# Thread: tests for normal data

1. ## Re: tests for normal data

If you accept that the power is weak, then there certainly is no rule against using them. If there was they would not exist Anderson-Darling is also used a lot.

2. ## Re: tests for normal data

Originally Posted by noetsi
If you accept that the power is weak, then there certainly is no rule against using them. If there was they would not exist Anderson-Darling is also used a lot.
I know about the power, but I don't have more data. I'm trying to do the best that is possible with the data that I have...

Thank you very much

3. ## Re: tests for normal data

Hi again.
Anf if my data is
Code:
`` 0   0   0   0   0   1   1   1   1   1  1   1   1   0   0   0  0  1  0  0  0  0  0   1   1   1   1   1   1  1   1   1   1  1   1   1   1   1   1   1   1  1  1  1  1  1  1  1´``
it's correct to apply a shapiro.test to test the normality of data?

4. ## Re: tests for normal data

Here's a hint - if the only values your data can take are 0 and 1 then the data itself is NOT normally distributed...

5. ## Re: tests for normal data

Hi.

I can have values like

Code:
`` 0   0   0   0   0   1   2   4   1   1  4   2  1   0   0   0  0  1  44  0  25  0  0   1   1   1   1   1   1  1   1   1   1  1   1   1   1   1   1   1   1  1  1  1  1  1  1  1´``
It's good to apply a shapiro.test?

6. ## Re: tests for normal data

It doesn't really matter. Your data isn't normal if it's like that. Once again I'll mention that it looks like you're working with count data. Maybe you should just look into a method that's appropriate for for you data instead of trying to figure out if your data is "normal" (Because it's not)

7. ## The Following 2 Users Say Thank You to Dason For This Useful Post:

GretaGarbo (05-06-2013), hlsmith (05-06-2013)

8. ## Re: tests for normal data

Originally Posted by Dason
It doesn't really matter. Your data isn't normal if it's like that. Once again I'll mention that it looks like you're working with count data. Maybe you should just look into a method that's appropriate for for you data instead of trying to figure out if your data is "normal" (Because it's not)

But the problem is that I have a lot of data like this. And I know to test the normality for each one, because after I will apply the test automatically. I can't look for 20000 groups of data to apply the test... You understand?

9. ## Re: tests for normal data

If this is count data then for a large enough mean the data will probably be approximately fairly well by a normal distribution. But like I mentioned before if this is count data then why are you so opposed to using a method that is more appropriate for count data? Something like a negative binomial regression or an overdispersed poisson regression. You might even have some zero inflation that you want to deal with.

But really knowing more about your data is the first place to start.

10. ## Re: tests for normal data

I'm not opposed to using other methods, I only didn't understand how can I use that. If you could help me how to do, thank you very much.
My pproblem is that I need to do a program to do that becausa I hava a lot of rows with data (a lot of groups of data) and I need to create some rules to apply my tests.
Thank you very much.

11. ## Re: tests for normal data

Are you working at a university? Could you possibly contact a statistician in your area and try to arrange a face to face? You've asked the same question many times and our advice has been the same every time. We're more than happy to help if you put some effort in but you just seem to ask the same question over and over. I offered up a few alternatives but we aren't going to run the analysis for you. If you have a local statistician you could meet with then that would be best since it would be a lot easier for them to get an idea about your data and help you out.

12. ## Re: tests for normal data

Ok, thank you