can model_parameters.aov return type 3 SSs?

[mattansb]

Currently gives type 1 SSs 😨

mtcars$am_f <- factor(mtcars$am)
mtcars$cyl_f <- factor(mtcars$cyl)

model <- aov(mpg ~ am_f * cyl_f, data = mtcars)

parameters::model_parameters(model)
#> Parameter  | Sum_Squares | df | Mean_Square |     F |      p
#> ------------------------------------------------------------
#> am_f       |      405.15 |  1 |      405.15 | 44.06 | < .001
#> cyl_f      |      456.40 |  2 |      228.20 | 24.82 | < .001
#> am_f:cyl_f |       25.44 |  2 |       12.72 |  1.38 | 0.269 
#> Residuals  |      239.06 | 26 |        9.19 |       |
anova(model)
#> Analysis of Variance Table
#> 
#> Response: mpg
#>            Df Sum Sq Mean Sq F value    Pr(>F)    
#> am_f        1 405.15  405.15 44.0641 4.847e-07 ***
#> cyl_f       2 456.40  228.20 24.8190 9.355e-07 ***
#> am_f:cyl_f  2  25.44   12.72  1.3832    0.2686    
#> Residuals  26 239.06    9.19                      
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
car::Anova(model, type = 2)
#> Anova Table (Type II tests)
#> 
#> Response: mpg
#>            Sum Sq Df F value    Pr(>F)    
#> am_f        36.77  1  3.9988   0.05608 .  
#> cyl_f      456.40  2 24.8190 9.355e-07 ***
#> am_f:cyl_f  25.44  2  1.3832   0.26861    
#> Residuals  239.06 26                      
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
car::Anova(model, type = 3)
#> Anova Table (Type III tests)
#> 
#> Response: mpg
#>              Sum Sq Df  F value    Pr(>F)    
#> (Intercept) 1573.23  1 171.1040 6.057e-13 ***
#> am_f          58.43  1   6.3549 0.0181761 *  
#> cyl_f        167.71  2   9.1200 0.0009978 ***
#> am_f:cyl_f    25.44  2   1.3832 0.2686140    
#> Residuals    239.06 26                       
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

^{Created on 2020-04-02 by the reprex package (v0.3.0)}

[strengejacke]

You can do this easily...

library(parameters)+
data(iris)
#> Error in library(parameters) + data(iris): non-numeric argument to binary operator
df <- iris
df$Sepal.Big <- ifelse(df$Sepal.Width >= 3, "Yes", "No")

model <- aov(Sepal.Length ~ Sepal.Big, data = df)
model_parameters(car::Anova(model, type = 3))
#> Parameter   | Sum_Squares |  df | Mean_Square |       F |      p
#> ----------------------------------------------------------------
#> (Intercept) |     2019.73 |   1 |     2019.73 | 2957.55 | < .001
#> Sepal.Big   |        1.10 |   1 |        1.10 |    1.61 | 0.207 
#> Residuals   |      101.07 | 148 |        0.68 |         |

^{Created on 2020-04-02 by the reprex package (v0.3.0)}

[mattansb]

Hmm true. Thinking about effectsize/issues/56 and about ANOVAs in general and the default people might expect...
Starting to really dislike ANOVAs :/

I guess this really isn't a "request" issue, but more of a "what do you think" issue 😊

[strengejacke]

I guess this really isn't a "request" issue, but more of a "what do you think" issue blush

If I see ANOVA I stop thinking

[mattansb]

[strengejacke]

My opinion is: the type argument is sufficient... @DominiqueMakowski ? You're the second psycho here... (not that this would mean your opinion counts anything, but... 😬 )

[mattansb]

My opinion is: the type argument is sufficient...

Agree. Closing this.