0
$\begingroup$

I have the following data: \begin{matrix} &Test Sample\ 1 & Test Sample\ 2 \dots & Test Sample \ 6& Control Sample\ 1 & Control Sample\ 2 \dots & Test Sample \ 6\\ measurements& 100 & 90 & & 30 & 20\\ \end{matrix}

If I want to normalize and compare between the test and control, should I normalize test and control samples separately or the whole matrix ?

Improve this question
$\endgroup$
1
  • $\begingroup$ When you are doing experiments I.e. generating experimental data why should you normalize ? $\endgroup$
    – Subhash C. Davar
    Commented May 23, 2020 at 7:57

1 Answer 1

Reset to default
1
$\begingroup$

An experimental sample is a group that receives the variable being tested in an experiment. The control group is the group in an experiment that does not receive the variable you are testing using the same experimental procedure. Therefore the control group, receiving no intervention, is used as a baseline to compare groups and assess the effect of that intervention.

I gave this as an object lesson to show that test and control must be linked by a similar experimental procedure. So yes, if you were to normalize then use the entire matrix.

Additional point (bonus round)

The purpose of data transformations is to make data easier to model and understand. One purpose of transforming data is to find outliers, reduce the amount of skew, etc.

Looking at a small part of your data, it seems like the values are between 0-100, Suggesting that normalization is not entirely necessary, IMHO.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.