Most technical presentations provide insights and promote discussions, and are not forums for dissecting raw data. Graphs and tables are the best way to summarize large quantities of raw data. Because of its visual nature, graphical information becomes memorable. However, complicated visual data causes undesirable impressions (presenter cannot interpret own data, is a disorganized person, or is being inconsiderate by talking down to the audience).

It is not unusual for the same data to be used in a journal article, a slide show and a poster show. However, each presentation type has different requirements. Design the graph for the type of presentation it will be used for. Graphs prepared for publication (where reader can linger over the data) are not ideal for slides or posters (where viewer gets only a glimpse of the image).

Poor Visual

Good Visual

Getting started with graphs and tables.

• Simplify the data.

• Show only the essential information needed to get the point across.

• Don’t put up an entire complicated chart or graph.

• Streamline the visual so it can be read quickly.

• Use only essential and functional linework in the graphic to orient the viewer.

• If you do have to show the whole graph, provide a handout so the audience can follow along.

• Be consistent in style and terminology.

• Be consistent in label size, font, type style, line width, colors, symbols, and layout.

• For posters, graphics should be done on matte surfaced paper to reduce glare from overhead lights.

Place the legends on the bars, lines, wedges of a pie, or the illustration rather than having an actual legend. This avoids double scanning involved in finding a relevant part of the key and then referring back to the graphic. If a key is used, put it within the picture area of the graph. Avoid enclosing the key in a box.

Good Example

Poor Example

• Labels should run horizontal from left to right whenever possible, with the exception of the Y-axis label. A vertical Y-axis label saves space. Vertical text is harder to read so make it large and simple enough to read when rotated. Axis labels should be placed outside the graph. X-axis label is centered under the X axis. The Y-axis label is centered parallel to the Y axis. Y-axisl labels should read from bottom to top.

• Axis labels should be placed close to the scale numbers, but not so close that they protrude into the imaginary rectangle occupied by the scale numbers.

• If have more than one Y axis, the labels should all read in the same direction (from bottom to top). It is helpful to include the data point or line pattern in the Y-axis labels to indicate which data relate to which Y axis.

• Numbers along Y- axis should be horizontal and aligned on the decimal point whether or not the decimal point is actually present.

• Label size should be large enough to be legible but not so large or heavy that they compete with the chart content for attention. Size of numbers should be smaller than axis label size. Title should be larger than axis labels. Explanatory labels like a legend should be smaller than axis labels.

• Bolding or increasing the text size of the headings can help to differentiate them from the body of the table. Use larger font sizes in preference to boldface text.

• For posters, text needs to be large enough so that the audience can read it comfortably from 2 meters. Labels should be set in at least 18-point text, upper and lowercase. Use generous leading on multiline blocks of text.

• For numeric labels along axes, use no more than 5-7 words on an average chart axis.

• For multichart formats or composite figures that share common X or Y axis, eliminate any redundant axis labeling.

• Spell out as many of the labels as possible but be brief. Too many obscure abbreviations and mysterious codes are distracting. Standard abbreviations can be used to keep axis labels short (% for percentages).

• Axis labels should include both the name of the variable and the unit of measurement. Conventionally the name of the variable is given first and then the unit of measurement is given immediately next to or below it in parentheses.

• Use units that are complete and easily understood.

• Use upper and lower case lettering.

• Use sans serif typefaces, such as Helvetica, for all labeling and text.

• Caps on error bars should match the diameter of the data point markers or symbols.

• Error bars should be thinner than the line used for data lines.

• Lines used for error bars should be thicker than lines used for the grid or frame but thinner than those used for data lines.

• Used to identify the type of scale, not to identify each data point. They should be long enough to be visible but not obtrusive.

• Tick marks on a linear scale should be the same length.

• Don’t label every tick mark or the axis will look too cluttered. Number at conventional intervals, for example 0, 2, 4, 6, 8....

• Tick marks at the beginning of each logarithmic cycle are longer than the others. Only tick marks at the beginning of cycles in logarithmic cycles are numbered.

• Axis should end at a tick mark.

• Tick marks can point in or out, but they should not cross the axis. Tick marks that point in direct the eye toward the data. Tick marks that point out keep the face of the graph clear and are particularly useful when data fall on the axis.

• Avoid special effects if they do not enhance the point to be made.

• Use of 3-D bars may make reading the data a little more difficult because it is not evident which part of the 3-D bar measures actual data.

• Use 3-D graphs if you have 3 axes of information.

• Use a build series on a 2D version of the chart to add some distinctive effect.

• Consider breaking the graph or figure into components (i.e., add segments, build, or use more slides to show the graph).

• You can show the entire graph and then focus in on a certain part that really displays the information you want to get across.