Deliberate errors and model reviews

An important part of the development of a model is a review of the model – either by a colleague or a more formal external review.

An important aspect of this is incorporating deliberate errors in the model. These errors are recorded, and the model version with these errors is clearly marked as such and not used for analysis. This provides two key benefits:

  • By finding all of the deliberate errors, the reviewer will demonstrate that they have carried out the review thoroughly and that it is likely that the genuine errors are discovered; and
  • Selecting the deliberate errors in the model provides an opportunity for reflection on model logic and assumptions.

The deliberate errors selected should reflect those which are likely to be included in the development model:

  • Formatting – that all cells comply with formatting rules.
  • Formula errors – referring to incorrect cells or ranges, using inappropriate calculations (e.g. adding rather than multiplying).
  • Logic errors – where the formula does not calculate the desired figure – for example multiplying by the incorrect currency.
  • Deviations from best practice – including hard coded values, repeating inputs etc.

Finally, before you submit a model for review, do a thorough review of the model yourself.

EconomicsLAB provides model review services. If you would like to discuss this, or any other issues in analytics and economics, please contact adam@economicslab.com.au.

 

Leave a Reply

Your email address will not be published. Required fields are marked *