrl = list(method = "nlminb", starttests = FALSE, kkt = FALSE)), family = binomial(link = "logit"))
MODEL INFO:
Observations: 6297
Dependent Variable: responses
Type: Mixed effects generalized linear regression
Error Distribution: binomial
Link function: logit
MODEL FIT:
AIC = 5808.26, BIC = 6003.95
Pseudo-R² (fixed effects) = 0.05
Pseudo-R² (total) = 0.31
FIXED EFFECTS:
------------------------------------------------------------------
Est. S.E. z val. p
----------------------------------- ------- ------ -------- ------
(Intercept) 1.11 0.38 2.94 0.00
conditionsNA1 -0.48 0.21 -2.31 0.02
conditionsMix 0.18 0.17 1.10 0.27
trialsitemA 0.47 0.53 0.90 0.37
trialsitemBL_A 0.74 0.53 1.40 0.16
trialsitemBL_NA 1.12 0.53 2.10 0.04
familiarityheld-out 0.26 0.31 0.85 0.40
conditionsNA1:trialsitemA 0.21 0.30 0.70 0.48
conditionsMix:trialsitemA -0.30 0.25 -1.20 0.23
conditionsNA1:trialsitemBL_A 0.43 0.31 1.36 0.17
conditionsMix:trialsitemBL_A 0.00 0.26 0.01 0.99
conditionsNA1:trialsitemBL_NA 0.20 0.33 0.61 0.54
conditionsMix:trialsitemBL_NA -0.10 0.28 -0.35 0.73
------------------------------------------------------------------
Emmeans
| contrast | trial_type | estimate | SE | z.ratio | p.value | |
|---|---|---|---|---|---|---|
| A1 - NA1 | itemNonA | 0.48 | 0.21 | 2.31 | 0.06 | |
| itemA | 0.27 | 0.24 | 1.13 | 0.78 | ||
| itemBL_A | 0.05 | 0.25 | 0.21 | 1.00 | ||
| itemBL_NA | 0.28 | 0.28 | 0.99 | 0.96 | ||
| A1 - Mix | itemNonA | -0.18 | 0.17 | -1.10 | 0.82 | |
| itemA | 0.11 | 0.20 | 0.57 | 1.00 | ||
| itemBL_A | -0.19 | 0.22 | -0.85 | 1.00 | ||
| itemBL_NA | -0.09 | 0.26 | -0.34 | 1.00 | ||
| NA1 - Mix | itemNonA | -0.67 | 0.21 | -3.20 | 0.00 | |
| itemA | -0.15 | 0.24 | -0.66 | 1.00 | ||
| itemBL_A | -0.24 | 0.26 | -0.93 | 1.00 | ||
| itemBL_NA | -0.37 | 0.29 | -1.27 | 0.61 |
Chi-Square Anova tests
2 main effects (without condition) vs 3 main effects
Models:
model3_b1: responses ~ trial_type + familiarity + (1 + trial_type | subject) + (1 + conditions| stimulus)
model4_b1: responses ~ conditions+ trial_type + familiarity + (1 + trial_type | subject) + (1 + conditions| stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model3_b1 21 5804.1 5945.8 -2881.1 5762.1
model4_b1 23 5800.7 5955.9 -2877.4 5754.7 7.3755 2 0.02503 *
2 main effects (without trials) vs 3 main effects
Models:
model2_b1: responses ~ conditions+ familiarity + (1 + trial_type | subject) + (1 + conditions| stimulus)
model4_b1: responses ~ conditions+ trial_type + familiarity + (1 + trial_type | subject) + (1 + conditions| stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model2_b1 20 5806.8 5941.8 -2883.4 5766.8
model4_b1 23 5800.7 5955.9 -2877.4 5754.7 12.092 3 0.007074 **
2 main effects (without familiarity) vs 3 main effects
Models:
model1_b1: responses ~ conditions+ trial_type + (1 + trial_type | subject) + (1 + conditions| stimulus)
model4_b1: responses ~ conditions+ trial_type + familiarity + (1 + trial_type | subject) + (1 + conditions| stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model1_b1 22 5799.4 5947.9 -2877.7 5755.4
model4_b1 23 5800.7 5955.9 -2877.4 5754.7 0.6758 1 0.411
3 main effects (without interaction) vs with interaction
Models:
model4_b1: responses ~ conditions+ trial_type + familiarity + (1 + trial_type | subject) + (1 + conditions| stimulus)
model5_b1: responses ~ conditions + trials + familiarity + conditions * trials + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model4_b1 23 5800.7 5955.9 -2877.4 5754.7
model5_b1 29 5808.3 6003.9 -2875.1 5750.3 4.4784 6 0.6122
B.0.2 2. Task2 - Familiar segmentation
Logistic regression model:
formula5_b2 <- responses ~ conditions + trials + familiarity + conditions * trials + (1 + trials|subject) + (1+ conditions |stimulus)
model5_b2 <- glmer(formula5_b2, data = data_s2, control=glmerControl(optimizer = "optimx", calc.derivs = FALSE, optCtrl = list(method = "nlminb", starttests = FALSE, kkt = FALSE)), family = binomial(link = "logit"))
MODEL INFO:
Observations:3141
Dependent Variable:responses
Type:Mixed effects generalized linear regression
Error Distribution:binomial
Link function:logit
MODEL FIT:
AIC= 2499.17, BIC = 2596.00
Pseudo-R² (fixed effects)= 0.42
Pseudo-R² (total)= 0.72
FIXED EFFECTS:
-----------------------------------------------------------------
Est. S.E. z val. p
---------------------------------- ------- ------ -------- ------
(Intercept) -1.86 0.30 -6.16 0.00
conditionsNA1 0.21 0.44 0.47 0.64
conditionsMix 0.61 0.50 1.22 0.22
trialsitemBL_A 4.47 0.41 10.84 0.00
familiarityheld-out 0.17 0.17 1.00 0.32
conditionsNA1:trialsitemBL_A -0.10 0.61 -0.17 0.86
conditionsMix:trialsitemBL_A -0.17 0.71 -0.24 0.81
-----------------------------------------------------------------
Emmeans:
| contrast | trial_type | estimate | SE | z.ratio | p.value | |
|---|---|---|---|---|---|---|
| A1 - NA1 | itemA | -0.21 | 0.44 | -0.47 | 1.00 | |
| itemBL_A | -0.61 | 0.50 | -1.22 | 0.67 | ||
| NA1 - Mix | item_A | -0.40 | 0.46 | -0.87 | 1.00 | |
| itemBL_A | -0.10 | 0.34 | -0.31 | 1.00 | ||
| A1 - Mix | itemA | -0.44 | 0.43 | -1.03 | 0.90 | |
| itemBL_A | -0.34 | 0.37 | -0.89 | 1.00 |
ChiSquare Anova test:
2 main effects (without condition) vs 3 main effects
model3_b2: responses ~ trials+ familiarity+ (1 + trials| subject) + (1 + conditions| stimulus)
model4_b2: responses ~ conditions+ trials+ familiarity+ (1 + trials| subject) + (1 + conditions| stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model3_b2 12 2494.2 2566.9 -1235.1 2470.2
model4_b2 14 2495.2 2580.0 -1233.6 2467.2 3.0025 2 0.2229
2 main effects (without trial) vs 3 main effects
Models:
model2_b2: responses ~ conditions+ familiarity+ (1 + trials| subject) + (1 + conditions| stimulus)
model4_b2: responses ~ conditions+ trials+ familiarity+ (1 + trials| subject) + (1 + conditions| stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model2_b2 13 2548.2 2626.9 -1261.1 2522.2
model4_b2 14 2495.2 2580.0 -1233.6 2467.2 54.995 1 1.209e-13 ***
2 main effects (without familiarity) vs 3 main effects
Models:
model1_b2: responses ~ conditions+ trials+ (1 + trials| subject) + (1 + conditions| stimulus)
model4_b2: responses ~ conditions+ trials+ familiarity+ (1 + trials| subject) + (1 + conditions| stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model1_b2 13 2494.1 2572.8 -1234.1 2468.1
model4_b2 14 2495.2 2580.0 -1233.6 2467.2 0.9062 1 0.3411
3 main effects (without interaction) vs with interaction
Models:
model4_b2: responses ~ conditions+ trials+ familiarity+ (1 + trials| subject) + (1 + conditions| stimulus)
model5_b2: responses ~ conditions + trials + familiarity + conditions * trials + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model4_b2 14 2495.2 2580 -1233.6 2467.2
model5_b2 16 2499.2 2596 -1233.6 2467.2 0.0597 2 0.9706
B.0.3 3. Task 3 - Novel segmentation
Logistic regression model:
MODEL INFO:
Observations:3146
Dependent Variable:responses
Type:Mixed effects generalized linear regression
Error Distribution:binomial
Link function:logit
MODEL FIT:
AIC= 2665.30, BIC = 2834.81
Pseudo-R² (fixed effects)= 0.37
Pseudo-R² (total)= 0.74
FIXED EFFECTS:
------------------------------------------------------------------
Est. S.E. z val. p
----------------------------------- ------- ------ -------- ------
(Intercept) -1.15 0.33 -3.47 0.00
conditionsNA1 0.46 0.50 0.92 0.36
conditionsMix 0.99 0.49 1.99 0.05
trialsitemA -1.72 0.34 -5.11 0.00
trialsitemBL_A 4.04 0.51 7.90 0.00
trialsitemBL_NA 3.51 0.44 7.93 0.00
conditionsNA1:trialsitemA -0.04 0.52 -0.08 0.93
conditionsMix:trialsitemA 0.48 0.50 0.96 0.34
conditionsNA1:trialsitemBL_A -0.19 0.77 -0.25 0.80
conditionsMix:trialsitemBL_A -0.83 0.77 -1.08 0.28
conditionsNA1:trialsitemBL_NA -1.19 0.67 -1.79 0.07
conditionsMix:trialsitemBL_NA -1.04 0.66 -1.57 0.12
------------------------------------------------------------------
Emmeans:
| contrast | trial_type | estimate | SE | z.ratio | p.value | |
|---|---|---|---|---|---|---|
| A1 - NA1 | itemNonA | -0.46 | 0.50 | -0.92 | 1.00 | |
| itemA | -0.42 | 0.63 | -0.66 | 1.00 | ||
| itemBL_A | -0.27 | 0.50 | -0.53 | 1.00 | ||
| itemBL_NA | 0.73 | 0.35 | 2.11 | 0.10 | ||
| A1 - Mix | itemNonA | -0.99 | 0.49 | -1.99 | 0.14 | |
| itemA | -1.46 | 0.62 | -2.35 | 0.06 | ||
| itemBL_A | -0.16 | 0.49 | -0.31 | 1.00 | ||
| itemBL_NA | 0.06 | 0.35 | 0.16 | 1.00 | ||
| NA1 - Mix | itemNonA | -0.53 | 0.50 | -1.05 | 0.88 | |
| itemA | -1.05 | 0.62 | -1.68 | 0.28 | ||
| itemBL_A | 0.11 | 0.50 | 0.22 | 1.00 | ||
| itemBL_NA | -0.68 | 0.33 | -2.05 | 0.12 |
ChiSquare Anova test:
1 main effect (without condition) vs 2 main effects
Models:
model3_b3: responses ~ trials + (1 + trials | subject) + (1 + conditions | stimulus)
model4_b3: responses ~ conditions + trials + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model3_b3 20 2664.8 2785.8 -1312.4 2624.8
model4_b3 22 2661.7 2794.8 -1308.8 2617.7 7.121 2 0.02843 *
1 main effects (without item) vs 2 main effects
Models:
model2_b3: responses ~ conditions + (1 + trials | subject) + (1 + conditions | stimulus)
model4_b3: responses ~ conditions + trials + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model2_b3 19 2723.3 2838.3 -1342.7 2685.3
model4_b3 22 2661.7 2794.8 -1308.8 2617.7 67.675 3 1.343e-14 ***
2 main effects (without interaction) vs with interaction
Models:
model4_b3: responses ~ conditions + trials + (1 + trials | subject) + (1 + conditions | stimulus)
model5_b3: responses ~ conditions + trials + conditions * trials + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model4_b3 22 2661.7 2794.8 -1308.8 2617.7
model5_b3 28 2665.3 2834.8 -1304.7 2609.3 8.3503 6 0.2135
B.0.4 4. Task 2 and 3 - Both segmentation tasks
Logistic regression model:
formula5_b4 <- responses ~ conditions + trials_sw + conditions * trials_sw + familiarity + (1 + trials|subject) + (1+ conditions|stimulus)
model5_b4 <- glmer(formula5_b4, data = data_s4, control=glmerControl(optimizer = "optimx", calc.derivs = FALSE, optCtrl = list(method = "nlminb", starttests = FALSE, kkt = FALSE)), family = binomial(link = "logit"))
MODEL INFO:
Observations:6287
Dependent Variable:responses
Type:Mixed effects generalized linear regression
Error Distribution:binomial
Link function:logit
MODEL FIT:
AIC= 4970.83, BIC = 5132.73
FIXED EFFECTS:
-------------------------------------------------------------
Est. S.E. z val. p
------------------------------ ------- ------ -------- ------
(Intercept) 2.58 0.22 11.73 0.00
conditionsNA1 -0.13 0.27 -0.48 0.63
conditionsMix 0.22 0.28 0.77 0.44
trials_swV -4.23 0.40 -10.47 0.00
familiarityheld-out 0.12 0.25 0.46 0.65
familiaritynovel -0.14 0.17 -0.84 0.40
conditionsNA1:trials_swV 0.35 0.57 0.60 0.55
conditionsMix:trials_swV 0.53 0.59 0.90 0.37
-------------------------------------------------------------
ChiSquare Anova test:
2 main effects (without condition) vs 3 main effects
Models:
model3_b4: responses ~ trials_sw + familiarity + (1 + trials | subject) + (1 + conditions | stimulus)
model4_b4: responses ~ conditions + trials_sw + familiarity + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model3_b4 20 4969.4 5104.3 -2464.7 4929.4
model4_b4 22 4967.6 5116.1 -2461.8 4923.6 5.7478 2 0.05648 .
2 main effects (without item) vs 3 main effects
Models:
model2_b4: responses ~ conditions + familiarity + (1 + trials | subject) + (1 + conditions | stimulus)
model4_b4: responses ~ conditions + trials_sw + familiarity + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model2_b4 21 5058.0 5199.7 -2508.0 5016.0
model4_b4 22 4967.6 5116.1 -2461.8 4923.6 92.368 1 < 2.2e-16 ***
2 main effects (without familiarity) vs 3 main effects
Models:
model1_b4: responses ~ conditions + trials_sw + (1 + trials | subject) + (1 + conditions | stimulus)
model4_b4: responses ~ conditions + trials_sw + familiarity + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model1_b4 20 4964.9 5099.8 -2462.4 4924.9
model4_b4 22 4967.6 5116.1 -2461.8 4923.6 1.2368 2 0.5388
3 main effects (without interaction) vs with interaction
Models:
model4_b4: responses ~ conditions + trials_sw + familiarity + (1 + trials | subject) + (1 + conditions | stimulus)
model5_b4: responses ~ conditions + trials_sw + conditions * trials_sw + familiarity + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model4_b4 22 4967.6 5116.1 -2461.8 4923.6
model5_b4 24 4970.8 5132.7 -2461.4 4922.8 0.8177 2 0.6644
B.0.5 5. Analysis of the subset of successful learners
Task 1 Emmeans:
> pairs(emm6_b1, adjust = "bonferroni")
trial = itemNonA:
contrast estimate SE df z.ratio p.value
A1 - NA1 0.49606 0.203 Inf 2.447 0.0432
A1 - Mix -0.08251 0.179 Inf -0.461 1.0000
NA1 - Mix -0.57857 0.205 Inf -2.817 0.0145
trial = itemA:
contrast estimate SE df z.ratio p.value
A1 - NA1 0.31351 0.237 Inf 1.322 0.5588
A1 - Mix 0.33733 0.209 Inf 1.616 0.3181
NA1 - Mix 0.02382 0.233 Inf 0.102 1.0000
trial = itemBL_A:
contrast estimate SE df z.ratio p.value
A1 - NA1 0.20357 0.247 Inf 0.823 1.0000
A1 - Mix -0.03067 0.226 Inf -0.136 1.0000
NA1 - Mix -0.23424 0.251 Inf -0.933 1.0000
trial = itemBL_NA:
contrast estimate SE df z.ratio p.value
A1 - NA1 0.35611 0.285 Inf 1.249 0.6346
A1 - Mix 0.34814 0.261 Inf 1.332 0.5481
NA1 - Mix -0.00797 0.277 Inf -0.029 1.0000
Results are averaged over the levels of: familiarity
Results are given on the log odds ratio (not the response) scale.
P value adjustment: bonferroni method for 3 tests
Task2 Emmeans
> pairs(emm6_b2, adjust = "bonferroni")
trial = itemA:
contrast estimate SE df z.ratio p.value
A1 - NA1 1.170 0.936 Inf 1.249 0.6345
A1 - Mix 1.572 0.824 Inf 1.908 0.1691
NA1 - Mix 0.402 0.728 Inf 0.552 1.0000
trial = itemBL_A:
contrast estimate SE df z.ratio p.value
A1 - NA1 -0.674 0.916 Inf -0.736 1.0000
A1 - Mix -0.246 0.763 Inf -0.322 1.0000
NA1 - Mix 0.428 0.811 Inf 0.529 1.0000
Results are averaged over the levels of: familiarity
Results are given on the log odds ratio (not the response) scale.
P value adjustment: bonferroni method for 3 tests
Task3 Emmeans
> pairs(emm4_b3, adjust = "bonferroni")
trial = itemNonA:
contrast estimate SE df z.ratio p.value
A1 - NA1 -1.977 1.496 Inf -1.322 0.5583
A1 - Mix -1.547 1.300 Inf -1.191 0.7014
NA1 - Mix 0.430 1.341 Inf 0.321 1.0000
trial = itemA:
contrast estimate SE df z.ratio p.value
A1 - NA1 1.664 1.448 Inf 1.149 0.7511
A1 - Mix 1.361 1.367 Inf 0.995 0.9591
NA1 - Mix -0.304 0.976 Inf -0.311 1.0000
trial = itemBL_A:
contrast estimate SE df z.ratio p.value
A1 - NA1 -0.537 0.961 Inf -0.559 1.0000
A1 - Mix -1.207 0.933 Inf -1.294 0.5874
NA1 - Mix -0.670 0.820 Inf -0.817 1.0000
trial = itemBL_NA:
contrast estimate SE df z.ratio p.value
A1 - NA1 0.414 0.849 Inf 0.487 1.0000
A1 - Mix 0.149 0.785 Inf 0.189 1.0000
NA1 - Mix -0.265 0.593 Inf -0.448 1.0000
Results are given on the log odds ratio (not the response) scale.
P value adjustment: bonferroni method for 3 tests
B.0.6 6. Analysis of the segmentation accuracy
Logistic regression model:
Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) [glmerMod]
Family: binomial ( logit )
Formula: responses ~ conditions + trials + acc + (1 + trials | subject) +
(1 + conditions | stimulus)
Data: data_s1
Control:
glmerControl(optimizer = "optimx", calc.derivs = FALSE, optCtrl = list(method = "nlminb",
starttests = FALSE, kkt = FALSE))
AIC BIC logLik deviance df.resid
5775.2 5930.4 -2864.6 5729.2 6274
Scaled residuals:
Min 1Q Median 3Q Max
-6.9966 0.1751 0.3267 0.5181 1.9099
Random effects:
Groups Name Variance Std.Dev. Corr
subject (Intercept) 0.15000 0.3873
trialsitemA 0.34288 0.5856 -0.45
trialsitemBL_A 0.40546 0.6368 -0.26 0.89
trialsitemBL_NA 0.55866 0.7474 0.05 0.86 0.78
stimulus (Intercept) 0.96203 0.9808
conditionsNA1 0.09705 0.3115 -1.00
conditionsMix 0.02924 0.1710 -0.42 0.42
Number of obs: 6297, groups: subject, 198; stimulus, 32
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.26268 0.36796 -0.714 0.475293
conditionsNA1 -0.33948 0.12798 -2.653 0.007985 **
conditionsMix -0.01529 0.12426 -0.123 0.902038
trialsitemA 0.71396 0.36022 1.982 0.047475 *
trialsitemBL_A 1.23056 0.36911 3.334 0.000856 ***
trialsitemBL_NA 1.32407 0.37787 3.504 0.000458 ***
acc 2.10837 0.37946 5.556 2.76e-08 ***
---
Logistic regression model 2
Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) [glmerMod]
Family: binomial ( logit )
Formula: responses ~ conditions + trials + acc + conditions * acc + (1 +
trials | subject) + (1 + conditions | stimulus)
Data: data_s1
Control:
glmerControl(optimizer = "optimx", calc.derivs = FALSE, optCtrl = list(method = "nlminb",
starttests = FALSE, kkt = FALSE))
AIC BIC logLik deviance df.resid
5777.6 5946.3 -2863.8 5727.6 6272
Scaled residuals:
Min 1Q Median 3Q Max
-6.9312 0.1750 0.3264 0.5177 1.9665
Random effects:
Groups Name Variance Std.Dev. Corr
subject (Intercept) 0.14147 0.3761
trialsitemA 0.34087 0.5838 -0.43
trialsitemBL_A 0.40245 0.6344 -0.26 0.89
trialsitemBL_NA 0.56340 0.7506 0.07 0.87 0.78
stimulus (Intercept) 0.96835 0.9840
conditionsNA1 0.09998 0.3162 -1.00
conditionsMix 0.02967 0.1723 -0.44 0.44
Number of obs: 6297, groups: subject, 198; stimulus, 32
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.6725 0.4964 -1.355 0.175486
conditionsNA1 0.3385 0.5525 0.613 0.540177
conditionsMix 0.4919 0.5733 0.858 0.390861
trialsitemA 0.7132 0.3588 1.988 0.046819 *
trialsitemBL_A 1.2297 0.3677 3.344 0.000825 ***
trialsitemBL_NA 1.3280 0.3769 3.524 0.000426 ***
acc 2.8380 0.7062 4.019 5.84e-05 ***
conditionsNA1:acc -1.1952 0.9453 -1.264 0.206123
conditionsMix:acc -0.8875 0.9523 -0.932 0.351346
Logistic regression model 3
Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) [glmerMod]
Family: binomial ( logit )
Formula: responses ~ conditions + trials + acc + trials * acc + (1 + trials |
subject) + (1 + conditions | stimulus)
Data: data_s1
Control:
glmerControl(optimizer = "optimx", calc.derivs = FALSE, optCtrl = list(method = "nlminb",
starttests = FALSE, kkt = FALSE))
AIC BIC logLik deviance df.resid
5777.0 5952.4 -2862.5 5725.0 6271
Scaled residuals:
Min 1Q Median 3Q Max
-7.3655 0.1712 0.3245 0.5219 1.9387
Random effects:
Groups Name Variance Std.Dev. Corr
subject (Intercept) 0.13318 0.3649
trialsitemA 0.30791 0.5549 -0.38
trialsitemBL_A 0.38594 0.6212 -0.18 0.87
trialsitemBL_NA 0.53793 0.7334 0.13 0.86 0.76
stimulus (Intercept) 0.95614 0.9778
conditionsNA1 0.09517 0.3085 -1.00
conditionsMix 0.03293 0.1815 -0.35 0.35
Number of obs: 6297, groups: subject, 198; stimulus, 32
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 0.15567 0.41776 0.373 0.70942
conditionsNA1 -0.33908 0.12773 -2.655 0.00794 **
conditionsMix -0.01339 0.12495 -0.107 0.91464
trialsitemA -0.05066 0.57922 -0.087 0.93030
trialsitemBL_A 0.37156 0.61750 0.602 0.54737
trialsitemBL_NA 0.52164 0.65476 0.797 0.42563
acc 1.37824 0.50783 2.714 0.00665 **
trialsitemA:acc 1.33260 0.78463 1.698 0.08944 .
trialsitemBL_A:acc 1.49199 0.85787 1.739 0.08200 .
trialsitemBL_NA:acc 1.39314 0.92746 1.502 0.13307
2 main effects (without acc score) vs 3 main effects
Models:
model1_b1: responses ~ conditions + trials + (1 + trials | subject) + (1 + conditions | stimulus)
model9_b1: responses ~ conditions + trials + acc + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model1_b1 22 5799.4 5947.9 -2877.7 5755.4
model9_b1 23 5775.2 5930.4 -2864.6 5729.2 26.178 1 3.113e-07 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
3 main effects (without interaction) vs with interaction of accuracy score and trials
Models:
model9_b1: responses ~ conditions + trials + acc + (1 + trials | subject) + (1 + conditions | stimulus)
model8_b1: responses ~ conditions + trials + acc + trials * acc + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model9_b1 23 5775.2 5930.4 -2864.6 5729.2
model8_b1 26 5777.0 5952.4 -2862.5 5725.0 4.2351 3 0.2372
3 main effects (without interaction) vs with interaction of accuracy score and condition
Models:
model9_b1: responses ~ conditions + trials + acc + (1 + trials | subject) + (1 + conditions | stimulus)
model10_b1: responses ~ conditions + trials + acc + conditions * acc + (1 + trials | subject) + (1 + conditions | stimulus)
npar AIC BIC logLik deviance Chisq Df Pr(>Chisq)
model9_b1 23 5775.2 5930.4 -2864.6 5729.2
model10_b1 25 5777.6 5946.3 -2863.8 5727.6 1.6376 2 0.441