CPA and ’normal’ disrtribution
In the Open Standard
David BordoliMember for 24 years 6 months
- Created
- 1
- 1
0
Hello
A question, and sorry at this stage it may seem a little ‘off topic’ but if anyone knows the answer I guess it will be you risk experts.
It is generally accepted that the duration of a single activity will have a positively skewed probability distribution. Is the result of a critical path analysis also skewed or will it be normal as a result of the Central Limits Theorem coming into play? That states, basically, that the means of successive random samples taken from a population will be normally distributed whatever the underlying parent distribution.
Regards
David
A question, and sorry at this stage it may seem a little ‘off topic’ but if anyone knows the answer I guess it will be you risk experts.
It is generally accepted that the duration of a single activity will have a positively skewed probability distribution. Is the result of a critical path analysis also skewed or will it be normal as a result of the Central Limits Theorem coming into play? That states, basically, that the means of successive random samples taken from a population will be normally distributed whatever the underlying parent distribution.
Regards
David
1 reply
Oldest first- 0Philip Rawlings ·David
The results of a risk analysis of a simple set of activities in series will ’tend’ towards a normal distribution as you state. BUT the tendency is lessened if the distributions are correlated together (depending on the degree of correlation), also depends on the shape of the network.
For a simple set of activities in parallel, you will probably get a left skewed distribution due to nodal/merger bias.
A real netwoek, of course, is more complex with a mix of parallel, series activities and some correlation so the effect is muted.
Log in or join to reply
ProjectControls.com is free to join. Members can post questions, reply, vote and follow discussions.
Existing members: your original username and password still work. Forgot password?