Laserfiche WebLink
<br />18 <br /> <br />second-hand sources -- either neighbors or officials involved in the <br />re 1 i ef effort n who knew of the acti ons people took during the flood. <br />Two modes of analysis were employed in this study: contingency <br />tables and discriminant analysis. <br />Contingency tables were used to examine whether or not various <br />actions taken during the flood were related to chances of survival. <br />Because the actions cannot be ranked in any meaningful way the contin- <br />gency analysis was as sophisticated as the data allowed. This mode of <br />analysis tells whether certain actions in this particular study more <br />frequently lead to survival than others. An example of a result of the <br />contingency analysis is that those who took an action were more likely <br />to live than die. The contingency tables are given as Tables 1-9 in <br />Appendix F. The results of the contingency analysis are shown as <br />generalizations inthe first half of Table 3. <br />For the variables which could be ranked, discriminant analysis, <br />a more sophisticated technique, was used. This technique is useful in <br />classifying data into categories. It tells us which variables were most <br />significant in predicting which class a particular case would belong. For <br />example, in determining which factors most influenced whether a group <br />took an action or did not take an action it was found that the number of <br />persons in the group was most significant. It must be stressed that <br />even though in this instance the variable "number in group" was most <br />influential in predicting whether a group would or would not take an <br />action, this cannot be generalized to all floods because the sample was <br />not randomly selected. The second half of Table 3 gives generalizations <br />from this study of the Big Thompson flood drawn from the actual data <br />analysis found in Tables 10-13 in Appendix F. <br />