## `summarise()` regrouping output by 'partID', 'trial', 'pondnum', 'strategy' (override with `.groups` argument)
## `summarise()` regrouping output by 'partID', 'strategy', 'cue' (override with `.groups` argument)
1. Documents & Dataframes
Readin data. TSTM data read in, in hidden chunk above (cause it’s long).
vi <- read.csv('data_out/tstm_vi_outcomes_20210615.csv', header = TRUE, na.strings = NA)
dccs <- read.csv('data_out/tstm_dccs_outcomes_20210615.csv', header = TRUE, na.strings = NA)
indDiffs <- merge(vi, dccs, by = "Subject")
Merge.
str(indDiffs)
## 'data.frame': 99 obs. of 14 variables:
## $ Subject : int 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 ...
## $ X.x : int 2 3 4 5 6 7 8 9 10 11 ...
## $ Age : num 0 390 0 0 0 0 0 0 0 0 ...
## $ Gender : Factor w/ 2 levels "female","male": 1 2 1 1 2 2 1 2 1 1 ...
## $ VisInspTime : num 28.4 28.1 29.7 208.8 27.3 ...
## $ TrialCount : int 14 18 9 27 18 18 15 18 11 2 ...
## $ VisInspTime_log: num 3.35 3.33 3.39 5.34 3.31 ...
## $ X.y : int 1 2 3 4 5 6 7 8 9 10 ...
## $ nsw : num 6.78 7.22 6.63 6.93 6.69 ...
## $ single : num 6.21 6.18 5.82 6.19 6.07 ...
## $ start : num 6.44 6.72 6.18 6.51 6.49 ...
## $ sw : num 6.75 7.05 6.65 6.88 6.76 ...
## $ sc_local : num 0.0291 0.1727 -0.0256 0.0472 -0.0767 ...
## $ mc : num 0.562 1.044 0.805 0.737 0.615 ...
Exclude extra participants (3051 and 3050).
ex2 <- c(3055, 3063, 3068, 3069, 3076, 3117, 3121, 3120, 3032, 3051, 3050, 5398, 9999)
indDiffs <- indDiffs[!(indDiffs$Subject %in% ex2),]
Shrink dataset
indDiffs <- indDiffs[,c(1,7,13,14)]
colnames(indDiffs) <- c('partID','mSOA','sc_local','mc')
Merge with TSTM.
df <- merge(tstm_measures, indDiffs, by = "partID"); str(df)
## 'data.frame': 2784 obs. of 13 variables:
## $ partID : int 3001 3001 3001 3001 3001 3001 3001 3001 3001 3001 ...
## $ trial : int 10 8 9 5 12 3 21 22 23 11 ...
## $ strategy: Factor w/ 3 levels "maint","prac",..: 3 3 3 3 3 3 1 1 1 3 ...
## $ cue : Factor w/ 2 levels "False","True": 1 1 1 1 2 1 1 1 1 2 ...
## $ P1 : num 5 5 5 1 5 5 5 5 5 5 ...
## $ P2 : num 5 0 5 5 5 0 5 5 5 5 ...
## $ all : num 10 5 10 6 10 5 10 10 10 10 ...
## $ age_grp : Factor w/ 3 levels "a","s","t": 1 1 1 1 1 1 1 1 1 1 ...
## $ bin : int 1 0 1 0 1 0 1 1 1 1 ...
## $ id : int 8 6 7 3 2 1 3 4 5 1 ...
## $ mSOA : num 3.35 3.35 3.35 3.35 3.35 ...
## $ sc_local: num 0.0291 0.0291 0.0291 0.0291 0.0291 ...
## $ mc : num 0.562 0.562 0.562 0.562 0.562 ...
summary(df)
## partID trial strategy cue P1
## Min. :3001 Min. : 3.00 maint :1392 False:1392 Min. :0.000
## 1st Qu.:3023 1st Qu.:10.75 prac : 0 True :1392 1st Qu.:4.000
## Median :3072 Median :18.50 switch:1392 Median :5.000
## Mean :3063 Mean :18.50 Mean :4.054
## 3rd Qu.:3097 3rd Qu.:26.25 3rd Qu.:5.000
## Max. :3119 Max. :34.00 Max. :5.000
## P2 all age_grp bin id
## Min. :0.000 Min. : 0.000 a:1024 Min. :0.0000 Min. :1.00
## 1st Qu.:5.000 1st Qu.: 6.000 s: 928 1st Qu.:0.0000 1st Qu.:2.75
## Median :5.000 Median :10.000 t: 832 Median :1.0000 Median :4.50
## Mean :4.122 Mean : 8.176 Mean :0.5636 Mean :4.50
## 3rd Qu.:5.000 3rd Qu.:10.000 3rd Qu.:1.0000 3rd Qu.:6.25
## Max. :5.000 Max. :10.000 Max. :1.0000 Max. :8.00
## mSOA sc_local mc
## Min. :3.275 Min. :-0.5574095 Min. :0.2446
## 1st Qu.:3.459 1st Qu.:-0.1243016 1st Qu.:0.6221
## Median :3.907 Median :-0.0617504 Median :0.7406
## Mean :4.021 Mean :-0.0651224 Mean :0.7475
## 3rd Qu.:4.480 3rd Qu.: 0.0004621 3rd Qu.:0.8666
## Max. :5.823 Max. : 0.3452465 Max. :1.1925
2. Basics
2.1 Correlations
Hmisc::rcorr(as.matrix(df[,11:13]))
## mSOA sc_local mc
## mSOA 1.00 0.04 -0.10
## sc_local 0.04 1.00 0.06
## mc -0.10 0.06 1.00
##
## n= 2784
##
##
## P
## mSOA sc_local mc
## mSOA 0.0357 0.0000
## sc_local 0.0357 0.0025
## mc 0.0000 0.0025
2.2 Descriptives
df %>% group_by(age_grp) %>% summarize(avgAcc = mean(bin))
## `summarise()` ungrouping output (override with `.groups` argument)
## # A tibble: 3 x 2
## age_grp avgAcc
## <fct> <dbl>
## 1 a 0.741
## 2 s 0.327
## 3 t 0.609
psych::describe.by(df, group = "age_grp")
## Warning: describe.by is deprecated. Please use the describeBy function
##
## Descriptive statistics by group
## group: a
## vars n mean sd median trimmed mad min max range
## partID 1 1024 3017.59 9.85 3017.50 3017.50 11.86 3001.00 3035.00 34.00
## trial 2 1024 18.50 9.24 18.50 18.50 11.86 3.00 34.00 31.00
## strategy* 3 1024 2.00 1.00 2.00 2.00 1.48 1.00 3.00 2.00
## cue* 4 1024 1.50 0.50 1.50 1.50 0.74 1.00 2.00 1.00
## P1 5 1024 4.50 1.35 5.00 4.91 0.00 0.00 5.00 5.00
## P2 6 1024 4.46 1.45 5.00 4.90 0.00 0.00 5.00 5.00
## all 7 1024 8.96 1.85 10.00 9.32 0.00 3.00 10.00 7.00
## age_grp* 8 1024 1.00 0.00 1.00 1.00 0.00 1.00 1.00 0.00
## bin 9 1024 0.74 0.44 1.00 0.80 0.00 0.00 1.00 1.00
## id 10 1024 4.50 2.29 4.50 4.50 2.97 1.00 8.00 7.00
## mSOA 11 1024 3.68 0.48 3.51 3.59 0.28 3.28 5.34 2.07
## sc_local 12 1024 -0.03 0.08 -0.04 -0.03 0.07 -0.21 0.17 0.38
## mc 13 1024 0.74 0.20 0.73 0.73 0.17 0.24 1.17 0.93
## skew kurtosis se
## partID 0.05 -1.08 0.31
## trial 0.00 -1.21 0.29
## strategy* 0.00 -2.00 0.03
## cue* 0.00 -2.00 0.02
## P1 -2.47 4.40 0.04
## P2 -2.42 4.13 0.05
## all -1.32 -0.05 0.06
## age_grp* NaN NaN 0.00
## bin -1.10 -0.79 0.01
## id 0.00 -1.24 0.07
## mSOA 1.66 2.49 0.02
## sc_local 0.34 -0.05 0.00
## mc 0.12 0.07 0.01
## ------------------------------------------------------------
## group: s
## vars n mean sd median trimmed mad min max range
## partID 1 928 3085.17 16.25 3087.00 3085.44 13.34 3054.00 3116.00 62.00
## trial 2 928 18.50 9.24 18.50 18.50 11.86 3.00 34.00 31.00
## strategy* 3 928 2.00 1.00 2.00 2.00 1.48 1.00 3.00 2.00
## cue* 4 928 1.50 0.50 1.50 1.50 0.74 1.00 2.00 1.00
## P1 5 928 3.47 1.98 5.00 3.71 0.00 0.00 5.00 5.00
## P2 6 928 3.58 1.97 5.00 3.85 0.00 0.00 5.00 5.00
## all 7 928 7.05 2.46 6.00 7.19 1.48 0.00 10.00 10.00
## age_grp* 8 928 2.00 0.00 2.00 2.00 0.00 2.00 2.00 0.00
## bin 9 928 0.33 0.47 0.00 0.28 0.00 0.00 1.00 1.00
## id 10 928 4.50 2.29 4.50 4.50 2.97 1.00 8.00 7.00
## mSOA 11 928 4.49 0.57 4.44 4.47 0.52 3.40 5.82 2.42
## sc_local 12 928 -0.10 0.16 -0.10 -0.10 0.13 -0.56 0.35 0.90
## mc 13 928 0.74 0.21 0.75 0.74 0.23 0.36 1.19 0.84
## skew kurtosis se
## partID -0.14 -0.57 0.53
## trial 0.00 -1.21 0.30
## strategy* 0.00 -2.00 0.03
## cue* 0.00 -2.00 0.02
## P1 -0.75 -1.18 0.06
## P2 -0.87 -1.00 0.06
## all -0.25 -0.57 0.08
## age_grp* NaN NaN 0.00
## bin 0.74 -1.46 0.02
## id 0.00 -1.24 0.08
## mSOA 0.39 -0.44 0.02
## sc_local -0.15 1.85 0.01
## mc 0.11 -0.51 0.01
## ------------------------------------------------------------
## group: t
## vars n mean sd median trimmed mad min max range
## partID 1 832 3094.04 19.39 3100.50 3095.65 17.79 3053.00 3119.00 66.00
## trial 2 832 18.50 9.24 18.50 18.50 11.86 3.00 34.00 31.00
## strategy* 3 832 2.00 1.00 2.00 2.00 1.48 1.00 3.00 2.00
## cue* 4 832 1.50 0.50 1.50 1.50 0.74 1.00 2.00 1.00
## P1 5 832 4.16 1.66 5.00 4.50 0.00 0.00 5.00 5.00
## P2 6 832 4.31 1.57 5.00 4.73 0.00 0.00 5.00 5.00
## all 7 832 8.47 2.02 10.00 8.71 0.00 4.00 10.00 6.00
## age_grp* 8 832 3.00 0.00 3.00 3.00 0.00 3.00 3.00 0.00
## bin 9 832 0.61 0.49 1.00 0.64 0.00 0.00 1.00 1.00
## id 10 832 4.50 2.29 4.50 4.50 2.97 1.00 8.00 7.00
## mSOA 11 832 3.92 0.52 3.78 3.87 0.48 3.28 5.09 1.81
## sc_local 12 832 -0.07 0.11 -0.06 -0.07 0.09 -0.27 0.22 0.49
## mc 13 832 0.77 0.17 0.75 0.76 0.16 0.47 1.16 0.69
## skew kurtosis se
## partID -0.63 -0.89 0.67
## trial 0.00 -1.21 0.32
## strategy* 0.00 -2.00 0.03
## cue* 0.00 -2.00 0.02
## P1 -1.56 0.62 0.06
## P2 -1.99 2.24 0.05
## all -0.69 -1.35 0.07
## age_grp* NaN NaN 0.00
## bin -0.45 -1.80 0.02
## id 0.00 -1.24 0.08
## mSOA 0.69 -0.70 0.02
## sc_local 0.16 0.28 0.00
## mc 0.28 -0.44 0.01
3. Models
3.1 Processing Speed
3.1.1 7-yos
ps_7 <- glmer(bin ~ id*cue*strategy*mSOA + (1|partID),
family = binomial(link = "logit"),
control=glmerControl(optimizer="bobyqa", optCtrl=list(maxfun=1000000)),
data = df[df$age_grp == "s",])
## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?
ps_7
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: bin ~ id * cue * strategy * mSOA + (1 | partID)
## Data: df[df$age_grp == "s", ]
## AIC BIC logLik deviance df.resid
## 741.7276 823.8892 -353.8638 707.7276 911
## Random effects:
## Groups Name Std.Dev.
## partID (Intercept) 1.294
## Number of obs: 928, groups: partID, 29
## Fixed Effects:
## (Intercept) id
## -6.5305 2.2541
## cueTrue strategyswitch
## 17.9581 9.5346
## mSOA id:cueTrue
## 0.7670 -3.2252
## id:strategyswitch cueTrue:strategyswitch
## -2.0127 -15.5918
## id:mSOA cueTrue:mSOA
## -0.4151 -3.1398
## strategyswitch:mSOA id:cueTrue:strategyswitch
## -2.7983 2.9631
## id:cueTrue:mSOA id:strategyswitch:mSOA
## 0.6671 0.4650
## cueTrue:strategyswitch:mSOA id:cueTrue:strategyswitch:mSOA
## 3.4770 -0.6962
## convergence code 0; 0 optimizer warnings; 1 lme4 warnings
3.1.1.1 LR Chi Sqr
car::Anova(ps_7, )
## Analysis of Deviance Table (Type II Wald chisquare tests)
##
## Response: bin
## Chisq Df Pr(>Chisq)
## id 23.2993 1 1.387e-06 ***
## cue 100.0194 1 < 2.2e-16 ***
## strategy 157.2618 1 < 2.2e-16 ***
## mSOA 7.3322 1 0.006773 **
## id:cue 7.0659 1 0.007857 **
## id:strategy 0.4310 1 0.511476
## cue:strategy 3.6076 1 0.057516 .
## id:mSOA 0.0933 1 0.759995
## cue:mSOA 0.0002 1 0.988130
## strategy:mSOA 0.3493 1 0.554486
## id:cue:strategy 0.0056 1 0.940416
## id:cue:mSOA 6.4850 1 0.010879 *
## id:strategy:mSOA 0.0648 1 0.799054
## cue:strategy:mSOA 0.1332 1 0.715154
## id:cue:strategy:mSOA 2.6521 1 0.103416
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
3.1.1.2 Exponentiated Coeff
jtools::summ(ps_7)
|
Observations
|
928
|
|
Dependent variable
|
bin
|
|
Type
|
Mixed effects generalized linear model
|
|
Family
|
binomial
|
|
Link
|
logit
|
|
AIC
|
741.73
|
|
BIC
|
823.89
|
|
Pseudo-R² (fixed effects)
|
0.51
|
|
Pseudo-R² (total)
|
0.68
|
|
Fixed Effects
|
|
|
Est.
|
S.E.
|
z val.
|
p
|
|
(Intercept)
|
-6.53
|
4.29
|
-1.52
|
0.13
|
|
id
|
2.25
|
0.73
|
3.09
|
0.00
|
|
cueTrue
|
17.96
|
5.17
|
3.47
|
0.00
|
|
strategyswitch
|
9.53
|
9.59
|
0.99
|
0.32
|
|
mSOA
|
0.77
|
0.94
|
0.81
|
0.42
|
|
id:cueTrue
|
-3.23
|
1.00
|
-3.21
|
0.00
|
|
id:strategyswitch
|
-2.01
|
1.56
|
-1.29
|
0.20
|
|
cueTrue:strategyswitch
|
-15.59
|
10.78
|
-1.45
|
0.15
|
|
id:mSOA
|
-0.42
|
0.16
|
-2.58
|
0.01
|
|
cueTrue:mSOA
|
-3.14
|
1.13
|
-2.78
|
0.01
|
|
strategyswitch:mSOA
|
-2.80
|
2.28
|
-1.23
|
0.22
|
|
id:cueTrue:strategyswitch
|
2.96
|
1.84
|
1.61
|
0.11
|
|
id:cueTrue:mSOA
|
0.67
|
0.22
|
3.02
|
0.00
|
|
id:strategyswitch:mSOA
|
0.47
|
0.37
|
1.27
|
0.20
|
|
cueTrue:strategyswitch:mSOA
|
3.48
|
2.53
|
1.37
|
0.17
|
|
id:cueTrue:strategyswitch:mSOA
|
-0.70
|
0.43
|
-1.63
|
0.10
|
|
Random Effects
|
|
Group
|
Parameter
|
Std. Dev.
|
|
partID
|
(Intercept)
|
1.29
|
|
Grouping Variables
|
|
Group
|
# groups
|
ICC
|
|
partID
|
29
|
0.34
|
3.1.1.3 Visualizations
plot_model(ps_7, type = "pred", terms = c("mSOA [all]"))
plot_model(ps_7, type = "pred", terms = c("mSOA [all]","cue","id"))
3.1.1.4 Pairwise
emmeans::emtrends(ps_7, pairwise ~ cue, var = "mSOA")
## NOTE: Results may be misleading due to involvement in interactions
## $emtrends
## cue mSOA.trend SE df asymp.LCL asymp.UCL
## False -1.45 0.643 Inf -2.71 -0.194
## True -1.42 0.497 Inf -2.39 -0.444
##
## Results are averaged over the levels of: strategy
## Confidence level used: 0.95
##
## $contrasts
## contrast estimate SE df z.ratio p.value
## False - True -0.0343 0.524 Inf -0.065 0.9478
##
## Results are averaged over the levels of: strategy
3.1.2 10-yos
ps_10 <- glmer(bin ~ id*cue*strategy*mSOA + (1|partID),
family = binomial(link = "logit"),
control=glmerControl(optimizer="bobyqa", optCtrl=list(maxfun=1000000)),
data = df[df$age_grp == "t",])
## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?
ps_10
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: bin ~ id * cue * strategy * mSOA + (1 | partID)
## Data: df[df$age_grp == "t", ]
## AIC BIC logLik deviance df.resid
## 712.6911 792.9963 -339.3456 678.6911 815
## Random effects:
## Groups Name Std.Dev.
## partID (Intercept) 1.317
## Number of obs: 832, groups: partID, 26
## Fixed Effects:
## (Intercept) id
## -2.346212 0.085210
## cueTrue strategyswitch
## 4.528889 -6.257927
## mSOA id:cueTrue
## -0.812763 -2.535200
## id:strategyswitch cueTrue:strategyswitch
## 1.682085 5.497196
## id:mSOA cueTrue:mSOA
## 0.270871 0.571290
## strategyswitch:mSOA id:cueTrue:strategyswitch
## 2.232540 0.208517
## id:cueTrue:mSOA id:strategyswitch:mSOA
## 0.472036 -0.597406
## cueTrue:strategyswitch:mSOA id:cueTrue:strategyswitch:mSOA
## -2.063920 0.002702
## convergence code 0; 0 optimizer warnings; 1 lme4 warnings
3.1.2.1 LR Chi Sqr
car::Anova(ps_10)
## Analysis of Deviance Table (Type II Wald chisquare tests)
##
## Response: bin
## Chisq Df Pr(>Chisq)
## id 40.9160 1 1.589e-10 ***
## cue 94.6257 1 < 2.2e-16 ***
## strategy 12.5563 1 0.0003949 ***
## mSOA 0.4291 1 0.5124305
## id:cue 29.5847 1 5.352e-08 ***
## id:strategy 14.2209 1 0.0001626 ***
## cue:strategy 6.4956 1 0.0108140 *
## id:mSOA 0.0532 1 0.8175991
## cue:mSOA 4.7463 1 0.0293619 *
## strategy:mSOA 3.4300 1 0.0640229 .
## id:cue:strategy 0.8932 1 0.3446033
## id:cue:mSOA 5.6540 1 0.0174155 *
## id:strategy:mSOA 5.4230 1 0.0198727 *
## cue:strategy:mSOA 3.2657 1 0.0707436 .
## id:cue:strategy:mSOA 0.0000 1 0.9958814
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
3.1.2.2 Exponentiated Coeff
jtools::summ(ps_10)
|
Observations
|
832
|
|
Dependent variable
|
bin
|
|
Type
|
Mixed effects generalized linear model
|
|
Family
|
binomial
|
|
Link
|
logit
|
|
AIC
|
712.69
|
|
BIC
|
793.00
|
|
Pseudo-R² (fixed effects)
|
0.54
|
|
Pseudo-R² (total)
|
0.70
|
|
Fixed Effects
|
|
|
Est.
|
S.E.
|
z val.
|
p
|
|
(Intercept)
|
-2.35
|
5.94
|
-0.39
|
0.69
|
|
id
|
0.09
|
1.11
|
0.08
|
0.94
|
|
cueTrue
|
4.53
|
6.98
|
0.65
|
0.52
|
|
strategyswitch
|
-6.26
|
6.51
|
-0.96
|
0.34
|
|
mSOA
|
-0.81
|
1.53
|
-0.53
|
0.60
|
|
id:cueTrue
|
-2.54
|
1.74
|
-1.45
|
0.15
|
|
id:strategyswitch
|
1.68
|
1.27
|
1.32
|
0.19
|
|
cueTrue:strategyswitch
|
5.50
|
8.20
|
0.67
|
0.50
|
|
id:mSOA
|
0.27
|
0.29
|
0.93
|
0.35
|
|
cueTrue:mSOA
|
0.57
|
1.83
|
0.31
|
0.75
|
|
strategyswitch:mSOA
|
2.23
|
1.67
|
1.34
|
0.18
|
|
id:cueTrue:strategyswitch
|
0.21
|
1.94
|
0.11
|
0.91
|
|
id:cueTrue:mSOA
|
0.47
|
0.48
|
0.99
|
0.32
|
|
id:strategyswitch:mSOA
|
-0.60
|
0.33
|
-1.82
|
0.07
|
|
cueTrue:strategyswitch:mSOA
|
-2.06
|
2.12
|
-0.97
|
0.33
|
|
id:cueTrue:strategyswitch:mSOA
|
0.00
|
0.52
|
0.01
|
1.00
|
|
Random Effects
|
|
Group
|
Parameter
|
Std. Dev.
|
|
partID
|
(Intercept)
|
1.32
|
|
Grouping Variables
|
|
Group
|
# groups
|
ICC
|
|
partID
|
26
|
0.35
|
3.1.2.3 Visualizations
plot_model(ps_10, type = "pred", terms = c("mSOA [all]","cue"))
plot_model(ps_10, type = "pred", terms = c("mSOA [all]","cue","id"))
plot_model(ps_7, type = "pred", terms = c("mSOA [all]","strategy","id"))
3.1.2.4 Pairwise
emmeans::emtrends(ps_10, pairwise ~ cue, var = "mSOA")
## NOTE: Results may be misleading due to involvement in interactions
## $emtrends
## cue mSOA.trend SE df asymp.LCL asymp.UCL
## False 0.178 0.573 Inf -0.945 1.30
## True 1.848 0.766 Inf 0.347 3.35
##
## Results are averaged over the levels of: strategy
## Confidence level used: 0.95
##
## $contrasts
## contrast estimate SE df z.ratio p.value
## False - True -1.67 0.649 Inf -2.572 0.0101
##
## Results are averaged over the levels of: strategy
emmeans::emtrends(ps_10, pairwise ~ strategy, var = "mSOA")
## NOTE: Results may be misleading due to involvement in interactions
## $emtrends
## strategy mSOA.trend SE df asymp.LCL asymp.UCL
## maint 1.754 0.775 Inf 0.236 3.27
## switch 0.272 0.556 Inf -0.817 1.36
##
## Results are averaged over the levels of: cue
## Confidence level used: 0.95
##
## $contrasts
## contrast estimate SE df z.ratio p.value
## maint - switch 1.48 0.639 Inf 2.319 0.0204
##
## Results are averaged over the levels of: cue
3.1.3 Adults
ps_adult <- glmer(bin ~ id*cue*strategy*mSOA + (1|partID),
family = binomial(link = "logit"),
control=glmerControl(optimizer="bobyqa", optCtrl=list(maxfun=1000000)),
data = df[df$age_grp == "a",])
## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?
ps_adult
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: bin ~ id * cue * strategy * mSOA + (1 | partID)
## Data: df[df$age_grp == "a", ]
## AIC BIC logLik deviance df.resid
## 699.0617 782.8967 -332.5308 665.0617 1007
## Random effects:
## Groups Name Std.Dev.
## partID (Intercept) 1.366
## Number of obs: 1024, groups: partID, 32
## Fixed Effects:
## (Intercept) id
## -0.53945 0.70911
## cueTrue strategyswitch
## -8.46010 -4.79045
## mSOA id:cueTrue
## -0.51157 1.82279
## id:strategyswitch cueTrue:strategyswitch
## 0.46159 18.64662
## id:mSOA cueTrue:mSOA
## 0.05954 3.90584
## strategyswitch:mSOA id:cueTrue:strategyswitch
## 1.01492 -2.82711
## id:cueTrue:mSOA id:strategyswitch:mSOA
## -0.65114 -0.19638
## cueTrue:strategyswitch:mSOA id:cueTrue:strategyswitch:mSOA
## -5.40730 0.81518
## convergence code 0; 0 optimizer warnings; 1 lme4 warnings
3.1.3.1 LR Chi Sqr
car::Anova(ps_adult)
## Analysis of Deviance Table (Type II Wald chisquare tests)
##
## Response: bin
## Chisq Df Pr(>Chisq)
## id 98.6081 1 < 2.2e-16 ***
## cue 131.0947 1 < 2.2e-16 ***
## strategy 67.5208 1 < 2.2e-16 ***
## mSOA 0.5340 1 0.4649161
## id:cue 13.8075 1 0.0002025 ***
## id:strategy 3.5246 1 0.0604650 .
## cue:strategy 0.1903 1 0.6626407
## id:mSOA 0.3108 1 0.5772183
## cue:mSOA 1.2804 1 0.2578324
## strategy:mSOA 0.0013 1 0.9716313
## id:cue:strategy 0.3219 1 0.5704705
## id:cue:mSOA 0.0486 1 0.8254369
## id:strategy:mSOA 0.0017 1 0.9673948
## cue:strategy:mSOA 1.7884 1 0.1811253
## id:cue:strategy:mSOA 1.9218 1 0.1656552
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
3.1.3.2 Exponentiated Coeff
jtools::summ(ps_adult)
|
Observations
|
1024
|
|
Dependent variable
|
bin
|
|
Type
|
Mixed effects generalized linear model
|
|
Family
|
binomial
|
|
Link
|
logit
|
|
AIC
|
699.06
|
|
BIC
|
782.90
|
|
Pseudo-R² (fixed effects)
|
0.54
|
|
Pseudo-R² (total)
|
0.71
|
|
Fixed Effects
|
|
|
Est.
|
S.E.
|
z val.
|
p
|
|
(Intercept)
|
-0.54
|
3.83
|
-0.14
|
0.89
|
|
id
|
0.71
|
0.87
|
0.81
|
0.42
|
|
cueTrue
|
-8.46
|
9.48
|
-0.89
|
0.37
|
|
strategyswitch
|
-4.79
|
4.59
|
-1.04
|
0.30
|
|
mSOA
|
-0.51
|
1.04
|
-0.49
|
0.62
|
|
id:cueTrue
|
1.82
|
2.01
|
0.91
|
0.37
|
|
id:strategyswitch
|
0.46
|
1.04
|
0.44
|
0.66
|
|
cueTrue:strategyswitch
|
18.65
|
10.32
|
1.81
|
0.07
|
|
id:mSOA
|
0.06
|
0.24
|
0.25
|
0.80
|
|
cueTrue:mSOA
|
3.91
|
2.69
|
1.45
|
0.15
|
|
strategyswitch:mSOA
|
1.01
|
1.23
|
0.82
|
0.41
|
|
id:cueTrue:strategyswitch
|
-2.83
|
2.18
|
-1.30
|
0.19
|
|
id:cueTrue:mSOA
|
-0.65
|
0.55
|
-1.19
|
0.23
|
|
id:strategyswitch:mSOA
|
-0.20
|
0.28
|
-0.70
|
0.48
|
|
cueTrue:strategyswitch:mSOA
|
-5.41
|
2.90
|
-1.87
|
0.06
|
|
id:cueTrue:strategyswitch:mSOA
|
0.82
|
0.59
|
1.39
|
0.17
|
|
Random Effects
|
|
Group
|
Parameter
|
Std. Dev.
|
|
partID
|
(Intercept)
|
1.37
|
|
Grouping Variables
|
|
Group
|
# groups
|
ICC
|
|
partID
|
32
|
0.36
|
3.2 Local Switch Costs
dumb <- lme4::glmer(bin ~ id*cue*strategy*sc_local + (1|partID),
family = binomial(link = "logit"),
control=glmerControl(optimizer="bobyqa", optCtrl=list(maxfun=1000000)),
data = df[df$age_grp == "s",])
3.2.1 7-yos
lsc_7 <- glmer(bin ~ id*cue*strategy*sc_local + (1|partID),
family = binomial(link = "logit"),
control=glmerControl(optimizer="bobyqa", optCtrl=list(maxfun=1000000)),
data = df[df$age_grp == "s",])
lsc_7
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: bin ~ id * cue * strategy * sc_local + (1 | partID)
## Data: df[df$age_grp == "s", ]
## AIC BIC logLik deviance df.resid
## 740.1232 822.2848 -353.0616 706.1232 911
## Random effects:
## Groups Name Std.Dev.
## partID (Intercept) 1.25
## Number of obs: 928, groups: partID, 29
## Fixed Effects:
## (Intercept) id
## -2.5751 0.2197
## cueTrue strategyswitch
## 2.8909 -3.8862
## sc_local id:cueTrue
## 4.0991 -0.0932
## id:strategyswitch cueTrue:strategyswitch
## 0.3317 1.5090
## id:sc_local cueTrue:sc_local
## -1.5318 -7.1346
## strategyswitch:sc_local id:cueTrue:strategyswitch
## -10.3291 -0.4626
## id:cueTrue:sc_local id:strategyswitch:sc_local
## 0.6604 2.4922
## cueTrue:strategyswitch:sc_local id:cueTrue:strategyswitch:sc_local
## 13.1226 -2.1691
3.2.1.1 LR Chi Sqr
car::Anova(lsc_7)
## Analysis of Deviance Table (Type II Wald chisquare tests)
##
## Response: bin
## Chisq Df Pr(>Chisq)
## id 23.4713 1 1.268e-06 ***
## cue 101.4493 1 < 2.2e-16 ***
## strategy 147.3795 1 < 2.2e-16 ***
## sc_local 5.9950 1 0.01435 *
## id:cue 4.5718 1 0.03250 *
## id:strategy 1.1660 1 0.28022
## cue:strategy 6.1232 1 0.01334 *
## id:sc_local 5.5991 1 0.01797 *
## cue:sc_local 4.7503 1 0.02929 *
## strategy:sc_local 5.4544 1 0.01952 *
## id:cue:strategy 0.8872 1 0.34625
## id:cue:sc_local 0.0147 1 0.90347
## id:strategy:sc_local 1.2631 1 0.26107
## cue:strategy:sc_local 0.0940 1 0.75914
## id:cue:strategy:sc_local 1.3900 1 0.23841
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
3.2.1.2 Exponentiated Coeff
jtools::summ(lsc_7)
|
Observations
|
928
|
|
Dependent variable
|
bin
|
|
Type
|
Mixed effects generalized linear model
|
|
Family
|
binomial
|
|
Link
|
logit
|
|
AIC
|
740.12
|
|
BIC
|
822.28
|
|
Pseudo-R² (fixed effects)
|
0.50
|
|
Pseudo-R² (total)
|
0.66
|
|
Fixed Effects
|
|
|
Est.
|
S.E.
|
z val.
|
p
|
|
(Intercept)
|
-2.58
|
0.64
|
-4.01
|
0.00
|
|
id
|
0.22
|
0.11
|
2.05
|
0.04
|
|
cueTrue
|
2.89
|
0.73
|
3.97
|
0.00
|
|
strategyswitch
|
-3.89
|
1.74
|
-2.23
|
0.03
|
|
sc_local
|
4.10
|
3.95
|
1.04
|
0.30
|
|
id:cueTrue
|
-0.09
|
0.14
|
-0.66
|
0.51
|
|
id:strategyswitch
|
0.33
|
0.27
|
1.23
|
0.22
|
|
cueTrue:strategyswitch
|
1.51
|
1.88
|
0.80
|
0.42
|
|
id:sc_local
|
-1.53
|
0.66
|
-2.33
|
0.02
|
|
cueTrue:sc_local
|
-7.13
|
4.53
|
-1.58
|
0.12
|
|
strategyswitch:sc_local
|
-10.33
|
9.94
|
-1.04
|
0.30
|
|
id:cueTrue:strategyswitch
|
-0.46
|
0.31
|
-1.50
|
0.13
|
|
id:cueTrue:sc_local
|
0.66
|
0.92
|
0.72
|
0.47
|
|
id:strategyswitch:sc_local
|
2.49
|
1.58
|
1.58
|
0.11
|
|
cueTrue:strategyswitch:sc_local
|
13.12
|
10.87
|
1.21
|
0.23
|
|
id:cueTrue:strategyswitch:sc_local
|
-2.17
|
1.84
|
-1.18
|
0.24
|
|
Random Effects
|
|
Group
|
Parameter
|
Std. Dev.
|
|
partID
|
(Intercept)
|
1.25
|
|
Grouping Variables
|
|
Group
|
# groups
|
ICC
|
|
partID
|
29
|
0.32
|
3.2.1.3 Visualizations
plot_model(lsc_7, type = "pred", terms = c("sc_local [all]"))
plot_model(lsc_7, type = "pred", terms = c("sc_local [all]","cue"))
plot_model(lsc_7, type = "pred", terms = c("sc_local [all]", "strategy"))
NOTE: Cue included in this figure for display purposes - no 3way indicated.
plot_model(lsc_7, type = "pred", terms = c("sc_local [all]", "cue", "id"))
3.2.1.4 Pairwise
emmeans::emtrends(lsc_7, pairwise ~ cue, var = "sc_local")
## NOTE: Results may be misleading due to involvement in interactions
## $emtrends
## cue sc_local.trend SE df asymp.LCL asymp.UCL
## False -2.35 2.50 Inf -7.25 2.55
## True -4.83 1.83 Inf -8.41 -1.25
##
## Results are averaged over the levels of: strategy
## Confidence level used: 0.95
##
## $contrasts
## contrast estimate SE df z.ratio p.value
## False - True 2.48 2.25 Inf 1.103 0.2701
##
## Results are averaged over the levels of: strategy
emmeans::emtrends(lsc_7, pairwise ~ strategy, var = "sc_local")
## NOTE: Results may be misleading due to involvement in interactions
## $emtrends
## strategy sc_local.trend SE df asymp.LCL asymp.UCL
## maint -4.88 1.80 Inf -8.40 -1.35
## switch -2.31 2.54 Inf -7.28 2.66
##
## Results are averaged over the levels of: cue
## Confidence level used: 0.95
##
## $contrasts
## contrast estimate SE df z.ratio p.value
## maint - switch -2.57 2.29 Inf -1.122 0.2618
##
## Results are averaged over the levels of: cue
3.2.2 10-yos
lsc_10 <- glmer(bin ~ id*cue*strategy*sc_local + (1|partID),
family = binomial(link = "logit"),
control=glmerControl(optimizer="bobyqa", optCtrl=list(maxfun=1000000)),
data = df[df$age_grp == "t",])
lsc_10
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: bin ~ id * cue * strategy * sc_local + (1 | partID)
## Data: df[df$age_grp == "t", ]
## AIC BIC logLik deviance df.resid
## 702.9959 783.3010 -334.4979 668.9959 815
## Random effects:
## Groups Name Std.Dev.
## partID (Intercept) 1.399
## Number of obs: 832, groups: partID, 26
## Fixed Effects:
## (Intercept) id
## -5.61061 1.15808
## cueTrue strategyswitch
## 6.98198 2.68702
## sc_local id:cueTrue
## -0.39519 -0.84694
## id:strategyswitch cueTrue:strategyswitch
## -0.56024 -2.48356
## id:sc_local cueTrue:sc_local
## 0.06202 -5.41512
## strategyswitch:sc_local id:cueTrue:strategyswitch
## 4.04833 0.27226
## id:cueTrue:sc_local id:strategyswitch:sc_local
## 1.62713 1.50034
## cueTrue:strategyswitch:sc_local id:cueTrue:strategyswitch:sc_local
## 8.52716 -3.13651
3.2.2.1 LR Chi Sqr
car::Anova(lsc_10)
## Analysis of Deviance Table (Type II Wald chisquare tests)
##
## Response: bin
## Chisq Df Pr(>Chisq)
## id 44.1495 1 3.042e-11 ***
## cue 104.8497 1 < 2.2e-16 ***
## strategy 14.2477 1 0.0001603 ***
## sc_local 2.7377 1 0.0980042 .
## id:cue 37.9700 1 7.184e-10 ***
## id:strategy 13.0832 1 0.0002980 ***
## cue:strategy 3.7669 1 0.0522759 .
## id:sc_local 2.7938 1 0.0946294 .
## cue:sc_local 0.3848 1 0.5350443
## strategy:sc_local 15.9864 1 6.380e-05 ***
## id:cue:strategy 5.1857 1 0.0227741 *
## id:cue:sc_local 0.1112 1 0.7387513
## id:strategy:sc_local 0.1579 1 0.6911375
## cue:strategy:sc_local 1.9920 1 0.1581370
## id:cue:strategy:sc_local 2.4759 1 0.1156046
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
3.2.2.2 Exponentiated Coeff
jtools::summ(lsc_10)
|
Observations
|
832
|
|
Dependent variable
|
bin
|
|
Type
|
Mixed effects generalized linear model
|
|
Family
|
binomial
|
|
Link
|
logit
|
|
AIC
|
703.00
|
|
BIC
|
783.30
|
|
Pseudo-R² (fixed effects)
|
0.47
|
|
Pseudo-R² (total)
|
0.67
|
|
Fixed Effects
|
|
|
Est.
|
S.E.
|
z val.
|
p
|
|
(Intercept)
|
-5.61
|
0.89
|
-6.29
|
0.00
|
|
id
|
1.16
|
0.16
|
7.20
|
0.00
|
|
cueTrue
|
6.98
|
1.02
|
6.86
|
0.00
|
|
strategyswitch
|
2.69
|
0.96
|
2.80
|
0.01
|
|
sc_local
|
-0.40
|
7.05
|
-0.06
|
0.96
|
|
id:cueTrue
|
-0.85
|
0.21
|
-4.01
|
0.00
|
|
id:strategyswitch
|
-0.56
|
0.19
|
-2.98
|
0.00
|
|
cueTrue:strategyswitch
|
-2.48
|
1.21
|
-2.05
|
0.04
|
|
id:sc_local
|
0.06
|
1.22
|
0.05
|
0.96
|
|
cueTrue:sc_local
|
-5.42
|
8.03
|
-0.67
|
0.50
|
|
strategyswitch:sc_local
|
4.05
|
7.97
|
0.51
|
0.61
|
|
id:cueTrue:strategyswitch
|
0.27
|
0.25
|
1.08
|
0.28
|
|
id:cueTrue:sc_local
|
1.63
|
1.57
|
1.04
|
0.30
|
|
id:strategyswitch:sc_local
|
1.50
|
1.54
|
0.97
|
0.33
|
|
cueTrue:strategyswitch:sc_local
|
8.53
|
10.04
|
0.85
|
0.40
|
|
id:cueTrue:strategyswitch:sc_local
|
-3.14
|
1.99
|
-1.57
|
0.12
|
|
Random Effects
|
|
Group
|
Parameter
|
Std. Dev.
|
|
partID
|
(Intercept)
|
1.40
|
|
Grouping Variables
|
|
Group
|
# groups
|
ICC
|
|
partID
|
26
|
0.37
|
3.2.2.3 Visualizations
plot_model(lsc_10, type = "pred", terms = c("sc_local [all]"))
plot_model(lsc_10, type = "pred", terms = c("sc_local [all]", "strategy"))
NOTE: Cue included in this figure for display purposes - no 3way indicated.
plot_model(lsc_10, type = "pred", terms = c("sc_local [all]", "cue","id"))
3.2.2.4 Pairwise
emmeans::emtrends(lsc_10, pairwise ~ cue, var = "sc_local")
## NOTE: Results may be misleading due to involvement in interactions
## $emtrends
## cue sc_local.trend SE df asymp.LCL asymp.UCL
## False 5.28 2.99 Inf -0.568 11.1
## True 4.40 2.95 Inf -1.394 10.2
##
## Results are averaged over the levels of: strategy
## Confidence level used: 0.95
##
## $contrasts
## contrast estimate SE df z.ratio p.value
## False - True 0.887 2.18 Inf 0.406 0.6847
##
## Results are averaged over the levels of: strategy
3.2.3 Adults
lsc_adult <- glmer(bin ~ id*cue*strategy*sc_local + (1|partID),
family = binomial(link = "logit"),
control=glmerControl(optimizer="bobyqa", optCtrl=list(maxfun=1000000)),
data = df[df$age_grp == "a",])
lsc_adult
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: bin ~ id * cue * strategy * sc_local + (1 | partID)
## Data: df[df$age_grp == "a", ]
## AIC BIC logLik deviance df.resid
## 694.3113 778.1463 -330.1556 660.3113 1007
## Random effects:
## Groups Name Std.Dev.
## partID (Intercept) 1.368
## Number of obs: 1024, groups: partID, 32
## Fixed Effects:
## (Intercept) id
## -2.60477 0.95968
## cueTrue strategyswitch
## 5.34456 -0.95080
## sc_local id:cueTrue
## -5.55058 -0.09754
## id:strategyswitch cueTrue:strategyswitch
## -0.30107 -0.75376
## id:sc_local cueTrue:sc_local
## 0.88938 5.71059
## strategyswitch:sc_local id:cueTrue:strategyswitch
## 4.23135 -0.28244
## id:cueTrue:sc_local id:strategyswitch:sc_local
## 3.89479 -1.55178
## cueTrue:strategyswitch:sc_local id:cueTrue:strategyswitch:sc_local
## -10.67330 -2.43307
3.2.3.1 LR Chi Sqr
car::Anova(lsc_adult)
## Analysis of Deviance Table (Type II Wald chisquare tests)
##
## Response: bin
## Chisq Df Pr(>Chisq)
## id 95.9972 1 < 2.2e-16 ***
## cue 123.4714 1 < 2.2e-16 ***
## strategy 61.9257 1 3.567e-15 ***
## sc_local 0.9157 1 0.3386126
## id:cue 13.5611 1 0.0002309 ***
## id:strategy 3.6725 1 0.0553160 .
## cue:strategy 0.1627 1 0.6866680
## id:sc_local 0.3705 1 0.5427499
## cue:sc_local 1.7901 1 0.1809146
## strategy:sc_local 1.9929 1 0.1580375
## id:cue:strategy 0.0410 1 0.8396212
## id:cue:sc_local 1.8016 1 0.1795204
## id:strategy:sc_local 1.9312 1 0.1646309
## cue:strategy:sc_local 4.7877 1 0.0286641 *
## id:cue:strategy:sc_local 0.4460 1 0.5042241
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
3.2.3.2 Exponentiated Coeff
jtools::summ(lsc_adult)
|
Observations
|
1024
|
|
Dependent variable
|
bin
|
|
Type
|
Mixed effects generalized linear model
|
|
Family
|
binomial
|
|
Link
|
logit
|
|
AIC
|
694.31
|
|
BIC
|
778.15
|
|
Pseudo-R² (fixed effects)
|
0.64
|
|
Pseudo-R² (total)
|
0.77
|
|
Fixed Effects
|
|
|
Est.
|
S.E.
|
z val.
|
p
|
|
(Intercept)
|
-2.60
|
0.53
|
-4.95
|
0.00
|
|
id
|
0.96
|
0.12
|
7.69
|
0.00
|
|
cueTrue
|
5.34
|
1.25
|
4.28
|
0.00
|
|
strategyswitch
|
-0.95
|
0.64
|
-1.49
|
0.14
|
|
sc_local
|
-5.55
|
5.83
|
-0.95
|
0.34
|
|
id:cueTrue
|
-0.10
|
0.45
|
-0.22
|
0.83
|
|
id:strategyswitch
|
-0.30
|
0.15
|
-2.05
|
0.04
|
|
cueTrue:strategyswitch
|
-0.75
|
1.37
|
-0.55
|
0.58
|
|
id:sc_local
|
0.89
|
1.44
|
0.62
|
0.54
|
|
cueTrue:sc_local
|
5.71
|
12.22
|
0.47
|
0.64
|
|
strategyswitch:sc_local
|
4.23
|
7.12
|
0.59
|
0.55
|
|
id:cueTrue:strategyswitch
|
-0.28
|
0.46
|
-0.61
|
0.54
|
|
id:cueTrue:sc_local
|
3.89
|
3.34
|
1.17
|
0.24
|
|
id:strategyswitch:sc_local
|
-1.55
|
1.72
|
-0.90
|
0.37
|
|
cueTrue:strategyswitch:sc_local
|
-10.67
|
14.11
|
-0.76
|
0.45
|
|
id:cueTrue:strategyswitch:sc_local
|
-2.43
|
3.64
|
-0.67
|
0.50
|
|
Random Effects
|
|
Group
|
Parameter
|
Std. Dev.
|
|
partID
|
(Intercept)
|
1.37
|
|
Grouping Variables
|
|
Group
|
# groups
|
ICC
|
|
partID
|
32
|
0.36
|
3.2.3.4 Pairwise
emmeans::emtrends(lsc_adult, pairwise ~ cue*strategy, var = "sc_local")
## NOTE: Results may be misleading due to involvement in interactions
## $emtrends
## cue strategy sc_local.trend SE df asymp.LCL asymp.UCL
## False maint -1.55 4.16 Inf -9.70 6.60
## True maint 21.69 8.99 Inf 4.07 39.31
## False switch -4.30 3.50 Inf -11.15 2.55
## True switch -2.69 3.98 Inf -10.48 5.11
##
## Confidence level used: 0.95
##
## $contrasts
## contrast estimate SE df z.ratio p.value
## False maint - True maint -23.24 9.04 Inf -2.571 0.0497
## False maint - False switch 2.75 3.62 Inf 0.759 0.8726
## False maint - True switch 1.14 3.98 Inf 0.286 0.9919
## True maint - False switch 25.99 8.76 Inf 2.967 0.0159
## True maint - True switch 24.37 9.00 Inf 2.708 0.0342
## False switch - True switch -1.62 3.42 Inf -0.472 0.9652
##
## P value adjustment: tukey method for comparing a family of 4 estimates
3.3 Mixing Costs
3.3.1 7-yos
mc_7 <- glmer(bin ~ id*cue*strategy*mc + (1|partID),
family = binomial(link = "logit"),
control=glmerControl(optimizer="bobyqa", optCtrl=list(maxfun=1000000)),
data = df[df$age_grp == "s",])
mc_7
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: bin ~ id * cue * strategy * mc + (1 | partID)
## Data: df[df$age_grp == "s", ]
## AIC BIC logLik deviance df.resid
## 750.5043 832.6659 -358.2522 716.5043 911
## Random effects:
## Groups Name Std.Dev.
## partID (Intercept) 1.414
## Number of obs: 928, groups: partID, 29
## Fixed Effects:
## (Intercept) id
## -5.78672 0.61961
## cueTrue strategyswitch
## 2.56441 -2.01823
## mc id:cueTrue
## 3.41903 0.10752
## id:strategyswitch cueTrue:strategyswitch
## 0.02954 -0.72525
## id:mc cueTrue:mc
## -0.26807 1.94394
## strategyswitch:mc id:cueTrue:strategyswitch
## -0.79241 -0.41199
## id:cueTrue:mc id:strategyswitch:mc
## -0.48430 -0.01210
## cueTrue:strategyswitch:mc id:cueTrue:strategyswitch:mc
## 0.30279 0.42211
3.3.1.1 LR Chi Sqr
car::Anova(mc_7)
## Analysis of Deviance Table (Type II Wald chisquare tests)
##
## Response: bin
## Chisq Df Pr(>Chisq)
## id 25.2713 1 4.981e-07 ***
## cue 103.2514 1 < 2.2e-16 ***
## strategy 156.9205 1 < 2.2e-16 ***
## mc 2.5166 1 0.112650
## id:cue 8.8466 1 0.002936 **
## id:strategy 0.1619 1 0.687410
## cue:strategy 4.3179 1 0.037714 *
## id:mc 4.2521 1 0.039203 *
## cue:mc 0.0555 1 0.813679
## strategy:mc 0.3429 1 0.558155
## id:cue:strategy 0.1735 1 0.677005
## id:cue:mc 0.6572 1 0.417547
## id:strategy:mc 0.4108 1 0.521551
## cue:strategy:mc 1.2864 1 0.256716
## id:cue:strategy:mc 0.1421 1 0.706165
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
3.3.1.2 Exponentiated Coeff
jtools::summ(mc_7)
|
Observations
|
928
|
|
Dependent variable
|
bin
|
|
Type
|
Mixed effects generalized linear model
|
|
Family
|
binomial
|
|
Link
|
logit
|
|
AIC
|
750.50
|
|
BIC
|
832.67
|
|
Pseudo-R² (fixed effects)
|
0.46
|
|
Pseudo-R² (total)
|
0.67
|
|
Fixed Effects
|
|
|
Est.
|
S.E.
|
z val.
|
p
|
|
(Intercept)
|
-5.79
|
2.09
|
-2.77
|
0.01
|
|
id
|
0.62
|
0.32
|
1.94
|
0.05
|
|
cueTrue
|
2.56
|
2.32
|
1.11
|
0.27
|
|
strategyswitch
|
-2.02
|
4.88
|
-0.41
|
0.68
|
|
mc
|
3.42
|
2.57
|
1.33
|
0.18
|
|
id:cueTrue
|
0.11
|
0.43
|
0.25
|
0.80
|
|
id:strategyswitch
|
0.03
|
0.77
|
0.04
|
0.97
|
|
cueTrue:strategyswitch
|
-0.73
|
5.39
|
-0.13
|
0.89
|
|
id:mc
|
-0.27
|
0.39
|
-0.68
|
0.49
|
|
cueTrue:mc
|
1.94
|
2.93
|
0.66
|
0.51
|
|
strategyswitch:mc
|
-0.79
|
6.17
|
-0.13
|
0.90
|
|
id:cueTrue:strategyswitch
|
-0.41
|
0.88
|
-0.47
|
0.64
|
|
id:cueTrue:mc
|
-0.48
|
0.54
|
-0.89
|
0.37
|
|
id:strategyswitch:mc
|
-0.01
|
0.98
|
-0.01
|
0.99
|
|
cueTrue:strategyswitch:mc
|
0.30
|
6.81
|
0.04
|
0.96
|
|
id:cueTrue:strategyswitch:mc
|
0.42
|
1.12
|
0.38
|
0.71
|
|
Random Effects
|
|
Group
|
Parameter
|
Std. Dev.
|
|
partID
|
(Intercept)
|
1.41
|
|
Grouping Variables
|
|
Group
|
# groups
|
ICC
|
|
partID
|
29
|
0.38
|
3.3.1.3 Visualizations
plot_model(mc_7, type = "pred", terms = c("mc [all]", "cue","id"))
3.3.2 10-yos
mc_10 <- glmer(bin ~ id*cue*strategy*mc + (1|partID),
family = binomial(link = "logit"),
control=glmerControl(optimizer="bobyqa", optCtrl=list(maxfun=1000000)),
data = df[df$age_grp == "t",])
mc_10
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: bin ~ id * cue * strategy * mc + (1 | partID)
## Data: df[df$age_grp == "t", ]
## AIC BIC logLik deviance df.resid
## 726.9583 807.2635 -346.4792 692.9583 815
## Random effects:
## Groups Name Std.Dev.
## partID (Intercept) 1.35
## Number of obs: 832, groups: partID, 26
## Fixed Effects:
## (Intercept) id
## -3.47109 0.74201
## cueTrue strategyswitch
## 6.65009 0.29540
## mc id:cueTrue
## -2.70900 -1.07789
## id:strategyswitch cueTrue:strategyswitch
## -0.34727 0.94987
## id:mc cueTrue:mc
## 0.52396 0.74749
## strategyswitch:mc id:cueTrue:strategyswitch
## 2.96554 -0.02709
## id:cueTrue:mc id:strategyswitch:mc
## 0.17723 -0.42389
## cueTrue:strategyswitch:mc id:cueTrue:strategyswitch:mc
## -5.11628 0.64203
3.3.2.1 LR Chi Sqr
car::Anova(mc_10)
## Analysis of Deviance Table (Type II Wald chisquare tests)
##
## Response: bin
## Chisq Df Pr(>Chisq)
## id 45.5423 1 1.494e-11 ***
## cue 104.5452 1 < 2.2e-16 ***
## strategy 18.0020 1 2.207e-05 ***
## mc 0.0376 1 0.84633
## id:cue 38.2573 1 6.200e-10 ***
## id:strategy 15.2907 1 9.217e-05 ***
## cue:strategy 4.2079 1 0.04024 *
## id:mc 4.6318 1 0.03138 *
## cue:mc 0.0811 1 0.77581
## strategy:mc 0.0342 1 0.85322
## id:cue:strategy 4.7543 1 0.02922 *
## id:cue:mc 1.2238 1 0.26861
## id:strategy:mc 0.0010 1 0.97432
## cue:strategy:mc 0.7348 1 0.39133
## id:cue:strategy:mc 0.2839 1 0.59416
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
3.2.2.2 Exponentiated Coeff
jtools::summ(mc_10)
|
Observations
|
832
|
|
Dependent variable
|
bin
|
|
Type
|
Mixed effects generalized linear model
|
|
Family
|
binomial
|
|
Link
|
logit
|
|
AIC
|
726.96
|
|
BIC
|
807.26
|
|
Pseudo-R² (fixed effects)
|
0.44
|
|
Pseudo-R² (total)
|
0.64
|
|
Fixed Effects
|
|
|
Est.
|
S.E.
|
z val.
|
p
|
|
(Intercept)
|
-3.47
|
3.55
|
-0.98
|
0.33
|
|
id
|
0.74
|
0.63
|
1.18
|
0.24
|
|
cueTrue
|
6.65
|
3.98
|
1.67
|
0.09
|
|
strategyswitch
|
0.30
|
3.92
|
0.08
|
0.94
|
|
mc
|
-2.71
|
4.70
|
-0.58
|
0.56
|
|
id:cueTrue
|
-1.08
|
0.78
|
-1.39
|
0.16
|
|
id:strategyswitch
|
-0.35
|
0.73
|
-0.47
|
0.64
|
|
cueTrue:strategyswitch
|
0.95
|
4.79
|
0.20
|
0.84
|
|
id:mc
|
0.52
|
0.83
|
0.63
|
0.53
|
|
cueTrue:mc
|
0.75
|
5.17
|
0.14
|
0.88
|
|
strategyswitch:mc
|
2.97
|
5.19
|
0.57
|
0.57
|
|
id:cueTrue:strategyswitch
|
-0.03
|
0.93
|
-0.03
|
0.98
|
|
id:cueTrue:mc
|
0.18
|
1.01
|
0.17
|
0.86
|
|
id:strategyswitch:mc
|
-0.42
|
0.96
|
-0.44
|
0.66
|
|
cueTrue:strategyswitch:mc
|
-5.12
|
6.18
|
-0.83
|
0.41
|
|
id:cueTrue:strategyswitch:mc
|
0.64
|
1.20
|
0.53
|
0.59
|
|
Random Effects
|
|
Group
|
Parameter
|
Std. Dev.
|
|
partID
|
(Intercept)
|
1.35
|
|
Grouping Variables
|
|
Group
|
# groups
|
ICC
|
|
partID
|
26
|
0.36
|
3.3.2.3 Visualizations
plot_model(mc_10, type = "pred", terms = c("mc [all]", "cue","id"))
3.3.3 Adults
mc_adult <- glmer(bin ~ id*cue*strategy*mc + (1|partID),
family = binomial(link = "logit"),
control=glmerControl(optimizer="bobyqa", optCtrl=list(maxfun=1000000)),
data = df[df$age_grp == "a",])
mc_adult
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: bin ~ id * cue * strategy * mc + (1 | partID)
## Data: df[df$age_grp == "a", ]
## AIC BIC logLik deviance df.resid
## 697.8578 781.6928 -331.9289 663.8578 1007
## Random effects:
## Groups Name Std.Dev.
## partID (Intercept) 1.427
## Number of obs: 1024, groups: partID, 32
## Fixed Effects:
## (Intercept) id
## -3.1304 1.3715
## cueTrue strategyswitch
## 6.5032 -2.7296
## mc id:cueTrue
## 0.8937 -0.2449
## id:strategyswitch cueTrue:strategyswitch
## -0.3982 1.9284
## id:mc cueTrue:mc
## -0.5693 -1.2609
## strategyswitch:mc id:cueTrue:strategyswitch
## 2.2034 -0.5422
## id:cueTrue:mc id:strategyswitch:mc
## -0.3239 0.1752
## cueTrue:strategyswitch:mc id:cueTrue:strategyswitch:mc
## -3.6583 0.8198
3.3.3.1 LR Chi Sqr
car::Anova(mc_adult)
## Analysis of Deviance Table (Type II Wald chisquare tests)
##
## Response: bin
## Chisq Df Pr(>Chisq)
## id 96.9261 1 < 2.2e-16 ***
## cue 129.0987 1 < 2.2e-16 ***
## strategy 68.2818 1 < 2.2e-16 ***
## mc 0.0039 1 0.9499994
## id:cue 14.3554 1 0.0001513 ***
## id:strategy 3.6394 1 0.0564260 .
## cue:strategy 0.9461 1 0.3307061
## id:mc 1.3328 1 0.2483013
## cue:mc 4.3999 1 0.0359413 *
## strategy:mc 4.6967 1 0.0302198 *
## id:cue:strategy 0.2022 1 0.6529364
## id:cue:mc 0.3206 1 0.5712510
## id:strategy:mc 0.4371 1 0.5085194
## cue:strategy:mc 0.0952 1 0.7576364
## id:cue:strategy:mc 0.3786 1 0.5383375
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
3.3.3.2 Exponentiated Coeff
jtools::summ(mc_adult)
|
Observations
|
1024
|
|
Dependent variable
|
bin
|
|
Type
|
Mixed effects generalized linear model
|
|
Family
|
binomial
|
|
Link
|
logit
|
|
AIC
|
697.86
|
|
BIC
|
781.69
|
|
Pseudo-R² (fixed effects)
|
0.56
|
|
Pseudo-R² (total)
|
0.73
|
|
Fixed Effects
|
|
|
Est.
|
S.E.
|
z val.
|
p
|
|
(Intercept)
|
-3.13
|
1.84
|
-1.70
|
0.09
|
|
id
|
1.37
|
0.44
|
3.10
|
0.00
|
|
cueTrue
|
6.50
|
3.68
|
1.77
|
0.08
|
|
strategyswitch
|
-2.73
|
2.24
|
-1.22
|
0.22
|
|
mc
|
0.89
|
2.35
|
0.38
|
0.70
|
|
id:cueTrue
|
-0.24
|
1.02
|
-0.24
|
0.81
|
|
id:strategyswitch
|
-0.40
|
0.53
|
-0.75
|
0.45
|
|
cueTrue:strategyswitch
|
1.93
|
4.28
|
0.45
|
0.65
|
|
id:mc
|
-0.57
|
0.55
|
-1.04
|
0.30
|
|
cueTrue:mc
|
-1.26
|
4.55
|
-0.28
|
0.78
|
|
strategyswitch:mc
|
2.20
|
2.81
|
0.78
|
0.43
|
|
id:cueTrue:strategyswitch
|
-0.54
|
1.13
|
-0.48
|
0.63
|
|
id:cueTrue:mc
|
-0.32
|
1.18
|
-0.27
|
0.78
|
|
id:strategyswitch:mc
|
0.18
|
0.67
|
0.26
|
0.79
|
|
cueTrue:strategyswitch:mc
|
-3.66
|
5.33
|
-0.69
|
0.49
|
|
id:cueTrue:strategyswitch:mc
|
0.82
|
1.33
|
0.62
|
0.54
|
|
Random Effects
|
|
Group
|
Parameter
|
Std. Dev.
|
|
partID
|
(Intercept)
|
1.43
|
|
Grouping Variables
|
|
Group
|
# groups
|
ICC
|
|
partID
|
32
|
0.38
|
3.3.3.3 Visualizations
plot_model(mc_adult, type = "pred", terms = c("mc [all]", "cue"))
plot_model(mc_adult, type = "pred", terms = c("mc [all]", "strategy"))
3.3.3.4 Pairwise
emmeans::emtrends(mc_adult, pairwise ~ cue, var = "mc")
## NOTE: Results may be misleading due to involvement in interactions
## $emtrends
## cue mc.trend SE df asymp.LCL asymp.UCL
## False -0.172 1.45 Inf -3.00 2.66
## True -2.875 1.98 Inf -6.76 1.01
##
## Results are averaged over the levels of: strategy
## Confidence level used: 0.95
##
## $contrasts
## contrast estimate SE df z.ratio p.value
## False - True 2.7 1.64 Inf 1.647 0.0995
##
## Results are averaged over the levels of: strategy
emmeans::emtrends(mc_adult, pairwise ~ strategy, var = "mc")
## NOTE: Results may be misleading due to involvement in interactions
## $emtrends
## strategy mc.trend SE df asymp.LCL asymp.UCL
## maint -3.0274 1.97 Inf -6.89 0.835
## switch -0.0201 1.44 Inf -2.85 2.809
##
## Results are averaged over the levels of: cue
## Confidence level used: 0.95
##
## $contrasts
## contrast estimate SE df z.ratio p.value
## maint - switch -3.01 1.61 Inf -1.862 0.0625
##
## Results are averaged over the levels of: cue