2
$\begingroup$

I have two models with the same outcome variable and the same independent variables, but Model 1 has two more independent variables than Model 2. I do an anova() test in R to check whether these two variables are important. According to the output they are.

  Res.Df     RSS Df Sum of Sq      F    Pr(>F)    
1  60914 1629530                                  
2  60916 1636407 -2   -6877.2 128.54 < 2.2e-16 ***

How do I have to present these results in an academic paper? I'd prefer mentioning it within the text, rather than adding an entire table.

Is something like this ok: F(60,916, 60,914) = 128.54, p < 0.001? I've seen some sources use this for anova tests which compare a variable across groups.

Improve this question
$\endgroup$
4
  • 3
    $\begingroup$ Charlotte, it depends on publication norm you use. For example, I use APA norms (American Psychological Association) and they recommend to report anova in this format: F(2, 60914) = 128.54, p < .001. Note that first degrees of freedom are the difference of d.f. between your model and its submodel. $\endgroup$
    – Daniel Dostal
    Commented Feb 26, 2021 at 9:58
  • $\begingroup$ Thank you @DanielDostal, I also use APA so this solves my question! $\endgroup$
    – Charlotte
    Commented Feb 26, 2021 at 10:07
  • $\begingroup$ Follow up question: Why is it F(2, 60914), and not F(2, 60916)? $\endgroup$
    – Dunen
    Commented Aug 1, 2022 at 10:51
  • $\begingroup$ @DanielDostal Your comment looks like an answer. I'm trying to reduce the number of unanswered questions. Could you post it as an answer? $\endgroup$
    – Peter Flom
    Commented Dec 17, 2023 at 12:26

0

Reset to default

Browse other questions tagged or ask your own question.