固定效应的交互作用显著:
model.21 <- glmer(ACC1 ~ CONGRUENCY * LEVEL + scale(collocationfrequency)
+ +(1|SUBJ)+(1+LEVEL|COLLOCATION), data = mydata2, family = "binomial")
summary(model.21) # 2548.5 拟合成功 最佳模型
Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
Family: binomial ( logit )
Formula: ACC1 ~ CONGRUENCY * LEVEL + scale(collocationfrequency) + (1 | SUBJ) + (1 + LEVEL | COLLOCATION)
Data: mydata2
AIC BIC logLik deviance df.resid
2548.5 2604.2 -1265.3 2530.5 3564
Scaled residuals:
Min 1Q Median 3Q Max
-10.9057 0.0863 0.2297 0.4246 1.7175
Random effects:
Groups Name Variance Std.Dev. Corr
SUBJ (Intercept) 0.1737 0.4168
COLLOCATION (Intercept) 0.4440 0.6663
LEVELintermediate 0.5944 0.7710 -0.08
Number of obs: 3573, groups: SUBJ, 64; COLLOCATION, 56
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 4.7215 0.3313 14.253 < 2e-16 ***
CONGRUENCYL2-only -3.1786 0.3549 -8.956 < 2e-16 ***
LEVELintermediate -2.2747 0.3653 -6.227 4.74e-10 ***
scale(collocationfrequency) 0.4719 0.1240 3.807 0.00014 ***
CONGRUENCYL2-only:LEVELintermediate 1.5261 0.3923 3.891 0.00010 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Correlation of Fixed Effects:
(Intr) CONGRUENCYL2-n LEVELn scl(c)
CONGRUENCYL2-n -0.876
LEVELntrmdt -0.763 0.661
scl(cllctn) 0.188 -0.191 -0.105
CONGRUENCYL2-: 0.649 -0.666 -0.834 0.061
主效应的交互作用显著:
Analysis of Deviance Table (Type II Wald chisquare tests)
Response: ACC1
Chisq Df Pr(>Chisq)
CONGRUENCY 72.776 1 < 2.2e-16 ***
LEVEL 29.192 1 6.556e-08 ***
scale(collocationfrequency) 14.496 1 0.0001404 ***
CONGRUENCY:LEVEL 15.137 1 0.0001000 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
但是,简单效应分析后是没有交互作用的,这是为什么啊?求大神们支招
$contrasts
CONGRUENCY = L1-L2:
contrast estimate SE df z.ratio p.value
advanced - intermediate 2.275 0.365 Inf 6.227 <.0001
CONGRUENCY = L2-only:
contrast estimate SE df z.ratio p.value
advanced - intermediate 0.749 0.220 Inf 3.403 0.0007
Results are given on the log odds ratio (not the response) scale.
$contrasts
LEVEL = advanced:
contrast estimate SE df z.ratio p.value
(L1-L2) - (L2-only) 3.18 0.355 Inf 8.956 <.0001
LEVEL = intermediate:
contrast estimate SE df z.ratio p.value
(L1-L2) - (L2-only) 1.65 0.307 Inf 5.376 <.0001
Results are given on the log odds ratio (not the response) scale.