I'll explain briefly the experimental set-up.
- One gruop of subject measured in two conditions:
A: with treatment (Instrumented gait analysis with foot orthosis)
B: without treatment (Barefoot instrumented gait analysis)
Before testing, subjects used the treatment (foot orthosis) for different periods, this may have an impact on the results. I need to verify if there are differences between the two conditions using the treatment duration as a covariate.
I guess the main question is: what is your research question (i.e., why the Wilcoxon and what do you hope to gain with covariate)? Also, what is the sample size and the distributions for past use and whatever your response variable is.
I assume since you said experimental set-up that this was an experiment (i.e., random allocation was used).
Thank you Daniel Wright
My research question is: Does the treatment (Foot orthosis) improves gait in children who walks on their toes?
The treatment (foot orthosis) may have long term effects.
Before data collection, the children wore the orthosis for different periods (ranging from 1 to >24 months).
So the time of use of the orthosis may impact the results of my test.
I would like to consider time of use of the orthosis before testing as a covariate in order to improve power of statistical test.
Daniel Wright
sample size is 21 subjects.
I used shapiro-wilk test to chek for normal distribution and some variables do not showed normal distribution
I am not an expert on instrumented gait analysis, but a quick search suggests that it yields several measures. Which one(s) are you using as your DV(s)? For the one(s) you are using, would it be sensible and defensible to use means and SDs for descriptive purposes? Thanks for clarifying.
Bruce Weaver Thank you for you answer,
I am using spatio-temporal parameters (e.g. candence, walking speed, step length) and kinematic (maximal angular excursion) and kinetic (joint moment and power) parameters.
I would describe them with mean and SD if their are normally distributed, otherwise i would use median and IQR
Thank you for clarifying, Paolo Brasiliano. The reason for Daniel Wright's reaction to the news that you have multiple response variables, I think, is that in addition to your original question, you now have the multiplicity problem to deal with. But first things first. I recommend that you read this BMJ Stats Note by Bland & Altman:
Article Practice Statistics Notes Analysis of continuous data from s...
One of the key points is that you are not the first one to have used these variables. So do not rely on your small sample (n=21) to tell you how these variables are distributed, or whether you can use means & SDs for description. What do you know about them from the literature?
Regarding multiplicity, two of my favourite articles remain the articles Schulz & Grimes (2005) published in the Lancet. I think they are very thoughtful and thought-provoking.
Article Schulz K, Grimes DMultiplicity in randomised trials I: endpo...
Article Multiplicity in randomised trials II: Subgroup and interim analyses
I hope this helps.
Cheers,
Bruce
PS- I should have added that using statistical tests of normality as precursors to parametric analysis is a bad idea. There are many threads on RG that talk about this, and here is a short conference presentation I gave about it several years ago.
Presentation Silly or Pointless Things People Do When Analyzing Data: 1. ...
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics