Tweet
Dear Biomech-L Recipients
I am writing this e-mail due to mass confusion, inside my head.
I am involved in various studies involving 3D analysis of movement,
primarily looking at full body motion analysis in sport.
Within my data, it is possible to analyse 100's of kinematic and
temporal variables as well as various muscle activation patterns from
collected EMG data
Reading many texts and Journals on the subject of statistics has left
me confused over the use of Bonferroni Type corrections (classic style
or Holm's step down proceedure) on my data. Within articles there
appears to be much confusion over which methodology to use and whether
Bonferroni Type adjustment of this type of data set, would be too
conservative.
A case in point, 10 players, 30 attempts at a set task each. Outcome
simply good or bad (maybe hit or miss), 100 measured kinematic and
temporal variables. When testing for significance of each variable
against outcome (hit or miss) i.e. using ANOVA, manual Bonferroni
adjustment would suggest 0.0005 (.05/100) significance level, but does
that overly inflate Type II error?
My question is I think a simple one, what solutions do biomechanists
use to analyse multiple dependant variables for significance, reducing
Type I and Type II errors?
Thanks in advance
P J Brooks
MPhil/PhD Student
University of Greenwich
Centre for Sport and Exercise Sciences
--------------------------------------------------------------------
University of Greenwich, a charity and company limited by guarantee,
registered in England (reg. no. 986729). Registered office:
Old Royal Naval College, Park Row, Greenwich, London SE10 9LS.
--------------------------------------------------------------------
I am writing this e-mail due to mass confusion, inside my head.
I am involved in various studies involving 3D analysis of movement,
primarily looking at full body motion analysis in sport.
Within my data, it is possible to analyse 100's of kinematic and
temporal variables as well as various muscle activation patterns from
collected EMG data
Reading many texts and Journals on the subject of statistics has left
me confused over the use of Bonferroni Type corrections (classic style
or Holm's step down proceedure) on my data. Within articles there
appears to be much confusion over which methodology to use and whether
Bonferroni Type adjustment of this type of data set, would be too
conservative.
A case in point, 10 players, 30 attempts at a set task each. Outcome
simply good or bad (maybe hit or miss), 100 measured kinematic and
temporal variables. When testing for significance of each variable
against outcome (hit or miss) i.e. using ANOVA, manual Bonferroni
adjustment would suggest 0.0005 (.05/100) significance level, but does
that overly inflate Type II error?
My question is I think a simple one, what solutions do biomechanists
use to analyse multiple dependant variables for significance, reducing
Type I and Type II errors?
Thanks in advance
P J Brooks
MPhil/PhD Student
University of Greenwich
Centre for Sport and Exercise Sciences
--------------------------------------------------------------------
University of Greenwich, a charity and company limited by guarantee,
registered in England (reg. no. 986729). Registered office:
Old Royal Naval College, Park Row, Greenwich, London SE10 9LS.
--------------------------------------------------------------------