WMOD00464 - Laserfiche WebLink

Home Browse Search

B-5 assessed by insuranl;e claims. The farmers who request crop insurance settlement are the same farmers who, through their cooperative association, sponsor the experiment. It is possible that their voluntary reports of crop loss for insurance purposes are influenced by their knowledge that a storm was seeded, their (positive) attitudes toward the hail-suppression efforts and their financial support of the study, and their assumptions concerning the effects of filing insurance claims for crop damage upon future insurance coverage. Such possible sources of bias seem inevitable when the dependent variable is influenced by the voluntary choices of interested par- ties in reporting results. * North Dakota * In the North Dakota pilot project, the random selection of no-seed days "was made in advance of the starting date each year and this information was communicated in advance to the meteorologists, pilots, and radar technicians on the project to help them in planning work schedules and equipment maintenance." Thus it appears that all personnel knew in advance whether or not an e:xperimental unit was to be seeded. To the extent that this knowledge cons- ciously or unconsc:iously influenced measurement pJrOcedures on seed or no-seed days, it represents a possibly troublesome source of bias. The use of randomization in anallysis, of course, provides no palliative for such bias. This study embodied interest both in rain production and hail suppression. On a day that had been selected for seeding, if the project meteorologist judged a storm to be approaching hail intensity on thc~ basis of reported hail on the ground or radar reflectivity beyond a predeter- mined threshold, the seeding rate was increased to the "hail suppression mode". Allowing this important change in procedures to be determined by a meteorologist's judgment at best restricts the generality of results. Were the hail suppression mode to be viewed as an important feature of the study, it would have been advisable to establish totally objective criteria for that mode, in advance, and to remove subjective factors from exerting influence on the decision. Crop insurance claims were used in the North Dakota study to help evaluate hail With all personnel aware of the identity of seed, no-seed days and of whether or not seed dllY~, were placed in the hail suppression mode, there is the possibility of bias from conscious or uncons- cious differences in collecting and recording insurance claims, depending on the experimental treatment that had been employed. * Tasmania .. One cause for concern here is that "if the wind was such that the silver iodide might pass over one of the control areas no seeding was carried out but the experiment was not suspended." (Since the rainfall for such periods was still recorded as "seeded", the only unfavorable consequence of such decisions would be a decrease in sensitivity. The action here was thus conservsltive and well-taken. Our noting it as a subjective decision implies NO corresponding ground for criticism of the experiment or its results.) 9. Operational Changes and Difficulties Experimental efforts to modify the production of rain or hail entail a variety of charac- teristic difficulties. "Suitable conditions" for influencing rain or hail in a given environment occur infrequently -- in some cases, about 25 days a year. The influence of experimental inter- vention is unlikely to change the effects of nature by more than a modest amount. The, varia- bility of precipitation from day to day is great, with or without experimental intervl~ntion, requiring a large number of observations to detect possible experimental effects. Consequently,