In the most general sense, any designer of on-screen displays should understand and design around some basic concepts. Any strategy for visualizing should begin with the following guidelines.
Know your audience. Their level of expertise will determine your design strategy. If this is a display for the general public, for example, a high degree of selection and generalization may be required, whereas expert viewers may desire a large amount of detail and depth.
Get your point across. Does the dominant visual feature on your map relate to the purpose of the map? Certain elements of a graphic object, like its hue, location, size, and contrast, tend to be more easily and more intuitively interpreted by human vision. Make sure that the dominant object on the display supports or at least represents the primary theme of interest of the graphic.
Avoid unnecessary bells and whistles. Including a visually striking digital terrain model on a map of voting results may make the display look better, but it in fact may also serve to distract the viewer to the extent that the main point of the display is lost. More information on a map, thus, is not always better.
Strive for visual balance. Use the space of your display to its fullest by avoiding large empty areas. However, donít crowd the image to the point of being difficult to interpret or even look at. Use a blend of text and graphic elements, and center your main image and its title to emphasize its importance.
from Craenen, et al., in GIS: Our Common Language, ESRI Map Book, Volume 12, p. 42.
The map above achieves these objectives to some extent. The map is visually balanced, with informative but not distracting or confusing text, a useable legend, and straightforward graphics. The message of the map, however, is unclear due to the color scheme used. The eye is drawn to the darkest (red) values, but we notice that this color represents the lowest of the attribute values. It is likely that low values of this theme are, in fact, the most important (showing regions of a low threshold), however, if this is in fact the case, the highest values (>2000) should not be represented by an even darker hue (black). The data is ordered, yet the color scheme used does not imply order.