Controlling for Multiple Tests. Issue 7 (August 2015)

• Record Type: Journal Article
• Authors: Maltenfort, Mitchell G.
• Is Part Of: Journal of spinal disorders & techniques. Volume 28:Issue 7(2015)
• Publication Date: 2015-08
• Subjects: Spinal cord -- Diseases -- Periodicals; Spinal cord -- Surgery -- Periodicals; Spinal Diseases -- therapy -- Periodicals; Cordotomy -- methods -- Periodicals; Spinal Cord Diseases -- therapy -- Periodicals; Spinal Fusion -- methods -- Periodicals; Spine -- surgery -- Periodicals; 616.73
• DOI: 10.1097/BSD.0000000000000301
• Languages: English
• ISSNs: 1536-0652

Abstract

<abstract> <title> <x xml:space="preserve">Abstract</x> </title> <sec> <p>In a statistical test, the <italic>P</italic>-value is the likelihood that the observed results are just due to chance, and by convention a <italic>P</italic>-value of &lt;0.05 is considered significant. However, this means that if a study performs 20 individual statistical analyses, assuming the null hypothesis is true in all of them, we expect one to fall below the <italic>P</italic>&lt;0.05 threshold and give us a false positive. Advanced statistical tests should be used to prevent this from occurring.</p> </sec> </abstract>