We could use the Bonferroni method. This would require us to calcuate the
number of relevant tests done, here at least 16. We would then multiply the P
values by this. If any is P value is then below 0.05, the relationship is
significant. If our initial hypothesis were specifically about ischaemic heart
disease, we would have four tests only, and the relationship between IHD and
back pain in younger men would have P= 4 times 0.02 = 0.08, not significant (* Intro* 9.10).
Back to question.

This page maintained by Martin Bland.

Last updated: 8 April, 2004.