What are type I and type II errors?

A: The null hypothesis is a cornerstone of scientific experiments. It is the hypothesis that an experiment tests. For example, if a hospital is testing a new drug, the null hypothesis will be that the drug has no effect.

In this scenario, scientists are vulnerable to making two kinds of mistakes. One, the type I error, happens when a scientist rejects a null hypothesis that is actually true: they conclude the drug works when in fact it does not. This is called a false positive. In a type II error, a scientist does not reject a null hypothesis even though it is actually false, i.e. a false negative.

There are different ways to measure, and prevent, these mistakes. Often the most effective one is to ensure all involved parties have more information during the test. This way, scientists do not have to work with a too-tight false positive limit or a too-loose false negative error limit.

The Election Commission of India (ECI) is currently conducting a ‘Special Intensive Revision’ (SIR) where both these errors are in play. Say a registered voter is eligible. A false positive error happens if ECI officials wrongly remove that voter’s name from the electoral rolls. A false negative error occurs if officials include a name on the rolls whereas that person is ineligible to vote (e.g. if they have died).

A good way to reduce both types of error rates would be for all stakeholders — including the ECI and the people — is to have more information, e.g. data from the new Census, opportunities to challenge decisions, and independent audits.