|
Statistical procedures for describing the major features of a data set, from which hypotheses for further testing may be generated. Whereas many statistical analyses follow the rules of scientific inference in the evaluation of hypotheses (see confirmatory data analysis), exploratory data analysis makes few initial assumptions about the expected findings.
Exploratory analyses are of particular value in geographical work for two reasons. First, much geographical investigation has a relatively weak theoretical base and its empirical expectations are consequently fairly imprecise in their goals: exploratory data analysis is sympathetic to such situations, because it lacks the constraints of formal hypothesis-testing. Secondly, few of the data sets employed by geographers are specifically collected for their purposes in properly controlled experimental conditions, so that the investigator has an incomplete appreciation of their structure. Exploratory data analysis, with its emphasis on graphical display, allows researchers to penetrate data sets, appreciate their peculiarities, and draw conclusions which are constrained neither by prior expectations nor by the limitations of inferential techniques. (RJJ)
Suggested Reading Sibley, D. 1990: Spatial applications of exploratory data analysis. Concepts and techniques in modern geography 49. Norwich: Environmental Publications. |
|