AARR(2/3): how to calculate the AARR
The standard calculation method, based on an exponential function, is well described in the UNICEF note for estimating AARR (see additional information below).
The formula is relatively simple but the result will highly depend on the parameters that are used to generate the AARR.
- AARR is a reflection of the trend of prevalence. If you use the data from the past 5 years or the past 15 years, the trend will differ and the AARR will not be the same. When generating a AARR value, the period under consideration needs to be exactly defined.
- AARR will obviously depend on the surveys included in the model. International initiatives usually rely on all the surveys of the JME (Joint Monitoring Estimates) database. But the EU relies on the EU database which is slightly different. The JME database may be slightly different from the database of National Statistical Offices. A decision must be made on whether to include or exclude some surveys depending on the data quality criteria (refer to the guidance note on data quality, section 3.3).
To calculate the “required AARR to reach the target”, additional parameters must be included in the model:
- The required AARR will obviously depend on the target (WHA targets; national target; targets from the Multisectoral Plan of Action for Nutrition…). This needs to be specified.
- If the target is a ’Number of children’, the required AARR will depend on the estimated number of children at baseline and at the time of the target. Different demographic models can be used, each with different scenarios and each are updated regularly. The model used need also to be specified. International initiatives usually use the WPP model (see additional information below).
When generating the AARR and the “required AARR”, 4 parameters must be specified:
- The time period under consideration
- The nutrition surveys which are included in/excluded from the calculation
- The target
- The demographic model (in case the target is a number of children)
A simple Excel spreadsheet or statistical software can be used for the regression to calculate the AARR.