Hi all,
I am not really good at stats and I have a question to ask;
I had two study groups of mice, one treatment and one non-treatment, with 6 mice in each group. I used student's t-test to calculate if there is a statistically significant difference between these groups in their body weight (using 2-tailed, equal variance).
However, today one mouse in the treatment group died, which has led to n=5 for treatment group and n=6 for non-treatment group. In this case, can I still use student's t-test with 2-tailed and 'unequal variance'?
No, Google Welch's test. That is what you want. Unequal sample sizes, normal pops. Use Satterthwait approximation for your critical value.
see the link below for details.
https://www.google.com/search?q=satterthwaite+approximation&ie=utf-8&oe=utf-8&client=firefox-b-1
Best wishes, David Booth
Hello Yinan,
I agree with Mirko. Your problem is probably not unequal sample sizes, but rather the sample size itself. You carefully need to check for normal distribution with such a small sample size and otherwise go for a non-parametric test like Mann-Whitney-U.
However, if it is indeed only about variance inhomogenity, you can go for Welch's test, as David already suggested.
Best,
Steffen
Agree with answers above. I would for non parametric one can do t-test with bootstrap confidence intervals which spss offers
Thanks for the answers everyone.
I am a bit confused about the term equal/unequal variances.
For example, if I want to know the difference in tumour volume between treatment and control group, is the tumour volume considered equal or unequal variance? Can anyone explain the difference between these terms without using technical terms?