2
$\begingroup$

I really hope someone can help me as i have been reading about this stuff and i am still really confused!

I have a dataset of 180 chemical compounds with 2 data (numerical) values each - i want to see if there is a difference between these two data points for each of the compounds for the whole dataset however my data is not normally distributed so my questions are:

Am i ok to use a matched paired wilcoxon rank test? Or can i get away with using a normal paired t test?

I have had been reading alot online an i just want some checking i am understanding! Any additional help/resources will be appreciated! Thanks in advance :)

$\endgroup$
  • 1
    $\begingroup$ I am not sure if I understand your question, but if you have only two data points for each compound, you can't do anything. But I doubt that is what you have. Could you post some sample data? $\endgroup$ – Peter Flom - Reinstate Monica Feb 23 '13 at 17:21
  • 1
    $\begingroup$ If there is only one data point and that data point is randomly chosen, then there isn't any statistics you can apply. The point, however, is your best guess of the mean, so you can just perform a subtraction between the two. Whichever higher is then higher. The shortcoming of this approach is that you will not be able to obtain any confidence interval, the differences will be final, though likely inaccurate or plain wrong. $\endgroup$ – Penguin_Knight Feb 23 '13 at 17:33
2
$\begingroup$
  1. If you mean to perform 180 statistical tests to compare two data points, then there is no suitable statistics. However, if the sampling is random, then your two data points are the available best guess of the two means you'd like to compare. You may simply perform a subtraction and see which one is higher.

  2. If you mean to compare the effect of two conditions applied to 180 different chemicals. For instance, you expose 180 chemicals to i) nothing and ii) high energy environment, either heat or radiation and then measure something. Then there are tests available.

The difference lies between whether you only have n=1 in each of the two environments or n=180 in each of the two environments. Either way, you'll need to clarify the design in your question.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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