I've been lurking for a bit and have enjoyed watching you get organized and start tackling this. I'm a long time user of the book and the lsr package in my teaching and actually thanked Danielle on a few occasions.
I'm not sure exactly how often or how much I can be of assistance but if this is indeed one of those cases where many hands make light work versus just add more complexity I'd love to try.
I'm much more a practitioner than a theorist but I do believe I can help even in small ways if you're resolved to move towards the tidyverse. Simple example is the etaSquared issue which was written when there were a lot fewer options out there. There's a nice clean solution in the sjstats package that is tidy and is easily dropped into the book to reinforce the teaching points around types of sums of squares and effect sizes...
Don't want to butt in since you all seem well-organized and moving forward but thought I would ask.
Chuck
xmtcars <- mtcars
xmtcars$cyl <- as.factor(xmtcars$cyl)
xmtcars$am <- as.factor(xmtcars$am)
sjstats::anova_stats(aov(mpg~am*cyl, xmtcars)) #type 1 order matters
#> term df sumsq meansq statistic p.value etasq partial.etasq
#> 1 am 1 405.151 405.151 44.064 0.000 0.360 0.629
#> 2 cyl 2 456.401 228.200 24.819 0.000 0.405 0.656
#> 3 am:cyl 2 25.437 12.718 1.383 0.269 0.023 0.096
#> 4 Residuals 26 239.059 9.195 NA NA NA NA
#> omegasq partial.omegasq cohens.f power
#> 1 0.349 0.574 1.302 1.000
#> 2 0.386 0.598 1.382 1.000
#> 3 0.006 0.023 0.326 0.298
#> 4 NA NA NA NA
sjstats::anova_stats(aov(mpg~cyl*am, xmtcars)) #type 1 order matters
#> term df sumsq meansq statistic p.value etasq partial.etasq
#> 1 cyl 2 824.785 412.392 44.852 0.000 0.732 0.775
#> 2 am 1 36.767 36.767 3.999 0.056 0.033 0.133
#> 3 cyl:am 2 25.437 12.718 1.383 0.269 0.023 0.096
#> 4 Residuals 26 239.059 9.195 NA NA NA NA
#> omegasq partial.omegasq cohens.f power
#> 1 0.710 0.733 1.857 1.000
#> 2 0.024 0.086 0.392 0.515
#> 3 0.006 0.023 0.326 0.298
#> 4 NA NA NA NA
sjstats::anova_stats(car::Anova(aov(mpg~am*cyl, xmtcars), type =2))
#> term sumsq meansq df statistic p.value etasq partial.etasq
#> 1 am 36.767 36.767 1 3.999 0.056 0.049 0.133
#> 2 cyl 456.401 228.200 2 24.819 0.000 0.602 0.656
#> 3 am:cyl 25.437 12.718 2 1.383 0.269 0.034 0.096
#> 4 Residuals 239.059 9.195 26 NA NA NA NA
#> omegasq partial.omegasq cohens.f power
#> 1 0.036 0.086 0.392 0.515
#> 2 0.571 0.598 1.382 1.000
#> 3 0.009 0.023 0.326 0.298
#> 4 NA NA NA NA
sjstats::anova_stats(car::Anova(aov(mpg~am*cyl, xmtcars), type =3))
#> term sumsq meansq df statistic p.value etasq partial.etasq
#> 1 (Intercept) 1573.230 1573.230 1 171.104 0.000 0.762 0.868
#> 2 am 58.430 58.430 1 6.355 0.018 0.028 0.196
#> 3 cyl 167.710 83.855 2 9.120 0.001 0.081 0.412
#> 4 am:cyl 25.437 12.718 2 1.383 0.269 0.012 0.096
#> 5 Residuals 239.059 9.195 26 NA NA NA NA
#> omegasq partial.omegasq cohens.f power
#> 1 0.754 0.838 2.565 1.000
#> 2 0.024 0.140 0.494 0.712
#> 3 0.072 0.330 0.838 0.975
#> 4 0.003 0.023 0.326 0.298
#> 5 NA NA NA NA
Hi,
I've been lurking for a bit and have enjoyed watching you get organized and start tackling this. I'm a long time user of the book and the
lsrpackage in my teaching and actually thanked Danielle on a few occasions.I'm not sure exactly how often or how much I can be of assistance but if this is indeed one of those cases where many hands make light work versus just add more complexity I'd love to try.
I'm much more a practitioner than a theorist but I do believe I can help even in small ways if you're resolved to move towards the
tidyverse. Simple example is theetaSquaredissue which was written when there were a lot fewer options out there. There's a nice clean solution in thesjstatspackage that istidyand is easily dropped into the book to reinforce the teaching points around types of sums of squares and effect sizes...Don't want to butt in since you all seem well-organized and moving forward but thought I would ask.
Chuck
Created on 2019-01-11 by the reprex package (v0.2.1)