# Hypothesis test for normality on the TI-84?

1. Jan 9, 2012

### moonman239

I'm trying to run a hypothesis test to see if a data set follows a normal distribution. I don't want to look at a plot. Is there any way? I tried to build a program that uses the built-in normal distribution functions to calculate how far the points on a normality plot would fall from forming a straight line, but I get errors.

Any ideas?

2. Jan 9, 2012

3. Jan 9, 2012

### moonman239

Here is the code I am using. Instead of filling List 3 with the normal CDF values, it fills List 3 with a bunch of 0's.

Code (Text):

SortA(L1)
N->dim(L1)
For(I,1,N)
(I/N)->L2(I)
normalcdf(0,L1(I),mean(L1),stdDev(L1))->L3(I)
End

4. Jan 9, 2012

### Staff: Mentor

Shouldnt the normalcdf be outside loop?

5. Jan 10, 2012

### moonman239

I don't think normalcdf will take lists for X values. But I will try that as soon as I can.
As it happens, I created another macro that just calculates the cumulative frequency. That one does work. I know how to have my normalcdf macro execute my cumulative frequency macro. For the reference, I just press "PRGM," in the EXEC highlight that macro and press "ENTER," that big button on the bottom-right.

6. Jan 13, 2012

### moonman239

Update: I changed the code.

Now it's this:

Code (Text):
prgmCUMFREQ
(L1-mean(L1))/stdDev(L1)->L3
For (I,1,dim(L1))
normalcdf(-1e99,L3(I),0,1)->L4(I)
End

Now it won't give me those zeroes.

But now I have another problem. When I fill in data from my statistics textbook, I get a different p-value than what my statistics textbook reads. I am reasonably certain I did not mistype any data.

UPDATE: Here's the code for my cumulative frequency macro:

Code (Text):
PROGRAM:CUMFREQ
SortA(L1)
For(I,1,dim(L1))
I/dim(L1)->L2(I)
End