Being a non-statistician, a geogapher working with climate change adaptation, and currently working 3 months at IIASA in Austria, I have a question related to correlation value for more than two data points, which is probably easy to answer for a spatial statistician. In my project I am looking at rainfall correlations between different regions in Ethiopia, and how they could be used to optimize a micro-insurance scheme against drought for farmers. I have proofed clearly that the less, or even negative some seasonal or annual rainfall correlation between two regions is, the more financial robust is a potential micro-insurance scheme. But what if I would like to pool three or four regions (or even more) together? The financial analysis showed already that I can get even higher financial robustness when increasing the number of regions in a pool, e.g. 3 regions or 4. But I am not able to link this to the rainfall correlations between the regions in a pool. E.g. when having sites A-B-C (each having diff. correlations, A-B, B-C, C-A), and some other sites B-C-D, how could I grasp the "combined" rainfall correlation between three sites in one value, to make insurance pool A-B-C with B-C-D comparable? This would be needed to make a regression (as I already did between two sites) to evaluate, whether lower and negative correlation between three (or even more sites) also results in higher economic robustness of the insurance scheme. I cannot simply take the mean of three correlation values (A-B, B-C, C-A) for all insurance pools and then make a regression with economic robustness, because the mean would shade a very positive and a very negative correlation between two sites when adding them up. I probably cannot weigh them either. Distance between the points is of no importance (at least not here). I cannot add them up either, b/c e.g. one positive, one neutral and one negative correlation (when having three sites) would result in zero (approximately). I would be grateful for any advise in a form which can be understood by a non-statistics expert. Eventual literature (teaching books) on this matter would be appreciated as well. sincerely Elisabeth Meze-Hausken YSSP Risk and Vulnerability IIASA International Institute for Applied Systems Analysis A-2361 Laxenburg, Austria Phone : +43 2236 807-255 Fax : +43 2236 71313
+ + To post a message to the list, send it to [email protected] + To unsubscribe, send email to majordomo@ jrc.it with no subject and "unsubscribe ai-geostats" in the message body. DO NOT SEND Subscribe/Unsubscribe requests to the list + As a general service to list users, please remember to post a summary of any useful responses to your questions. + Support to the forum can be found at http://www.ai-geostats.org/
