I think correlation or regression analysis may be useful

Hi Paul,

I get  right back to your questions.

1. we want to test  a polymer formulation on the probands lower arm.The other arm serves as a control, getting the formulation without polymers. The scoring discribes the skin reaction  during  the testing  period,which is 4 weeks.

2. see 1

3.it is a intraindividual test, 20 subjects (control and  test on one arm accordingly)

4. good question.  We did not figure that out, However I assume,we will colled the scroing data after two weeks and at the end after 4 weeks.

5. there is no condition, we are testeing the tolaranceof  a new excipient not of an active compound

6.actually we don´t. The polymer is beeing used in cosmetics since several years

7. I do not know something about a baseline,sorry, I´m new to this.

8. We think not.

9.The hypothesis is, that we will not find a different response  to the polymer formulation compared to the formutlion without polymer. We also expect a no to mild reaction in the test person. In IGA scoring it would be a 0 or a 1 to 2, depending on the scaling maximum (0-5 or 0-7)

thank you  so much for taking the time.I really appriciate it! Best Diana

Dear Paul,

thank you so much for taking the time!!! I will discuss all you points and suggestions with my bosses today. We, unfortunatly, are not as experienced in solving statistical tasks, actually not experienced at all. I hope I can get back to you in case I get stuck somewhere?

Have a great day and thanks again!

Kind regards

Diana

Hi Diana,

I am impressed by Pauls' exhaustive analysis of your challenge and only want to add one probably minor comment:

You state that: "9.The hypothesis is, that we will not find a different response to the polymer formulation compared to the formutlion without polymer." as Pauls response already implies the Hypothesis is actual a Null-Hypothesis which can be rejected based on a certain alpha-level. However, if this is indeed your Hypothesis it is not possible to show this with inferential statistics, as a p-value larger than the alpha-level is not a prove of the Null-Hypothesis. Some much more competent people describe this here: http://mathnstats.com/index.php/hypothesis-testing/112-reject-or-fail-to-reject.html very nicely. I hope this is somewhat useful.

Cheers, Markus

Hi Paul,

Thank you for your warm wishes I just had to follow up on your last comment. All the best from little less but still cold Germany and a wonderful Thanksgiving to all of you.

Cheers, Markus

Dear Paul,

I had some time to digest most of your thoughts and suggestions, some of them did not enter my active brains site or at least it did not stick there, an area that was not much used for statistical questions (-:.

However, you tried to include the time here as a vector and this is indeed something I considered to do by using a repeated meassure Anova, as you suggested. Since the timing and the corresponding effects are not the most important point for us, we might also go with  the Wilcoxon sighned rank test, but only if we collect two data sets (after 2 and 4 weeks).

However it looks like we might have 4 data sets or data collected at 4 different time points

day 0= Skin condition-no treatment at all;

day 1= Skin condition after 24h after first Treatment;

day 14= Skin cond...

day 28 Skin condition at endpoint

Would you include the day 0 data set in the Anova? I feel like that this might not be correct, since there is no treatment at that time point, only right after the scoring we will applicate the test solutions.

Paul, thank you again for your great effort and help on this. I did not expect that, that was very kind and helpful!

Have a great day, here it is already dark outside and I´m going home now.

Kind regards

Diana

Ps: Is there any way I will ever understand what a N-of-1 approach is? Sorry, I did not get that...

Dear Paul,

thank you again very much for your great help! I hope you had a lovely thanksgiving!

Marcus, thanks for your comment and the link. It is really nicely explain what a pickle situation one get into.

I wish you, Marcus and your Anova expert friend a happy Christmas time.

Stay well.

Diana

Ps: And Paulinus also, thank you for your suggestions.