FLOOD09697 - Laserfiche WebLink

Home Browse Search

indirecl method cannol be defended as reliable for model application but was selected as the mosl prac- ncal melhod for supplying Ihe needed data for cali- brating Ihe model. Conlent Analysis of Agency Information Techniques of contenl analysis were used 10 de- velop a method for scoring and tabulaling Ihe dala obtalned in agency and group interviews. Content analysis has been used 10 render unlike data into like forms for comparison and to oblain quantilative mea- sures from qualitative dala for mathematical analyses. Variations permit one to compare several sources of information or one or more sources at different per- iods of time (see also Holsti, 1968). One lechnique employed here was a simple count of each lime an interviewee mentioned an agency, public group, or individual associated with an agency or group. It was thought thai such a count would reveal which agen- cies are considered most important. The type of relalionship one agency had with others that it mentioned was also explored. Each slalemenl that referenced a Iype of relationship was extracted and the proper nouns removed. Later "judges" were asked to evaluate these statements and rank the legal and hierarchical aspects of the relation- ships. For example, interview notes might contain a statement such as. "We must submit our plans to the County Commission for approval." The statement would be put in the following form: "X must submit its plans to Y for approval." This and all other state- ments which mighl have some bearing on the legal and hierarchical relationship between two agencies were then grouped and considered by the judges. An example of such a set of statements might be as fol- lows: "X must submit its plans to Y for approval." !ly determines which problems X will study." !ly has X hold monthly public meetings." Wilhin such a set of statements, X is always the same agency and Y is always Ihe same agency. The judges were individually asked to review the set of statements and Ihen classify the relationship of X to Y as an hierarchical or horizontal power relationship, etc. Olher stalements were similarly extracted and Ihe judges were asked 10 evaluate the quality a~d in- tensity of the relationship, the involvement of each group in flood control, Ihe intensity of the involve- ment, the time orientation of the group (Le., if its plans were of a short term or a long term nature), and the group reaction to various flood conlrol proposals. The choice of information to be probed by con- tent analysis was limited by the time available for inter- views. It took some time to develop a feel for Ihe kinds of questions to ask officials. One interview did not always conlain answers to queslions probed in other interviews, and it was not until interviews were well along Ihal the list of questions was able to be ~ta.ndardized. Some inlerviews occurred in public or ill mteragency meetings where the researcher was not at full liberty to ask queslions. Anolher deficiency in this approach is thai inlerviews are recorded by nole taking, and quotations may nol be complete or may be paraphrased. Ideally, every inlerview should be recorded verbatim. Conl.ent analysis, however, is a viable method for quantifying qualitative dala. Data Analysis for Sociological Variables Stalislical Tesls Used and Identification of Variables After the public interviewing was completed, a number of steps were necessary to prepare the data to develop relationships for Ihe sociological compon- ent of the model. Each response to each ilem was coded and punched on cards for compuler process- ing. Responses to each ilem were first tabulaled and analyzed for distribution and number of no-answers. A second sel of decks was then made for cross- labulation of pairs of items and the results were ana- lyzed for significance by several non-paramelric and correla~ion tests. Chi square, Cramer's V, Contingency CoeffiCient, and Gamma rank-order (Nie el aI., 1970: 275-277; also note James et aI., 1971: 57). The ob- jective was to identify significanl variables and estab- lish the relative imporlance of Ihose identified. The principal method used in quantifying social relationships for inclusion in the model was multiple regression analysis. Multiple regression equations were developed for the significant relationships for inclu- sion in the sociological component of the model. Cer- tain variables which had been assumed to be indepen- dent, however, appeared in several equations. As an example, the variable titled, "Knowledge of flood con- trol projects" was found to be correlated with whether persons favored or opposed particular projects. In order to increase understanding of the "knowledge variables, and those variables which might be corre- lated with it was run as a dependent variable with some of the other variables being used as independent, or In other words, as predictors of knowledge." Through lhis method. knowledge of interrelationships among .variables was increased. Another example is shown m the model presented in Chapter Y, where Concern About Flooding is the dependenl variable and Perceived likelihood of Flooding is an indepen- dent variable in Equation I (shown later in Table 5.9). Several studies demonstrate the versatilily of re- gression analysis in the social sciences. Techniques Slmilar 10 those used elsewhere were applied in this 12