I could not find a question related to my problem so I am asking in a new thread. If the question has been asked before please direct me to the concering thread.
I have no experience in statistics so my question might be easy to answer/stupid.
So i performed an experiment measuring the expression of a gene in Wildtype (WT) and Knock-Out (KO) Cells. Upon analyzing the data, with Graph Pad Prism 8, i tested for normal distribution first and found that my WT-samples are non-gaussian and my KO-samples are gaussian. I now want to compare these two samples, but which test would be appropiate for that? Do i need to use a non-parametric, like the Mann-Whitney test, since one sample has a non-normal distribution? Can they be compared properly?
Thank you very much in advance,
Adrian
I think the Mann_Whitney test can be applied in this case.
Regards
Mr. Fuchs,
Please, consider the references [126,127] cited in work [1] shown, below. These literature sources are among the best written statistical analysis for non-mathematicians.
[1] Steroids, 164 (2020) 108750
Stochastic dynamic mass spectrometric quantification of steroids in mixture — Part II; Bojidarka Ivanova, Michael Spiteller
Thanks for the quick answers. I have one more question concerning the Mann-Whitney-Test.
Due to the n numbers of my experiments beeing low (n=3), the experiments were very time and resource consuming, i wanted to avoid using the Mann-Whitney-Test.
" If you have small samples, the Mann-Whitney test has little power. In fact, if the total sample size is seven or less, the Mann-Whitney test will always give a P value greater than 0.05 no matter how much the groups differ. " - Citation of the Graph Pad Prism 8 Statistical Guide
Can I justify that using the t-test, because it is pretty robust to violations of normality? What makes the t-test robust?
Mr. Fuchs,
From your last question I can deduce that you understand that the t-test evaluates the normality of the datasets. But, in fact, the t-test is applicable to normally distributed two or more sets of variables. It evaluates the equality of groups of variables.
Next, the Mann-Whitney-Wilcoxon test is applicable to independent sets of variables. Do your datasets of variable independent? Which method you use in the latter context?
In addition: Could you provide statistical parameters from W-statistics of your data? It is applicable to small sample size.
As well as, the t-test is one of the first tests, which is provided to the basic education for BSc students from almost all disciplines of the Natural Sciences at the Universities in EU. Could you provide more detail on your basic educational discipline; in addition to, some detain on your current position at Friedrich-Alexander-University of Erlangen-Nürnberg?
Sandra
A escolha do teste mais apropriado depende de vários fatores:
. tamanho da amostra.
. homocedasticidade ou heterocedasticidade.
. na amostra com distribuição não-normal é preciso perceber o grau de enviesamento (assimetria, curtose e gráfico).
Espero ter ajudado.
Saúde,
DS
- Badges
- Science topic