Applications of Public Health Informatics in Public Health Surveillance -
Spatial and Temporal Analysis
Conducting surveys, analysis of data, information management and presentation are areas given considerable attention by the expert staff of ACC. This document will provide summary information regarding the application of techniques with examples of studies conducted by ACC staff.
Data Analysis
There are numerous methods for the analysis of primary and secondary data. The analytical process often is to generate simple numbers to describe distributions, either grouped or ungrouped. Most often statistical measurements form the basis of such analysis. Statistical analysis has two functions: it can add to our understanding of the data that make up the distribution, and it can substitute for (be used instead of) the distribution. We can divide major descriptive statistical methodology into two groups: measures of central tendency and measures of dispersion. In addition, often statistical measures are employed to present the trends and the significance of the inference variables.
However, as statistical analysis and other increasingly powerful and complex tools for analyzing data are worth noting, some data analysts concluded that the underlying structure of the data is often ignored in favor of sophisticated - but largely inconclusive -analytical methods. The most effective response to these arguments comes from the statistician John Tukey in his 1977 book Exploratory Data Analysis. In his book he states, "We must be careful that in the use of traditional statistical analytics, the extremes get dealt with too harshly; there is too much emphasis on the central sections of the distribution. We define values well away from the centre as 'outliers', and treat them as somehow abnormal, or as unduly influencing the measures of centrality and dispersion." This must be kept in mind when analyzing and presenting the findings.
Employment of the most sophisticated analytical tools with complete and accurate primary and secondary data is not the key to maximizing its value, having the appropriate presentation methodology is vital for a successful outcome.
The Presentation
In recent decades many researchers have realized that they have an obligation: to make the results of their work known to a much larger community, outside the somewhat restricted group of fellow researchers, including Congress, OMB, and federal, state and community program and project administrators. One example is to present the data that is aggregated, spatially or temporally, at a level that makes sense to its audience or stakeholders. This communication of science to government officials and the general public which may be aided by mass media, has become increasingly significant in recent years - not least due to the pressure on researchers that accompanies the competitive search for scarce research funds.
What many researchers have found, however, is that the techniques that they can successfully employ to distribute information to their peer groups are at best ineffective, and at worst totally useless, when used to communicate with non-specialist audiences. Unfortunately, most researchers still receive little or no training in this process.
The process of communication with an audience is no different when dealing with researchers than with any other group. In principle the process can be studied and analyzed, and strategies developed for maximizing the impact of the presentation on the audience.
Analyzing the audience
All communication studies emphasize the pivotal role of a clear understanding of the characteristics of the audience. These characteristics include factors such as:
- The extent to which the audience has similar degrees of knowledge and educational levels,
- Whether the audience is likely to want detailed material,
- How visually literate the audience is, which will determine the number, style and content of illustrations,
- How numerically literate the audience is, which will determine the use of equations and formulae, and
- Present the data that is aggregated, spatially or temporally, at a level that makes sense to its audience or stakeholders.
Balancing the components
The skill in developing any communication is in balancing the components: the support material, the language and jargon, and the overall structure.
Even with experience most researchers find this process at least as complex (and probably more mystifying) than the actual research. They are most comfortable when dealing with the 'traditional' formats such as journal articles and conference papers, and least comfortable when using new formats such as electronic publishing, or when communicating those who are outside the specific research community.
Effective presentations
When evaluating the effectiveness of any presentation (in whatever format) there are two major areas to consider. This can be described as either 'internal' or 'external'. Internal factors relate to the contents and structure of the presentation, and external factors relate to the impact of the presentation on the audience. The two are obviously linked, in that the audience will not usually be in a position to compensate for deficiencies in the internal factors. Similarly, the best-prepared material will be ineffective if presented to the wrong audience. Effective presentations must combine an appropriate choice of the correct amount of the right level of material, with an audience receptive to that material.
The presentation of survey and performance information
It would be wrong to imply that presenting survey and performance information requires a peculiar approach solely because it deals with numbers and equations, or because the general public knows relatively little about the data variables and their significance. On the other hand, many researchers are still uncomfortable with the more relaxed style that is expected in presentations to wider audience, a style that is becoming increasingly accepted in specialized research publications.
Presentation formats
The most skilled researcher involved in the field of survey and performance data presentation is aware of the many forms of presentation and the means for achieving such presentation, such as, for instance, temporal and spatial analysis. Such researcher knows that each presentation method has specific restrictions and possibilities that affect how they are used to convey certain types of information to certain audiences, written or oral or both. We are going to limit our discussion to the use of Text and Table Presentation considerations.
Text and tables
The use of text and before looking in some depth at graphical presentation techniques, one need to understand the relative merits of describing data (which are largely numerical) using words, numbers or pictures: knowing this one can select the right combination for each particular data analysis task. The power of properly constructed illustrations re-enforces the text when conveying research material, to both researchers and non-researchers alike. Such presentation methods include: tables, frequency distributions, histograms, pie charts, area charts, polar charts, 3D plots, triangular charts, flow charts, organization charts, and temporal and spatial illustrations. Each presentation method must consider the audience and in so doing give attention to type, size, color, and structure in:
- Defining the contribution to the presentation process that graphical displays of data can make,
- Describing the essential elements of graphical integrity in data graphics,
- Listing and distinguishing the major forms of data graphs, and
- Listing and distinguishing between the major forms of non-quantitative (structural) graphics.
The contribution of graphical data display
Given that one of the hallmarks of most scientific research is the collection of data, and given that researchers need to analyze and interpret those data, what methods are available for examining data graphically? Such methods, when used in conjunction with the numerical data summarization will form the first line of attack for interpreting data.
The rather tired cliché 'a picture is worth a thousand words' has less relevance in data presentation than in other areas, but it is true that effective graphs can markedly increase a reader's comprehension of complex data sets.
What constitutes 'effectiveness' in graphical data presentation remains a matter of some dispute. The advent of inexpensive, feature-rich graphing software on personal computers has made it easy to create eye-catching charts that often have no real value, and may actually mislead the viewer (perhaps deliberately). We see large numbers of graphics in the media that tend to emphasis form over content, usually because of an inadequate appreciation of numerical techniques on the part of the designer of the graphics. To create what the American graphic designer Edward Tufte called graphical excellence demands a blend of statistical rigor and graphical design skills that is unfortunately rare.
Nevertheless, it is possible, by knowing the basic types of data graphs and understanding their limitations, one can create visually-pleasing charts that are also founded in sound statistical principles.
Graphical integrity
The key to creating effective data graphics is the combination of good graphic design and appreciation of statistics that Edward Tufte (in his seminal books The Visual Display of Quantitative Information and Envisioning Information) defined as graphical integrity. He suggested that application of the following principles could lead to both graphical integrity and graphical excellence:
- It is essential to focus on the substance (contents) of the graph - not on the design, methodology or technology.
- It is essential to avoid distortion induced by the format of the graph.
- It is desirable to aim for what he termed high data density, whereby the graph is used to present - as no other method can - large amounts of data in a coherent manner.
- It is worthwhile constructing graphs that not only present data in a `static' from, but that encourage comparisons between variables, location, or time periods.
- It is desirable to allow the viewer to discover levels of detail within the graph.
- It is necessary to make the graph serve a single, clear purpose, such as data description, tabulation, exploration or decoration.
- It is important to integrate the graph with statistical and textual descriptions.
- Above all, it is vital that the graph should show the data, not the technical skills of its creator.
The following are examples of temporal and spatial analysis that were taken from studies conducted by Dr. Hsu of American Cybernetic Corporation.
|
