## Definition

Parity indices require data for the specific groups of interest. They represent the ratio of the indicator value for one group to that of the other. Typically, the likely more disadvantaged group is the numerator. A value of exactly 1 indicates parity between the two groups.

## Data source

The sources are the same as for the underlying indicators for this goal.

## Source definition

UNESCO Institute for Statistics.

## Calculation method

Parity index:
The indicator value of the likely more disadvantaged group is divided by the indicator value of the other sub-population of interest.
Adjusted parity index:
The parity index does not indicate whether improvement or regression is due to the performance of one of the groups.
The default calculation method for the parity index yields an indicator that is not symmetrical around 1 and that has no upper limit. This limitation can be overcome with a simple transformation, by inverting ratios that exceed 1 and subtracting them from 2. This adjusted parity index is symmetrical around 1 and lies in the range 0-2, which makes interpretation easier.

## Data required

The indicator vales for the sub-populations of interest.

## Interpretation

The further from 1 the parity index lies, the greater the disparity between the two groups of interest. For indicators that should ideally increase in values (e.g. gross enrolment ratios, completion rates, participation rates, etc), a parity index value less than 1 indicates disparity in favour of the advantaged group and a value higher than 1 indicates disparity in favour of the disadvantaged group. The interpretation of the parity index is the other way around for indicators that should ideally approach 0%, like out-of-school rates.

## Limitations

See section "Calculation method"
Starting in September 2020, all parity indices are calculated with the formula for adjusted parity indices.

## Purpose

To measure the general level of disparity between two sub-populations of interest with regard to a given indicator.

## Types of disaggregation

None because the parity indices directly compare two sub-populations of interest.