Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: score ~ group * condition + (1 | subject) + (1 | token_set)
Data: EN_JT_1
REML criterion at convergence: 512.7
Scaled residuals:
Min 1Q Median 3Q Max
-3.2745 -0.3231 0.1870 0.6697 1.8834
Random effects:
Groups Name Variance Std.Dev.
subject (Intercept) 0.021850 0.14782
token_set (Intercept) 0.003202 0.05658
Residual 0.091070 0.30178
Number of obs: 852, groups: subject, 71; token_set, 24
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) 1.00047 0.03727 85.51605 26.844 < 2e-16 ***
groupHS -0.06865 0.05296 106.58618 -1.296 0.19773
groupSB -0.03172 0.05862 106.58450 -0.541 0.58959
conditionEN-GJT-R-RES -0.28400 0.03813 48.55384 -7.449 1.42e-09 ***
groupHS:conditionEN-GJT-R-RES -0.08721 0.04796 756.84544 -1.818 0.06938 .
groupSB:conditionEN-GJT-R-RES 0.14858 0.05308 756.82431 2.799 0.00525 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Correlation of Fixed Effects:
(Intr) gropHS gropSB cEN-GJ gHS:EN
groupHS -0.569
groupSB -0.514 0.361
cEN-GJT-R-R -0.512 0.228 0.206
gHS:EN-GJT- 0.257 -0.453 -0.164 -0.503
gSB:EN-GJT- 0.233 -0.164 -0.453 -0.455 0.361
I have run the mixed-model above, where I crossed two fixed variables (group and condition) with each other to look for any effects of each as well as any interactions between the two.
I have three groups (EN, HS, and SB) and two conditions (EN-GJT-R-RES and EN-GJT-R-GAP).
I have two questions:
-
I'm not sure how to understand the output under "Fixed Effects" -- specifically the final two rows. I believe in the three rows below (Intercept) each group or condition is compared to the reference group or condition: in other words, in the second row HS is compared to EN, in the third row SB is compared to EN, and in the fourth row EN-GJT-R-RES is compared to EN-GJT-R-GAP. Is that correct? I'm not sure about the last two rows. What is being compared to what?
-
How do I run a post-hoc test that will compare all groups and all conditions with each other, as well as show all interactions? I found a line of code here that will let me run a Tukey test for groups, and a separate one for conditions (see below), but I'm not sure how I can run a Tukey test to show interactions.
> summary(glht(model_1, linfct = mcp(group = "Tukey")), test = adjusted("holm"))
Simultaneous Tests for General Linear Hypotheses
Multiple Comparisons of Means: Tukey Contrasts
Fit: lmer(formula = score ~ group * condition + (1 | subject) + (1 |
token_set), data = EN_JT_1)
Linear Hypotheses:
Estimate Std. Error z value Pr(>|z|)
HS - EN == 0 -0.06865 0.05296 -1.296 0.585
SB - EN == 0 -0.03172 0.05862 -0.541 1.000
SB - HS == 0 0.03693 0.06322 0.584 1.000
(Adjusted p values reported -- holm method)
> summary(glht(model_1, linfct = mcp(condition = "Tukey")), test = adjusted("holm"))
Simultaneous Tests for General Linear Hypotheses
Multiple Comparisons of Means: Tukey Contrasts
Fit: lmer(formula = score ~ group * condition + (1 | subject) + (1 |
token_set), data = EN_JT_1)
Linear Hypotheses:
Estimate Std. Error z value Pr(>|z|)
EN-GJT-R-RES - EN-GJT-R-GAP == 0 -0.28400 0.03813 -7.449 9.41e-14 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
(Adjusted p values reported -- holm method)