4

I can get the significance of pairwise comparisons with the following code

m <- lmer(angle ~ recipe*temp + (1|replicate), data=cake)
emtrends(m, pairwise~recipe, var="temp")

$emtrends
 recipe temp.trend         SE  df   lower.CL  upper.CL
 A       0.1537143 0.02981898 250 0.09498586 0.2124427
 B       0.1645714 0.02981898 250 0.10584300 0.2232999
 C       0.1558095 0.02981898 250 0.09708110 0.2145379

$contrasts
 contrast     estimate        SE  df t.ratio p.value
 A - B    -0.010857143 0.0421704 250  -0.257  0.9641
 A - C    -0.002095238 0.0421704 250  -0.050  0.9986
 B - C     0.008761905 0.0421704 250   0.208  0.9765

However, what if I'm interested in whether the trend for each recipe is significant on its own? How can I get the significance of the $emtrends?

locus
  • 387
  • 2
  • 9

2 Answers2

5

The simplest way is to do

test(emtrends(m, ~recipe, var="temp"))
Russ Lenth
  • 5,922
  • 2
  • 13
  • 21
1

You can get p-values from t-values:

out.emtrends <- emtrends(m, pairwise~recipe, var="temp")
emtr <- as.data.frame(out.emtrends$emtrends)
tvalues <- emtr$temp.trend/emtr$SE
dfs <- emtr$df
(pvalues <- 2*pt(-abs(tvalues), dfs))

# 5.158650e-07 8.514898e-08 3.669655e-07
Marco Sandri
  • 23,289
  • 7
  • 54
  • 58