Conflicting results in anova_test() for formula and argument input for two-way and higher order designs

[benediktclaus]

When conducting a between-subjects ANOVA with two or more factors using anova_test(), the results depend on how the model was defined (as formula vs. as function arguments). As checked with JASP and SPSS, the results are only correct, if the model is defined via function arguments. The formula call gives wrong results.

library(tidyverse)
library(rstatix)
#> 
#> Attache Paket: 'rstatix'
#> The following object is masked from 'package:stats':
#> 
#>     filter
library(datarium)

satisfaction <- datarium::jobsatisfaction

satisfaction %>% 
  anova_test(score ~ gender*education_level, type = 3)
#> Coefficient covariances computed by hccm()
#> ANOVA Table (type III tests)
#> 
#>                   Effect DFn DFd       F        p p<.05   ges
#> 1                 gender   1  52   1.547 2.19e-01       0.029
#> 2        education_level   2  52 132.432 3.92e-21     * 0.836
#> 3 gender:education_level   2  52   7.338 2.00e-03     * 0.220

satisfaction %>% 
  anova_test(dv = score, between = c(gender, education_level), type = 3)
#> Coefficient covariances computed by hccm()
#> ANOVA Table (type III tests)
#> 
#>                   Effect DFn DFd       F        p p<.05   ges
#> 1                 gender   1  52   0.586 4.48e-01       0.011
#> 2        education_level   2  52 189.172 1.37e-24     * 0.879
#> 3 gender:education_level   2  52   7.338 2.00e-03     * 0.220

^{Created on 2020-05-05 by the reprex package (v0.3.0)}

[kassambara]

fixed now, thank you for pointing this out