Probabilism in Geological Data Analysis

The way in which probability can enter the interpretation of geological data is outlined. Probabilistic models are introduced in data synopses to allow statistical inferences about samples and population parameters. Inferences of this type are a special application of mathematical theory and have little to do with testing scientific generalizations. The logic of statistical inference, as used in operations research, is contrasted with that of geological inference. Geology is a cumulative science; its methodology is not that of experimentation with formal replication; inferences are developed from observations and tested and refined against new data, often by different investigators at different times and in different places. Because of the way in which inferences are drawn, sampling in geology is purposive. Random sampling, necessary for inferences based on sampling theory, is unattainable in most geological contexts. Different classes of geological measurements require careful consideration as to their appropriate form of synopsis, and, particularly, as to whether a parametric probability model is applicable.