No abstract

Concentration distributions of air contaminants and meteorological variables in New Jersey and New York for workdays (Mondays through Fridays, omitting holidays) and Sundays are compared by means of quantile-quantile plots. The ozone distributions are slightly higher on Sundays, and the primary pollutant distributions are lower. These results raise serious questions about the validity of current concepts underlying ozone reduction in urban atmospheres.

A contingency table specifies the joint distribution of a number of discrete variables. The numbers in a contingency table are represented by rectangles of areas proportional to the numbers, with shape and position chosen to expose deviations from independence models. The collection of rectangles for the contingency table is called a mosaic. Mosaics of various types are given for contingency tables of two and more variables.

Harmonically related peaks in the spectrum of a stationary stochastic process may indicate the presence or wave components that are not sine-shaped, i.e., whose Fourier expansions contain phase-locked higher order terms. But the spectrum itself suppresses phase relations, and more refined methods are needed to decide such questions. Moreover, phase relations might also exist outside of the peaks. We discuss proposals for testing the presence of phase relations and for extracting them quantitatively by means of numerical bispectrum analysis, and we derive their statistical properties and compare their relative merits. Applications of these methods to EEG signals will be presented.

Exploratory data analysis (EDA) techniques are particularly relevant to the geophysical sciences, since geophysical systems typically possess multiple correlations of a variety of parameters on widely differing time scales, often in conjunction with temporal patterns of low intensity but high importance. In this article a variety of EDA techniques (robust statistics, scatterplots, histograms, stem-and-leaf diagrams, box plots, trend analysis programs, and quantile-quantile plots) are applied to geophysical data to demonstrate the successful elucidation of information difficult or impossible to extract from the data by more formal and traditional statistical approaches. I. INTRODUCTIONObservational data in many of the natural sciences are acquired under rather carefully controlled conditions, often with variation in only a single parameter between consecutive observations. Under such conditions, routine statistical tools are often sufficient for analysis. In the geophysical sciences, however, controlled or replicate field experiments are the exception rather than the rule. The data of interest contain diurnal, seasonal, annual, and multiyear patterns, in addition to the more dramatic effects that take place on shorter time scales. Geophysical data systems also abound in weak regular patterns over a very broad rfinge of time scales, any of which may be of interest. These properties suggest that innovative treatment of the data, termed exploratory data analysis (EDA), may bring substantial rewards. In this article we hope to demonstrate that such a' statistical approach is well within the feasibility of modern computational systems and that considerable useful information can result from taking such an approach to the analysis of geophysical data. Exploratory analysis is designed to find out 'what the data are telling us.' Its basic intent is to search for interesting rela-tiOnships and structures in a body of data and to exhibit the results in such a way as to make them recognizable. This process involves 'summarization,' perhaps in the form of a few simple statistics (e.g., mean and variance of a set of data) or perhaps in the form of a simple plot (such as a scatterplot). It also involves 'exposure,' that is, the presentation of the data so as to allow one to see both anticipated and unexpected characteristics of the data. In general, finding the unexpected turns out to be much more rewarding than merely confirming the suspected.Although EDA is by no means a new subject, it is only in the recent past that it has become an important part of statistical analysis. A major factor in this development has been the rapidly increasing availability of computer hardware and software, together with the challenge of larger bodies bf data in many fields and the accelerating emphasis on quantification in a growing variety of disciplines. Despite its utility, however, EDA has yet to be adequately incorporated into formal statistical theory. Perhaps the best succinct introduction to the subject that can currently be given is to ...