## Monday, October 18, 2010

### Post Hoc, Ergo Propter Hoc

Post Hoc, Ergo Propter Hoc
Literally: "After this, therefore because of this":
This type of false cause occurs when the [project manager] mistakenly assumes that, because the first event preceded the second event, it must mean the first event caused the later one. Sometimes it does, but sometimes it doesn't

In fact, this error--false cause--is a common confusion between correlation--that is, a relationship--and cause-and-effect [causation].  The latter requires the former, but not the other way 'round.

Correlation
Correlation merely establishes a relationship but not necessarily a causation between two activities, conditions, or events. Correlation can be tested and measured from empirical obserations by mathematical analysis. The analysis can determine the degree of correlation [strong or weak, for instance] and the direction [one goes up when the other goes down]. Correlation usually occurs in a close proximity of time when circumstances affect all elements.

One commonly accepted idea in project management is that most tasks in sequence in a schedule have some degree of correlation--that is, the metrics of one task, like duration +/-,  may be observed in the next, if for no other reason than they share the effects of a common project environment of policy, doctrine, values, and resource talents/capabilities in a common temporal framework.  Many others, not in direct sequence, may also have some degree of correlation.

So it is common to construct a correlation matrix with one task on the vertical and the other on the horizontal of the matrix. [Take notice I do not call this a 'causation matrix'].  In many projects, it's 'doctrine' to assume about 0.2 positive correlation of behavior between tasks. As a practical matter, as observed in Monte Carlo simulations, such a correlation will spread the variance a small amount.  However, the purpose of the simulation is to show effects of similar behavior in the same timeframe, not to establish cause-and-effect between tasks.  See Stephen Book's and Raymond Covert's presentation on page 162 of this deck.

Causation
Causation is the greater error for project managers. The true cause the true effect are correlated. But who knows what the real cause is for some observed effect?

The error PMs make is to assume that because certain conditions existed in some other successful project, then simply replicating those conditions should bring success again. MAYBE!, but maybe not.

Causation--particularly causation that repeats--is devishly hard to prove without controlled trials and careful consideration of all the circumstances. Think of all the Malcom Baldridge winners that subsequently failed in business.  And, think of all the number of level 4 or 5 maturity model program offices that turn in failures.  The error of false cause is especially troublesome when importing one methodology or set of practices from one situation to another. [See: Agile]. It's usually a lot more than just the practices, so don't be fooled into a false cause-and-effect story!

Assume relationships, but doubt Post Hoc, Ergo Propter Hoc.

Bookmark this on Delicious