Open RenyBB opened 2 months ago
m<- lmerTest::lmer(mpg ~ wt+ (1| gear), data=mtcars)
parameters::parameters(anova(m))
#> Parameter | Sum_Squares | df | Mean_Square | F | p
#> -----------------------------------------------------------
#> wt | 526.21 | 1 | 526.21 | 62.20 | < .001
#>
#> Anova Table (Type 3 tests)
Created on 2024-07-09 with reprex v2.0.2
I reckon that's because parameters() doesn't return one by default (@mattansb expert of anovas)
We've only tested for lme4, which doesn't include the denominator df. Package lmerTest, however, does. This should be fixed in parameters now (https://github.com/easystats/parameters/pull/988).
report still needs to be fixed, though.
m <- lmerTest::lmer(mpg ~ wt + (1 | gear), data = mtcars)
parameters::parameters(anova(m))
#> Parameter | Sum_Squares | df | df (error) | Mean_Square | F | p
#> ------------------------------------------------------------------------
#> wt | 526.21 | 1 | 21.92 | 526.21 | 62.20 | < .001
#>
#> Anova Table (Type 3 tests)
report::report(anova(m))
#> The ANOVA suggests that:
#>
#> - The main effect of wt is statistically significant and large (F(1) = 62.20, p
#> < .001; Eta2 (partial) = 0.74, 95% CI [0.56, 1.00])
#>
#> Effect sizes were labelled following Field's (2013) recommendations.
Created on 2024-07-09 with reprex v2.1.1
@RenyBB also note that in-general, dfs for mixed models are a tricky business (https://github.com/easystats/parameters/issues/989)
For example, the following code...
... provides a written summary that is missing the denominator degrees of freedom:
" The ANOVA suggests that:The main effect of wt is statistically significant and large (F(1) = 62.20, p < .001; Eta2 (partial) = 0.74, 95% CI [0.56, 1.00]) "