What statistical test to use?

#1
Dear,

I am Mathieu Bossuyt, a last year dentistry student at the University of Ghent, Belgium. I have some difficulties finding the right statistical test for my master thesis and was wondering if you could maybe help me, would appreciate this a lot.

1668859780355.png
Study flow, showed above is used. The data collected at different time points is tongue posture (in sagittal and frontal view) and swallowing pattern. Following distinction is made:

1668859780367.png 1668859780378.png
  • Swallowing pattern -> somatic pattern vs visceral swallowing (2 options)
  • Research question -> groups have to be compared in time, but also horizontally.
    ‘Is there a change in tongue posture and tongue function after a unilateral posterior crossbite?’
I am stuck with al the variables, groups and different time stamps. Could you please help me out?

Kind regards
Mathieu Bossuyt
 

Karabiner

TS Contributor
#2
What exactly do you mean by "finding the right statistical test"?
The situation looks a bit more complex, and most of it is not explained to the readers.

By the way, is this a thought experiment? Experimentation with humans usually requires involvement of an expert in data analysis.

With kind regards

Karabiner
 
Last edited:
#4
What exactly do you mean by "finding the right statistical test"?
The situation looks a bit more complex, and most of it is not explained to the readers.

By the way, is this a thought experiment? Experimentation with humans usually requires involvement of an expert in data analysis.

With kind regards

Karabiner
A little bit more information about the study flow:
"All subjects included are evaluated for tongue posture and function. The children with an infantile swallowing pattern are randomly assigned to two groups: The experimental group (EXP + OMT) (receiving orthodontic maxillary expansion and OMT) or the control group (EXP) (receiving orthodontic maxillary expansion only). The children presenting a normal swallowing pattern, are assigned to a third group (CtEXP) (orthodontic maxillary expansion only).
Data registration takes place at different timepoints. The first registration moment is at baseline before any treatment has taken place (pre-treatment, T0). After the active expansion phase and before start of the retention phase, with or without OMT, new data is collected (post-treatment, T1). Once OMT is completed (this takes on average 10-20 weeks) and the patient has been in retention for at least 3 months the post-retention data were collected (T2). The data that are collected at the before mentioned time points, is tongue posture and -function. Finally, after a minimum of 6 months, new data is collected (T3) to evaluate the stability of the treatment result. The date collected for given time points is tongue posture and tongue function. The study flow is shown schematically below."

I want to look at significants, inter-rater reliability... What would you suggest to do with the data?

Kind regards
 

Karabiner

TS Contributor
#5
Ok, thank you, this is very much clearer now. So, at baseline, children differ
with respect to the outcome variables (sagittal and frontal/transverse)?

Groups can be compared at each data collection point (T1, T2, T3, respectively)
using Chi² tests (for the group x outcome crosstable).

Admittedly, I don't know how a repeated-measures design with a multicategorical
dependent variable (or two of them) can be analysed in such a way that all information
can be fully included. Or, is there something like a binary primary outcome criterion in
the sense of "treatment goal achieved yes/no" or something like that?

Again, I'm a bit surprised that experimentation with children is seemingly carried out
without a statistical expert on board.

With kind regards

Karabiner