Dear Allstats,
I have a set of data from a recently carried out survey of 100
volunteers.
The aim of the study was to assess the adhesion & irritancy of 5
different types of sticky plasters.
Each volunteer was instructed to apply each plaster and remove it
after 24 hours and then instructed to complete a questionnaire
(Questions were regarding the ease of application/removal,
comfort, irritancy etc etc using five point rating & intensity
scales). The order of application of the plasters were randomly
assigned to each volunteer, with a rest period of 24 hours
between any two applications.
A typical table of results for say easy of application is:
Plaster 1 Plaster 2 Plaster 3
Plaster 4 Plaster 5
Very Easy
Easy
Neither Easy/Difficult
Difficult
Very Difficult
>From the statistical analysis I want to determine if there any
significant differences in the 5 different types of plasters.
The typical Chi squared test for a r X c frequency table cannot
be considered as one of the variables (rating scale) is ordered.
Also need to take into account the fact that each volunteer
applied each of the 5 plasters and therefore have dependent
samples.
Am I correct in assuming the non-parametric Friedman's Test would
be appropriate when trying to determine if there are any
significant differences between the 5 different types of
plasters. If there are significant differences then multiple
comparison tests can be performed to compare between each pair of
plasters.
Any suggestions or guidance in the analysis would be greatly
appreciated.
Thanks
Prafull
*****************************************
Prafull Mistry
Medical Statistician
Nexan Telemed Ltd
Tel: (01223) 713500
Fax: (01223) 713501
e-mail: [log in to unmask]
*****************************************
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
