Clinical Biostatistics: Lucky for Spurs? Suggested answer 4

This website is for students following the M.Sc. in Evidence Based Practice at the University of York.

Is it "Lucky for Spurs if the year ends in one"?

Whether Totenham Hotspur won a trophy by last digit of year at the end of the season
Ends in a “1” Doesn’t end in a “1” Total
Trophy 8 8 16
No trophy 1 73 74
Total 9 81 90
Fisher's exact test: P<0.00001
(1915-18 and 1940-45 excluded)

**Whether Totenham Hotspur won a trophy by last digit of year at the end of the season**
	Ends in a “1”	Doesn’t end in a “1”	Total
Trophy	8	8	16
No trophy	1	73	74
Total	9	81	90
Fisher's exact test: P<0.00001
(1915-18 and 1940-45 excluded)

4. Could we use a Bonferroni correction to adjust this P value?

Suggested answer

The Bonferroni correction for multiple testing puts a lot of tests of significance together. We test a composite null hypothesis by saying that if any individual test has a sufficiently low P value, there is evidence against the composite.

With a data-generated hypothesis, we cannot say how many different tests there might be. It would not just be the final digit of the year, but any conceivable pattern of years.

We cannot, therefore, apply the Bonferroni correction.

Back to question.

Back to Clinical Biostatistics index.

To Martin Bland's home page.

This page maintained by Martin Bland.
Last updated: 31 July, 2006.