[R-meta] Variability ratio in studies with different designs

[Fernando Klitzke Borszcz]

Hello everyone,

I am carrying out a meta-analysis that involves studies of different
designs, as follows:

1) parallel controlled; where ES = pre to post change of control group -
pre to post change of experimental group.

2) crossover;
a) ES = pre to post change in condition 1 - pre to post change in condition
2;
b) ES = post condition 1 - post condition 2.

I am interested in investigating the variability of responses between
conditions using the lnVR (logarithm of the ratio between standard
deviations [SD]).

In this case, for the design:
1- lnVR = ln(SD change control / SD change experimental), where, SD change
is the SD of the pre to post changes;

2a- lnVR = ln(SD change condition 1 / SD change condition 2);

2b- Is appropriate in this condition to use: lnVR = ln(SD between subjects
in condition 1 / SD between subjects in condition 2)? Assuming that if the
SD between subjects in one condition was greater than in another,
individual responses were greater in one condition than in another, because
there are only parameters measured after each condition.

Thank you!

Att.

Fernando Klitzke Borszcz
Doutorando em Educação Física
Laboratório de Esforço Físico - LAEF
Centro de Desportos - CDS
Universidade Federal de Santa Catarina- UFSC

	[[alternative HTML version deleted]]