blog/two-way-anova-in-r/

[giscus[bot]]

blog/two-way-anova-in-r/

Learn how to do a two-way ANOVA in R. You will also learn its aim, hypotheses, assumptions, and how to interpret the results of the two-way ANOVA

https://statsandr.com/blog/two-way-anova-in-r/

[technocrat]

Thanks for finally clearing that up. I had learned much of this piece wise but never connected the dots coherently.

[AntoineSoetewey]

Glad you find it useful @technocrat !

[003136]

This was extremely helpful! My mentor asked that I rerun the tests, but control for two covariates (Age and Gender). If there a way to add covariates to the ggbetweenstats function for it runs an ANCOVA instead of an ANOVA?

[AntoineSoetewey]

Hello,

No unfortunately the {ggstatsplot} package does not allow to run ANCOVA nor two-way ANOVA.

These two posts may be helpful in your case:

• Two-way ANOVA in R
• Multiple linear regression in R

Hope this helps.

Regards,
Antoine

[adamczycha]

This blog is marvellous. I'm so glad that I found you! Thank you for sharing the knowledge 😍

[AntoineSoetewey]

Thank you for your kind words @Marcyyyl !

[fiuzatayna]

Hey, thank you for the walkthrough! Much appreciated. 😄

I was wondering if you know of any visualization package like ggstatsplot that accepts working with Dunnett's test results.

[AntoineSoetewey]

Dear @fiuzatayna,

It seems to be on the to do list of the maintainer of the {ggstatsplot} package (see IndrajeetPatil/ggstatsplot#500).

There is also the {ggpubr} package, but as far as I know, I don't think there is a possibility to display results of Dunnett's test.

In the meantime, you could try adding the p-values manually to a ggplot with annotate() or geom_text(). You can also check this tutorial https://www.datanovia.com/en/blog/ggpubr-how-to-add-p-values-generated-elsewhere-to-a-ggplot/, but it is not as straightforward and easy as with {ggstatsplot}.

If you find a good way, feel free to let me know; I am also interested in this feature!

Regards,
Antoine

[AndrewSalvatore]

Hello,

I'm trying to understand the process of deciding upon and implementing type 1, 2, or 3 ANOVA. I explored the external link you provided but am still having difficulties understanding this concept.

In your example, you use a type 1 ANOVA, which assumes a balanced design. However, the sample sizes of sex and species in the penguin dataset are unbalanced. Could you elaborate on this?

Additionally, you say that type 2 should be used when there are no significant interactions and type 3 when there are. In that case, one would have to conduct an ANOVA first to know if a significant interaction exists or not. Does this mean you first conduct a type 1 ANOVA, and then reconduct a type 2 or 3 based on the results of the first ANOVA?

Any help would be greatly appreciated,
Andrew

[AntoineSoetewey]

Dear @AndrewSalvatore,

Thanks for your questions.

You are right: type I ANOVA assumes a balanced design, and this is the reason I wrote "we assume a balanced design here" (even though it is unbalanced, but I wanted to use the penguins dataset for this post).

Regarding your second question about the interaction: it's best to check whether there is an interaction, and then use the appropriate type of ANOVA. What I usually do is a type III ANVOVA (or a linear model with interaction (i.e., lm(y ~ a * b)) and keep it if the interaction is significant, otherwise simplify the model by running a type II ANOVA (without the interaction).

Hope this is more clear.

Regards,
Antoine

[nop9898]

Thank you for your extremely helpful blog.

[AntoineSoetewey]

You're welcome @nop9898!