[BlindRUG] Interpreting QQ Plot nonvisually or alternative approaches

[Jonathan Godfrey]

Hello Miso,

I get asked this question a lot; far too often to be honest.

You mentioned residuals, so I am assuming your question relates to fitting linear models (of any kind) using the `lm()` function.

There are four assumptions for the linear model, least important of which is the normality of residuals. It is however the easiest one for sighted people to assess courtesy of the normal quantile plot.

A lack of normality is often a symptom that another assumption is also questionable.

Not having the correct model, constant variance, and independence of residuals all have greater impact than the lack of normality.

You can investigate most assumptions by way of fitting alternative models or making use of various hypothesis tests (variance and independence), not to forget the tests for normality. Many of these are used in the `UniDesc()` function in the BrailleR package.

Jonathan






From: BlindRUG <blindrug-bounces at nfbnet.org> On Behalf Of Miso Kwak via BlindRUG
Sent: Monday, November 27, 2023 10:58 AM
To: 'Blind R Users Group' <blindrug at nfbnet.org>
Cc: misokwak12 at gmail.com
Subject: [BlindRUG] Interpreting QQ Plot nonvisually or alternative approaches

Hello,
Is there any way to understand a QQ Plot I produce in R nonvisually?
Or is there any comparable alternative that would allow me to make interpretation on the distribution of residuals?

Thank you in advance.
Miso



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://nfbnet.org/pipermail/blindrug_nfbnet.org/attachments/20231126/8bcbe813/attachment.html>