Risk management - Simple estimation problems
Simple estimation problems
Obtaining good estimates for particular activities is not easy.
Honesty and integrity
Many people have self interest at heart when making estimates. Human nature tends to make personnel overestimate so as not to look foolish if a lower estimate proves incorrect. Use of experienced personnel and consultants will alleviate the problem. It is good practice not to offer either directly or indirectly any incentives when estimating.
Even if persons are truthful with no ulterior motives estimation can vary greatly from person to person. Every individual will base his estimate on personal experience or data which in itself contains particular assumptions.
Politics can create hurdles. Putting a colleague in an unfavourable light may not be seen as a good career move.
Anchoring
People tend to be influenced by what they already know.
For example, if someone who is very confident puts forward that a delay will be 15 months then other expressed views may be similar. It may be hard for someone to ‘challenge’ this by suggesting an apparently ‘ridiculous’ value of 4 months.
In a situation where an expected value has been suggested perception of higher and lower estimates will tend to be narrow around the original expected estimate. The simple system of taking the estimation through additional steps to refine the upper and lower ends is designed to avoid this.
Activity dependency
When you have a chain of activities which are running in series it is very easy to underestimate the risk of failure somewhere along the line.
Let us say you have 6 activities leading to a particular goal and each one has a 98% chance of success. These seem very high.
However, all 6 will need to complete to achieve the objective. The probability of all 6 being successful is:
0.98 x 0.98 x 0.98 x 0.98 x 0.98 x 0.98 = 0.886 or 88.6%.
Whilst still quite high this leaves a margin of over 11% for failure!
Recent events
If a similar experience has recently occurred this may influence estimation if personnel do not want to be ‘caught out again’.
For example, if a particular risk on another project was judged to have a 10% chance of materialising and it did occur with a resultant delay of 20 weeks it may be hard for people to accept a in a current project. This particular risk suddenly seems more likely to occur.
If the same risk was identified in the current project there may be a tendency to increase the potential likelihood to say 20% (which is illogical) and record a potential delay of 20 weeks.
This may be valid when reviewing the details of the occurrence but it may be equally valid that the expected delay will be only 10 weeks based upon the experience of handling the risk occurrence before.
Even similar events recently experienced are likely to influence the perception of the risk.