Robert Holmes wrote:Pattern is a necessary condition for causality but it is not sufficient. You also need relevance.Nicholas Thompson wrote: It depends what you want to predict and from what data. If you measured both Santa Fe and south Florida, you might get data that looked like: Bell SF No Bell SF Bell FL No Bell FL Attack 0 0 1 5 No Attack 0 0 10 5 If, overall, you wanted to measure the effectiveness of bell ringing, a non-random relationship appears likely in this data (pvalue of 0.06 for a one sided Fisher exact test). If Nick wants to avoid being attacked by a croc, in general he should ring that bell. But add in the extra parameter of where, and the pvalue is 1 for Santa Fe in isolation. |
============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org
