Quadrilateral learning experiments: Sequences
Kyle MacDonald and Michael Frank
January 22, 2016
Load data.
Descriptives
How many participants in each condition?
## Source: local data frame [2 x 2]
##
## training_condition n()
## (chr) (int)
## 1 active_active 20
## 2 passive_passive 20How long did each condition take?
## Source: local data frame [2 x 3]
##
## training_condition m_train_time m_exp_time
## (chr) (dbl) (dbl)
## 1 active_active 0.6112942 4.434711
## 2 passive_passive 0.4713317 3.475504
Visualization
Sanity check on pretest questions
Overall accuracy analysis
Overall accuracy analysis collapsing across entity and relational tests
Within subjects change scores
Accuracy on the learned shape
Within subjects change scores across learned shape questions
Models
Predict overall accuracy based on condition and test type
mm1 <- glmer(correct ~ training_condition * trial_type * block + (1|subids),
data = filter(d1, trial_type %in% c("relational", "entity")),
family=binomial,
nAGQ = 1,
control = glmerControl(optimizer = "bobyqa"))
summary(mm1)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: correct ~ training_condition * trial_type * block + (1 | subids)
## Data: filter(d1, trial_type %in% c("relational", "entity"))
## Control: glmerControl(optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 2906.4 2957.8 -1444.2 2888.4 2231
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.9014 -1.0126 0.5517 0.9632 1.0646
##
## Random effects:
## Groups Name Variance Std.Dev.
## subids (Intercept) 0 0
## Number of obs: 2240, groups: subids, 40
##
## Fixed effects:
## Estimate
## (Intercept) -0.1252
## training_conditionpassive_passive 0.1502
## trial_typerelational 1.4104
## blockpretest 0.1502
## training_conditionpassive_passive:trial_typerelational -0.8708
## training_conditionpassive_passive:blockpretest -0.1001
## trial_typerelational:blockpretest -0.2458
## training_conditionpassive_passive:trial_typerelational:blockpretest 0.5185
## Std. Error
## (Intercept) 0.1120
## training_conditionpassive_passive 0.1583
## trial_typerelational 0.1926
## blockpretest 0.1583
## training_conditionpassive_passive:trial_typerelational 0.2600
## training_conditionpassive_passive:blockpretest 0.2238
## trial_typerelational:blockpretest 0.2700
## training_conditionpassive_passive:trial_typerelational:blockpretest 0.3689
## z value
## (Intercept) -1.117
## training_conditionpassive_passive 0.949
## trial_typerelational 7.323
## blockpretest 0.949
## training_conditionpassive_passive:trial_typerelational -3.349
## training_conditionpassive_passive:blockpretest -0.447
## trial_typerelational:blockpretest -0.910
## training_conditionpassive_passive:trial_typerelational:blockpretest 1.406
## Pr(>|z|)
## (Intercept) 0.263849
## training_conditionpassive_passive 0.342718
## trial_typerelational 2.43e-13
## blockpretest 0.342736
## training_conditionpassive_passive:trial_typerelational 0.000811
## training_conditionpassive_passive:blockpretest 0.654515
## trial_typerelational:blockpretest 0.362624
## training_conditionpassive_passive:trial_typerelational:blockpretest 0.159852
##
## (Intercept)
## training_conditionpassive_passive
## trial_typerelational ***
## blockpretest
## training_conditionpassive_passive:trial_typerelational ***
## training_conditionpassive_passive:blockpretest
## trial_typerelational:blockpretest
## training_conditionpassive_passive:trial_typerelational:blockpretest
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) trnn__ trl_ty blckpr tr__:_ trn__: trl_t:
## trnng_cndt_ -0.708
## trl_typrltn -0.582 0.412
## blockpretst -0.708 0.501 0.412
## trnng_cn_:_ 0.431 -0.609 -0.741 -0.305
## trnng_cnd_: 0.501 -0.707 -0.291 -0.707 0.430
## trl_typrlt: 0.415 -0.294 -0.713 -0.586 0.528 0.415
## trnng_c_:_: -0.304 0.429 0.522 0.429 -0.705 -0.606 -0.732Predict accuracy for learned shape based on condition and test type
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: correct ~ training_condition * block * trial_type + (1 | subids)
## Data:
## filter(d_shape_learned, trial_type %in% c("relational", "entity"))
## Control: glmerControl(optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 1363.6 1408.2 -672.8 1345.6 1031
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.8559 -1.0742 0.6288 0.8660 1.0144
##
## Random effects:
## Groups Name Variance Std.Dev.
## subids (Intercept) 0 0
## Number of obs: 1040, groups: subids, 40
##
## Fixed effects:
## Estimate
## (Intercept) 0.143101
## training_conditionpassive_passive -0.171674
## blockpretest 0.144581
## trial_typerelational 1.093662
## training_conditionpassive_passive:blockpretest -0.001597
## training_conditionpassive_passive:trial_typerelational -0.409213
## blockpretest:trial_typerelational -0.453357
## training_conditionpassive_passive:blockpretest:trial_typerelational 0.462420
## Std. Error
## (Intercept) 0.169456
## training_conditionpassive_passive 0.239351
## blockpretest 0.240580
## trial_typerelational 0.276590
## training_conditionpassive_passive:blockpretest 0.339281
## training_conditionpassive_passive:trial_typerelational 0.376987
## blockpretest:trial_typerelational 0.383017
## training_conditionpassive_passive:blockpretest:trial_typerelational 0.529194
## z value
## (Intercept) 0.844
## training_conditionpassive_passive -0.717
## blockpretest 0.601
## trial_typerelational 3.954
## training_conditionpassive_passive:blockpretest -0.005
## training_conditionpassive_passive:trial_typerelational -1.085
## blockpretest:trial_typerelational -1.184
## training_conditionpassive_passive:blockpretest:trial_typerelational 0.874
## Pr(>|z|)
## (Intercept) 0.398
## training_conditionpassive_passive 0.473
## blockpretest 0.548
## trial_typerelational 7.68e-05
## training_conditionpassive_passive:blockpretest 0.996
## training_conditionpassive_passive:trial_typerelational 0.278
## blockpretest:trial_typerelational 0.237
## training_conditionpassive_passive:blockpretest:trial_typerelational 0.382
##
## (Intercept)
## training_conditionpassive_passive
## blockpretest
## trial_typerelational ***
## training_conditionpassive_passive:blockpretest
## training_conditionpassive_passive:trial_typerelational
## blockpretest:trial_typerelational
## training_conditionpassive_passive:blockpretest:trial_typerelational
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) trnn__ blckpr trl_ty trn__: tr__:_ blck:_
## trnng_cndt_ -0.708
## blockpretst -0.704 0.499
## trl_typrltn -0.613 0.434 0.432
## trnng_cnd_: 0.499 -0.705 -0.709 -0.306
## trnng_cn_:_ 0.449 -0.635 -0.317 -0.734 0.448
## blckprtst:_ 0.442 -0.313 -0.628 -0.722 0.445 0.530
## trnng_c_::_ -0.320 0.452 0.455 0.523 -0.641 -0.712 -0.724Predict change scores for learned shape based on condition and test type
## Linear mixed model fit by REML ['lmerMod']
## Formula: m_diff_score ~ training_condition * trial_type + (1 | subids)
## Data: ss_acc_learned
##
## REML criterion at convergence: -12.5
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.1828 -0.6269 0.1383 0.6534 2.5252
##
## Random effects:
## Groups Name Variance Std.Dev.
## subids (Intercept) 0.004229 0.06503
## Residual 0.038405 0.19597
## Number of obs: 80, groups: subids, 40
##
## Fixed effects:
## Estimate
## (Intercept) -3.571e-02
## training_conditionpassive_passive -1.371e-17
## trial_typerelational 9.405e-02
## training_conditionpassive_passive:trial_typerelational -9.167e-02
## Std. Error t value
## (Intercept) 4.617e-02 -0.774
## training_conditionpassive_passive 6.530e-02 0.000
## trial_typerelational 6.197e-02 1.518
## training_conditionpassive_passive:trial_typerelational 8.764e-02 -1.046
##
## Correlation of Fixed Effects:
## (Intr) trnn__ trl_ty
## trnng_cndt_ -0.707
## trl_typrltn -0.671 0.475
## trnng_cn_:_ 0.475 -0.671 -0.707