Data coding
Full copies of the 14 qualified papers were obtained and
their statistical data, including age, gender, duration of the
treatment, regularity of practice, and type of anxiety test
used, were extracted, entered into the Comprehensive MetaAnalysis
Program (CMA)31 and a table, and checked by a
proofreader. Study quality was assessed by using the
CLEAR score check list.26 We were able to contact all but one author for study details on such areas as randomization
procedures, allocation concealment, and assessor blinding.
Pretest anxiety scores in percentiles were calculated for each
study group relative to norms for working adults, which was
set at the 50th percentile.{
The test-retest reliabilities of the anxiety tests for different
populations at different test-retest intervals (which
are needed to calculate the effect sizes) were taken from
test manuals, principally the STAI manual, Table 11 (page
30).32