Adjusted p values from afex::aov_4() not retained by model_parameters()

[strengejacke]

Fixes #1101

[strengejacke]

I reverted the change for re-computing the ANOVA, because it returned wrong p-values?

data(laptop_urry, package = "afex")
afx <- afex::aov_4(
  overall ~ condition * talk + (1 | pid),
  data = laptop_urry,
  anova_table = list(p_adjust_method = "bonferroni")
)
#> Contrasts set to contr.sum for the following variables: condition, talk

afx
#> Anova Table (Type 3 tests, bonferroni-adjusted)
#> 
#> Response: overall
#>           Effect     df    MSE        F  ges p.value
#> 1      condition 1, 132 220.46     0.52 .004   >.999
#> 2           talk 4, 132 220.46 7.60 *** .187   <.001
#> 3 condition:talk 4, 132 220.46     2.20 .063    .216
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '+' 0.1 ' ' 1
anova(afx)
#> Anova Table (Type 3 tests, bonferroni-adjusted)
#> 
#> Response: overall
#>                num Df den Df    MSE      F      ges    Pr(>F)    
#> condition           1    132 220.46 0.5217 0.003937    1.0000    
#> talk                4    132 220.46 7.6018 0.187229 4.579e-05 ***
#> condition:talk      4    132 220.46 2.2045 0.062619    0.2157    
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

afex::aov_4(
  overall ~ condition * talk + (1 | pid),
  data = laptop_urry
)
#> Contrasts set to contr.sum for the following variables: condition, talk
#> Anova Table (Type 3 tests)
#> 
#> Response: overall
#>           Effect     df    MSE        F  ges p.value
#> 1      condition 1, 132 220.46     0.52 .004    .471
#> 2           talk 4, 132 220.46 7.60 *** .187   <.001
#> 3 condition:talk 4, 132 220.46   2.20 + .063    .072
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '+' 0.1 ' ' 1

^{Created on 2025-05-08 with reprex v2.1.1}

[strengejacke]

I'm not sure why we extract model$anova_table in one case, and model$Anova in the other case, but the first returns adjusted p-values, the latter not. I fixed the code, so p-adjustment is only applied to the latter case. Should be correct now.