Higgins et al developed a new quantity called I-squared which is a measure of inconsistency in studies included in a meta-analysis. The authors are from Institute of Public Health at Cambridge and Institute of Health Sciences at Oxford in England.
Parameters:
(1) Cochran's Q for heterogeneity
(2) number of studies in the meta-analysis
degrees of freedom =
= (number of studies) - 1
I-squared =
= MAX(0,100%*(1 - ((degrees of freedom) / Q)))
where:
• A negative value is set to 0.
Interpretation:
• minimum I-squared: 0%
• maximum I-squared: 100%
• The higher the I-squared the greater the inconsistencies between the studies.
• A meta-analysis with an I-squared of 0% is composed of studies that are consistent with each other.
To read more or access our algorithms and calculators, please log in or register.