By valid, we mean that the data should be an authentic representation of what occurred throughout the project.
You should also ensure that the data you are looking at relates specifically to the project being reviewed, and that you are not collecting or using information outside your terms of reference.
Reliable data is information that can be looked at by a number of different people who would all understand it in the same way (even though they might draw different conclusions from it).
For example, earned value management is a reliable technique that tracks project performance, ensuring that the team have a common understanding of what 10 or 20 per cent behind schedule means.
We all might take different actions based on that information; however, the data itself is not in dispute.
Current means that the information you gather reflects the most up-to-date impressions or understanding of the project.
Reviewing different configurations of the project plan ensures currency of the data; whereas, a project that has not adequately tracked delivery could be said to have a non-current understanding of its performance.
Finally, sufficiency means that you should gather data until you have enough information to draw reliable conclusions and make high quality recommendations.
Quite often, multiple data sets will throw up conflicting information.
For example, two stakeholders might have very different impressions of project success, while the project plan tells a different story again.
The skill of the forensic reviewer lies in triangulating these multiple points of data to arrive at the ‘true’ picture of what went on in the project.