0

I am pretty struggling to understand the concept of ANOVA, why do we need such a process? I also notice during the model utility test(full model vs reduced model/ check whether any of estimated parameter is not important) we also call anova(). what is this test relationship with analysis of variance?

Appreciate for any comments.

LJNG
  • 205
  • 1
  • 7
  • Does this answer your question? [Logic behind the ANOVA F-test in simple linear regression](https://stats.stackexchange.com/q/8237/930) – chl Oct 22 '20 at 06:37
  • Thank you for your reply. No really, I know the details about comparing nested model and assessing deviance, but confusing its relationship with analysis of variance, comparing different group's variations. how come they call the same code in R. – LJNG Oct 23 '20 at 09:33
  • 1
    These are the same, though. The `aov` function is just a wrapper around `lm` with extra arguments allowed in the formula (e.g., controlling the `Error` term), so you could use `lm` directly to fit a two-way ANOVA, much like you could use `glm` to fit a simple linear regression, after specifying the correct `family` (gaussian). In the case of a linear regression model, you can provide a table of regression coefficients (with their individual significance level) and/or a table of regression for the model as a whole, as described in the above link, depending on what you're interested in. – chl Oct 23 '20 at 17:04

0 Answers0