These data are from a study to determine whether a certain training 
regimen can counteract bone loss in women with postmenopausal 
osteopenia.  The data are strength measurements for five muscle groups 
taken before and after six months of training.

    LP -- leg press
    HF -- hip flexor
    HE -- hip extensor
    AB -- arm abductor
    AD -- arm adductor

It's never easy deciding what to do with such data (even ignoring the 
lack of a control group!).  One could perform five separate 
comparisons, but when you perform many tests, there's a good chance 
something will appear to be statistically significant if only because 
probability theory guarantees that 5% of the time the observed 
significance level (P value) will be less than 0.05 even when there is 
no effect.  One possibility in situations like this is to construct a 
measure of "total strength", that is, the sum of all five measures, 
too, but I notice that LP will dominate the other measures. 

Another option is the Bonferroni adjustment, dividing the preferred 
significance level by the number of tests.  In this case, since there 
are 5 tests, each would be performed at the 0.01 level of 
significance.  Another way to perform the Bonferroni adjustment is 
to multiply each P value by the number of tests.  Continuing this 
example, if the P value is multiplied by 5 and is still less than 
0.05, then the P value before multiplication must have been less than 
0.01, so the two ways of performing the adjustment are equivalent.