MATLAB Answers

# Why do I get an NaN error with the ttest?

19 views (last 30 days)
Barbara Schlaepfer on 21 Jul 2019
Commented: Adam Danz on 10 Dec 2019
I am trying to do a statistical analysis. I have two sets of data and try to find out if one is always a little higher. For the analysis I also need the significance.
I was trying the ttest, but for several answers I get NaN.
How can I fix that, do I have to use a different test?
Thanks for any help.

#### 1 Comment

Adam Danz on 10 Dec 2019
ANKUSH JAMTHIKAR's answer moved here as a comment
I have two binary vectors with same elements. I mean, if X and Y are the two binary vectors, then X=Y. Now when I am running ttest(X,Y), it is giving me h=NaN and p=NaN. Does anyone know, how to handel this issue? Is there any alternate test to do this job?
I am a bigginner in statistics. So kindly help me to understand this.
Thank you in advanced!

Sign in to comment.

### Answers (2)

Star Strider on 21 Jul 2019
The degrees-of-freedom are 0. It seems you likely do not have enough data to perform a t-test (or likely any other test of individual or group differences).

#### 2 Comments

Barbara Schlaepfer on 22 Jul 2019
I have 18 data points for each set, Do you think that is not enough?
Star Strider on 22 Jul 2019
You are apparently not presenting them to the ttest function as single-precision or double-precision numeric values. (They cannot be symbolic, categorical, strings, characters, or anything other than numeric values.) There appears to be no data in your vectors that ttest can use.
EDIT — The only way I can reproduce the results you get is with:
[h,p,ci,stat] = ttest(0, 1)

Sign in to comment.

Adam Danz on 10 Dec 2019
Edited: Adam Danz on 10 Dec 2019
Rather than copying the answer to here, see this review of possible causes of NaN output in ttest().

#### 0 Comments

Sign in to comment.

### Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by