[BlindRUG] I can't go on learning R like this!

[Jonathan Godfrey]

Hello Amy,

The two points you make about a hypothesis test are true for every hypothesis test, no matter what the context.

In reality though, if the sample is small there are all sorts of problems you can't overcome.

If the sample is large, the impact of the problems even if they do exist is reducing. Some of the (old-fashioned) tests do account for size though. They are being used in scenarios that have more data than the standard size for taught courses.

If the sample is massive then the tools being used to validate our models etc. ought to change to accommodate the massiveness of the data. That takes you not data mining or machine learning territory.

A quick and dirty test for homogeneity is to split your data in half and test the variance of the response variable for the two halves. If the standard deviation is in a ratio of 2:1 (either way) or greater, then you have a problem. You ought to expect the ratio to be different to 1:1 through sheer random luck.

I'll take a look at your other stuff after the class I'm about to start.

HTH,
Jonathan


-----Original Message-----
From: BlindRUG <blindrug-bounces at nfbnet.org> On Behalf Of Amy Albin via BlindRUG
Sent: Wednesday, 9 March 2022 7:01 am
To: blindrug at nfbnet.org
Cc: Amy Albin <amyralbin at gmail.com>
Subject: Re: [BlindRUG] I can't go on learning R like this!

>From the Regression chapter of the Blind R book


There are non-graphical substitutes for the standard residuals analysis graphs that show the presence of non-constant variance; they are seldom taught as part of the standard introduction to regression analysis. A formal hypothesis test for non-constant variance can be conducted using either the Breusch-Pagan test found using the bptest() function, the Goldfeld-Quandt test found using the gqtest() function, both from the lmtest package, or the Non-constant Variance test found using the ncvTest() from the car package. These tests are simple to use if you are comfortable with hypothesis testing, but you will need to

My professor said these types of tests are bad because:

1. If the sample is large, we will find significance even if it doesn't matter.
2. If the sample is small, we may miss significance when there actually is a problem.

I've never heard of the tests mentioned above, but would the same principles apply?

Is the real answer to this question that blind people don't have as good methods available as sighted people?

On 3/8/22, Amy Albin <amyralbin at gmail.com> wrote:
> Hello All,
>
> I keep trying to learn R but keep running into problems I just can't 
> figure out how to solve.
>
> Some of the problems are specific to how to make things screen reader 
> friendly.
>
> In other situations, I'm doing exactly what the script says (bootstrap
> function) but it doesn't work! I'm being a good student and trying to 
> logically trouble shoot and type it different ways but it still 
> doesn't work!
>
> For this entire thing, I've probably spent about 2 hours on it, and I 
> don't even know how to read the regression coefficients or test for 
> homosastecity.
>
> It is so discouraging because in my field, we are expected to be able 
> to do these things! And I am dedicated! And I do put in the hours! But 
> I still can't figure it out.
>
> I am thinking of getting a tutor at my school, but they probably 
> wouldn't know how to make things screen reader friendly.
>
> Can someone hear read the files I attached and give me suggestions?
>
> Thanks!
>
> Best,
>
> Amy Albin
> Pronouns: she/her/hers
>

_______________________________________________
BlindRUG mailing list
BlindRUG at nfbnet.org
https://apc01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fnfbnet.org%2Fmailman%2Flistinfo%2Fblindrug_nfbnet.org&data=04%7C01%7CA.J.Godfrey%40massey.ac.nz%7Caae13b3ecc8c460cb34408da012dc2d3%7C388728e1bbd0437898dcf8682e644300%7C1%7C0%7C637823593310667243%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=xtp%2BezkmSTxxDcywnGPZo8cBelKzAZjv6e%2BIODGjWUs%3D&reserved=0
To unsubscribe, change your list options or get your account info for BlindRUG:
https://apc01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fnfbnet.org%2Fmailman%2Foptions%2Fblindrug_nfbnet.org%2Fa.j.godfrey%2540massey.ac.nz&data=04%7C01%7CA.J.Godfrey%40massey.ac.nz%7Caae13b3ecc8c460cb34408da012dc2d3%7C388728e1bbd0437898dcf8682e644300%7C1%7C0%7C637823593310667243%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=Dg%2BdkWwtt5Adc1IElWT8icj0RYdxPnZV7nGGSYXx8bY%3D&reserved=0
The list archive can be viewed at:
https://apc01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.nfbnet.org%2Fpipermail%2Fblindrug_nfbnet.org&data=04%7C01%7CA.J.Godfrey%40massey.ac.nz%7Caae13b3ecc8c460cb34408da012dc2d3%7C388728e1bbd0437898dcf8682e644300%7C1%7C0%7C637823593310667243%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=tXHtLb%2BD0%2FBZYKw2jr7aW7IUYB5w6inac9895q4x6bo%3D&reserved=0
More information and useful links about using R as a blind person can be obtained at:
http://R-Resources.massey.ac.nz

Look for help using R commands by reading the accessible e-book "Let's Use R Now" compiled by Jonathan Godfrey at:
http://R-Resources.massey.ac.nz/lurn/front.html