Saturday, September 23, 2006

A and B are in “success”… if you misspell it

Andrew Robinson emails this example of dodgy reasoning:

I was at a conference last Friday, and two presenters made the following argument.

1) We note that certain projects are successful.

2) We examine the characteristics of successful projects, and we find A and B in common to all the successful projects.

3) Therefore A and B are desirable factors for projects.

Obviously, in failing to examine the unsuccessful projects, the reasoner misses the factors that are common and unique to all the successful projects. So, they fail to adequately distinguish the successful from the unsuccessful projects.

Does that fallacy have a label?

Talking to Jef, we agreed this is an example of the False Cause; Correlation Error (post hoc ergo propter hoc), as they haven't actually established a causal relationship between "A and B" and successful projects. To establish this they would need to alter "A and B" to see if there is any marked effect. If so, it would be reasonable to suppose that "A and B" are causal agents.

However, even if "A and B" are causal agents, the presenters are still guilty of Observational Selection; as pointed out by Andrew, they haven't looked to see if "A and B" were also characteristics of unsuccessful projects. (If they have looked, but ignored counter evidence, it would be Stacking the Deck.)

This opens up the deeper philosophical question; how can one establish a causal process? I’ll ramble on about causal processes in my next post.

Update - here's that post on causal processes.
Tags: , ,

Email via Tim van Gelder