Analyzing the results of a cloud-seeding experiment in tasmania |
| |
Authors: | AJ Miller DE Shaw LG Veitch EJ Smith |
| |
Institution: | 1. CSIRO Division of Mathematics &2. Statistics , SydneySydney and Adelaide respectively;3. CSIRO Division of Cloud Physics , Sydney, Australia |
| |
Abstract: | We considered the analysis or a randomized cloud seeding experiment of lusmania, Where an distributed analysis had been specified before the experiment was carried one. We compare “classical” regression analyses, with and without transformations, with permutation tests based upon double-ratio and variance-ratio statistics. Regression residuals are used to compare the merits of various other alternative test-statistics including some based upon gamma distribution assumptions. We conclude that more attention needs to be paid to the relative weights which various rest statistics implicitly attach to low, medium and high rainfalls, and that, irrespective ot distributional assumptions and theory, significant testing should be done using permutation tests. |
| |
Keywords: | cloud-seeding weather modification permutation tests randomization tests resi duals analysis transformations |
|
|