Principle 8: Use small multiples (easy on your viewers).
Discussion: Bar charts, for example, act as information architecture. With small multiples on these charts we can demonstrate mastery of detail. As you know, I always ask you to label the axes, both horizontal and vertical. You may need a legend. By labeling every appropriate multiple, you do not run the risk of someone accusing you of evidence selection. You need to understand the credibility of the information.
You have to make a selection first about whether you will use a pie, bar, line, surface, or flowchart. Once that decision has been made and you have determined the chart's purpose, you are ready to choose a question from your survey or interview guide. You should not choose too much data for the chart presentation.
A student was recently faced with an interview guide. What questions were most important to present what the apartment complex should do to increase tenancy? The student had to examine his problem question as well as his purpose. Eventually, the student decided that the use of media recommended by the tenants and the manager might be the most important information to present. The student could use a horizontal bar chart (bar chart) to present the data with the vertical axis showing the different media and the horizontal axis showing the number of respondents for each kind of media. Here's how the student arrived at all these decisions:
- Determine the chart's purpose.
- Determine the categories (by you) the interviewees talked about.
- Arrive at a set list of categories.
- Determine how to create the value and category axis for a program, such as Excel.
- Plot the appropriate numbers with an appropriate chart from Excel, for example.
- Conceive title for chart. Make sure you write a talking caption.
- Pick out the most important bar or slice for writing your title.
- Conceive a conclusion for the title.
In the latest Style Guide for Business and Technical Communication from Franklin-Covey, a well-known scheduling book and time management company providing seminar services, the issue of which chart for which occasion occurred. You may also recognize Covey as the author of several books on Effective Habits for individuals and families. The book is devoted to all kind of business communication situations, but the chart section of the reference was particularly strong. We can classify certain charts as to their uses. First, the pie charts are considered:
Advantages and Disadvantages
Comparison and Contrast
Decisions and Alternatives
Costs
Trends
This list suggests you should consider a pie chart when you are expressing the previous items. For example, 100 percent of some costs might necessitate a pie chart. You could also use a line chart, depending on your preference and what you wanted to show.
You could compare or contrast certain slices of the pie and percentages. Think of an example, such as percentage cost to run each department of a retail store. You comparison and contrast works well in that case.
For line chart selection you have a number of alternatives. The uses include:
Chronology
Costs
Trends
Decisions and Alternatives
Comparison and Contrasts
Chronology particularly interested me in this list. You see all the time the chronology of certain stock transactions in the newspaper. You see the history of the stock as well as the ups and downs of the market prices. Can you think of a line chart that would show decisions and alternatives?
One semester a student did a fascinating report on store tastings for a particular chain store. I did not know anything about store tastings. Apparently, store tastings occur when the employees are asked to sample different beverages or other merchandise before the product is made available to the public. According to the student, a store tasting "provides a time set aside for employees to either sample new products entering the store, or items that have not been sampled before to help increase familiarity about the products." Therefore, the student was interested in presenting data about whether employees engaged in a free for all by taking a free lunch instead of tasting the products. Out of that desire to present grew a pie chart.
As we remember a pie chart, you need to have 100 percent of something. You usually deal with percentages in a pie chart. You try to avoid more than six segments or slices in the pie for readability. The student presented data about attitudes toward the store tastings. The student calculated the percentages of responses for three major categories: (1)Store Tastings Are a Free for All; (2)Store Tastings Are Sometimes a Free for All; and (3)Store Tastings Are Not a Free for All. The student discovered that the largest percentages occurred in the first two categories. Even though the student only had 13 respondents or employees, the percentages made the point quickly. That is what you must do with your chart selected. Make your point quickly.
In the newly revised analytical report memo you are asked to provide an explanation tie-in for the table and figure. You are technically writing part of the Considerations section of the memo. You are doing more than saying the table and chart contain data. Your explanation in paragraph memo form with captions means you tell the significance of the numbers. You talk about percentages and averages. Your work will probably encompass close to a page of single-spaced keyboarding. You don't have to report every single number. However, you do not let the table or chart speak for themselves.
I suspect it would be helpful to see some actual paragraphs from a report and how the student handled part of the Considerations. In the report you are seeing the student surveying elementary school youngsters about their languages used and homework help obtained. You see one of the paragraphs called the explanation tie-in:
Attachment B, Table 1, describes people helping the sample students with their homework. The results show that 12 students (46 percent), the largest number, do not obtain assistance with their homework by any member of their home. They do homework by themselves. Five students (15 percent) are assisted by a third relative (aunt, uncle, or cousin). Three students (12 percent) obtain help from their siblings (brothers, sisters, or both).
In the second illustration we see explanation tie-in for charts, especially a pie chart. The student is trying to find out how to improve training in an investment banking firm. We will call the firm HEI. In the pie chart the student report writer had given the statistics of Strongly familiar (one slice), 23 percent, and Somewhat familiar (another slice or segment), 31 percent. From this pie chart the student wrote the following as part of the Considerations:
Figure 1 (Attachment B) presents a graphic display of the answers given to the HEI operations question. Out of the 13 people surveyed, only seven respondents (54 percent) were able to state they had some familiarity with HEI. The survey further revealed that some employees are not even aware that other key agencies in the settlement process exist.
Think about: Did you spot that 54 percent was arrived at by adding 23 and 31 percents respectively? It is always wise to combine numbers where appropriate in pie and horizontal bar charts. If we looked at the rest of the pie chart, we would see Just get-by, 23 percent, and Don't know, 23 percent. Did you notice the writer did not have to repeat the question to present the data? The report writer simply stated the "HEI operations question" and let the reader refer to the Attachments for more information about the exact wording of the question.
Exercise: Draw the chart you have just read about.
In the analytical report we place, for the most part, the tables and charts in the Attachments. Each table has its own table number and table title written this way:
TABLE 1
STORE TASTING PREPARATION SURVEY FOR EMPLOYEES
For charts you should use the Figure number and the Figure title. The figure title must be a talking caption. The figure number and figure title must be centered over the visual. Let's take an example:
FIGURE 1
CUSTOMERS WANT CIGARETTE VENDING MACHINE REMOVED
What you do in a figure is find some piece of data that stands out. That might be a pie chart, a line chart, or a bar chart. You make that piece of data your title. You avoid saying some of the following phrases:
SURVEY IDENTIFIES PROBLEMS
DATA SHOWS AN INCREASE
CHART SHOWS AN INVESTMENT OPPORTUNITY
These previous, vague titles do not let the reader know what is going on. You need to have your caption "talk" to the reader.
Tables require columnar presentation of data. Often, selecting specific questions or a whole range of interview of questionnaire questions will give you the data for a table. You may even build a quantitative data table from the interviews. You simply conceive the categories and place in the data in some workable fashion with the columns and rows. For example, columns for advantages and disadvantages of some particular plan may be a helpful way to present the qualitative data, categories you conceived from open-ended questions..
Qualitative Data Present Choices
Perhaps an actual example of a student facing a dilemma with presenting data will further explain the use of qualitative data. The student's report originally dealt with cash flow and advertising's impact. He interviewed two major department heads, including the customer service representative who gave him superb data about suggested solutions to the cash flow problem. The student did not have quantitative data; he had two interviews. Then, a light bulb dawned in the student's mind. Why couldn't he set up vertical columns headed "Department," "Problem," and "Suggested Solution"?
The student made choices. He knew management would appreciate a capsule form of the data shown in a qualitative data table. The customer representative provided particularly good data for presenting in the table. The student could place "n=2" after the title in the table.
In quantitative tables, such as numbers and percentages, you may want to provide each question on the left side of the table. You may shorten or paraphrase the question to accommodate the space. Then, your columns can be numbers, percentages, and total. Never forget to include a total column.
Don't forget that every table and figure should have "n=25 or number of respondents" placed in parentheses immediately after the table title. You need to clarify how many people responded. The reader should not be left to guess how many percentages or have to calculate the appropriate statistics.
In class we looked at five different problems and the charts that would satisfy each situation. You are asked to first say whether the chart should be component, item, time series, frequency distribution, or correlation. Use these exact words in explaining the chart. Then, rough out the chart by labeling Y and X axis and a title for each chart. Remember to use different kinds of plot lines for correlation (if appropriate) than for a horizontal bar chart. The exercises now follow:
- Sales are forecast to increase over the next 10 years.
- The largest number of employees earns between $30,000 and $35,000.
- In September, the turnover rates for the six divisions were about the same.
- The sales manager spends only 15 percent of the time in the field.
- Size of merit increases is not related to tenure.
- Region C ranks last in productivity.
Certain students over the semester have asked about the tables and charts assignment with the Tufte criteria. I have chosen selected paragraphs from previous students' memos to talk about clarification:
Quantitative Table Needs Investigation
The table shows heads for the rows and columns. The figures given are in percentages and in dollars in the first item in each column. Percentages are given for the first column and dollar signs for the next two columns. The title clearly states the main point. The graphic shows chronological order from 4.25% to 8.25%. It compares different interest rates, monthly payments and total costs. A source is shown at the bottom of the graphic. The research was performed by Times Research.
The table shows exact figures, and serves its purpose. However, the table needs a line between interest rate and monthly payment.
Think about: You are doing content thinking, not analysis thinking. We don't care what the table shows. We are interested in Tufte's principles.
What evidence is shown through footnotes and text that suggest you can trust the data?
In this case I looked at three variables, interest rate, monthly payment, and total cost. How clearly were those variables presented? Can you find the information you need to discover? How?
Principle 4 talks about integration of number, image, and word. As you look at the table, how effectively were those three elements integrated into the visual? How do you know? Give some specifics.
Think about: What kind of data comparisons are shown? Are the figures arranged in such a way that a clear presentation is given?
Think about: How could better content be achieved in the table or the graph?
Principle 8 talks about ease of reading. How clear were the numbers or the figures in understanding the visual? Did you find the table or the chart appealing to the eye? How do you know? Give some specific examples.
Charles Osgood, now narrator of Sunday Morning on CBS, featured a special Osgood File about how we are drowning in numbers. We have license permit numbers and PIN numbers. Osgood asks us if we are tired of numbers. We have social security numbers and tax numbers. Numbers define and locate us. You have an access code. You have area codes and telephone numbers to remember. You have so many numbers to remember.
Let's say you are looking at the evening news one day on the TV--even though college students don't have time for TV. You note the newly hired anchor for ABC News, Charles Gibson. He is explaining why many residents and evacuees from New Orleans have not returned from various cities around the United States. How does one show that graphically? Here's what Gibson did. He visually took the huge telephone directory and yellow pages from New Orleans before Hurricane Katrina. He placed those books on a table before the viewer. Then, he contrasted the telephone directory now (half the size and white pages packed with yellow)to prove his point about whether evacuees had returned. That dramatic and simple illustration in front of the camera brought home the problem. Sure, Gibson used props, but he visually described the dilemma. Simplicity was the key, and he made his point with graphics.
A number of students have asked for an example of the memo evaluating tables and charts. As you look at this example, please realize it is not perfect. Further, I will add certain Notes after certain paragraphs to clarify what I think the reader is doing. Let's try this example:
To: Dr. G. Jay Christensen
From: Gustave Gomez
buddyperson@aol.com
Date: Current
Subject: EVALUATION OF TABLES AND CHARTS
Visuals Receive Attention
This memo is written to critique and analyze two tables and two charts. Table
1 in Attachment A presents a qualitative explanation of seven various types of
automotive fuel cells. Table 2 in Attachment B contains a breakdown of the top
ten box office earners. All elements of the daily weather of North America are
illustrated in Attachment C. Attachment D is a bar chart demonstrating the
costs associated with ownership of a Mercedes-Benz E65 AMG. Each of these
illustrations attempts to conquer the difficult task of clearly presenting
an explanation through a chart or table.
Illustrations Show Comparison
Qualitative Table 1 (Attachment A)presents various data being compared. This
table sufficiently compares data by comparing and explaining five facets of
seven various options. Advantages are provided alongside disadvantages in a
non-biased manner. The table clearly and understandably compares the data found
within the research. The creator of the table successfully transfers
understanding of the topic onto the reader. Table 2 (Attachment B)also provides
data comparisons on an array of important information. The data shown in the
table cover every bit of information a reader would expect to see in such a
table. The illustration fairly and adequately compares various objective details
of the films. Figure 1 (Attachment C) also properly shows ample data
comparison for the reader. The chart contains multiple legends and explanations
necessary for this type of chart. Without the graphic assistance the average
reader would not be able to comprehend the chart and, therefore, miss the
purpose. Figure 2 (Attachment D)includes data comparison as well as the other
three but with less data to compare. Illustrations A, B, and C completed a higher
task of data comparison because of the excessive amount of information given
in each.
Note: The writer could have helped the reader by putting information about the
intended purpose of each table and chart. What did the writer or the preparer
intend?
Figures Boast Causality
All four figures allow a reader to easily learn about the presented evidence.
Table 1 (Attachment A)provides clear column headings to complement the data
given within the table. This method further improves the chance a reader will
have of understanding the point of the table. This explanation allows the reader
to learn about the evidence uncovered during the research and testing of the
subjects. The setup of Table 2 (Attachment B)creates a similar effect.
Attachment C allows the reader to understand the daily weather. This allowance
is made possible by the previously mentioned legend. The legend enhances the
reader's understanding of the illustration. The reading of Figure 1
(Attachment C) is quite simple. This simplicity of the chart allows the
reader an opportunity to easily learn the presented evidence.
Dr. G. Jay Christensen
Page 2
Current Date
Multiple Variables Require Appearance
A seemingly infinite number of variables exist in Table 1 (Attachment A). The
headings for each column allow a wide array of data responses as entered data.
The use of multiple variables in this chart guarantees a complete explanation
of the subject. Attachment B also addresses a number of important variables.
Each of the included variables builds on the next. This system of variables
helps the reader understand the flow and significance of the variables on one
another. Many variables are also included in Figure 1 (Attachment C). The data
in this type of chart vary from one area to another. Each presented variable
is necessary to construct a complete chart. Each of the variables is
accompanied by an explanation to inform the reader of the reason for the
variable's presence. The simplicity of Attachment D is obtained as a direct
result of the low number of variables. An excessive number of variables in
this chart would only cause confusion and defeat the purpose. Only a few
variables are necessary to understand the few items that need to be examined
in this bar chart.
Note: As I read over the previous paragraph, I am left with questions. Why is
there so much that the visual preparer had to include as variables? Did you
notice the sentences became a little long with so many prepositions?
Understanding Demands Integration
The flow of the variables in Attachment A results in beautiful integration.
The data from one area relate directly to the data in the next. The structure of
Table 2 (Attachment B) immediately forms an understanding of the topic in the
reader's head. Each column of data relates to the next column to the right.
This pattern stays consistent within the entire table. Attachment C is
flawlessly drawn. The reader can credit understanding of this chart to the
creator's integration of words, images, and explanations. Various colors are
used to assist in conveying the information to the reader. Figure 2
Attachment D)contains text to explain the expressed data entries. Integration of
of the data is not too difficult in this case because of the minimal variables.
Note: The writer has made a sincere attempt to use words that convey meaning. You
can tell the person has studied the visuals. The Tufte criterion is applied,
and the use of color is also mentioned. That brings up a point. Make sure your
table or chart appears in its original color. Color is important in trying to
convey meaning. Also, take the time to explain the charts rather than let them
speak for themselves.
Documentation Provides Support
Table 1 (Attachment A)is supported with extensive text and research. A large
amount of research was performed over the span of a few decades before the data
were entered into this table. The data in Attachment B is taken from another
source. This source is cited at the bottom of the table. To form a table such
as this one, the original source must already have a system in place to gather
the extensive amount of data. Data were retrieved from thousands of
significant sources to present the data in this table. The validity of the
data given in Figure 1 (Attachment C)depends on the technology used to gather
it. The more advanced the weather tracking device, the more accurate the data.
The exact sources for the construction of the data in Attachment D are given.
Mathematical calculations are used to obtain the data in this type of chart.
Some variables within the calculations are estimated or assumed. As a result,
the information provided is only as accurate as the assumptions used to obtain
it.
Dr. G. Jay Christensen
Page 3
Current Date
Chartjunk Decreases Value
One column appearing in Table 1 (Attachment A) provides a unique, but trivial
facts. These facts disrupt the flow of the chart more than they contribute.
The vague relevance of the added facts may cause a reader to lose focus of
the table's main points. The breakdown of data in Table 2 (Attachment B)
ensures the reader will only read quality, relevant information. Unnecessary
information are not given to obstruct the view of the table's reader. Impor-
tant and relevant data are all that are provided in this table. Though only
a small portion of the data may be relevant to reader of Figure 1 (Attachment
C), it must all be provided to cover the majority of the possible readers.
Again, the quality of the data depends on the quality of the research tools
used to gather the data. Figure 2 (Attachment D) uses assumptions to figure
the data entered. The integrity and quality of the data depend on the accuracy
of these assumptions. The creator of this chart could easily sway the numbers
one way or the other by manipulating the original assumptions.
Note: As I originally commented on the paper, "Much better analysis." The
writer is starting to go on a tangent, however, with the discussions about
assumptions and accuracy. One has to be careful in this kind of the table "not to
ramble." Say what you have to say, and stop.
Information Stacking Breeds Confusion
Tables 1 and 2 (Attachments A and B)insert data into distinct boxes, all of
which are the same size. The uniformity of the tables increases the ease of
reading them. Figure 1 (Attachment C)mixes much of the data together. This
mixing of the data creates difficulty in the reader's understanding. To
interpret the data given within this chart, a reader would need much more time to
separate the data and anlyze what exactly is going on. The data provided in
Figure 2(Attachment D) is shown adjacent in space. This setup permits easier
comprehension by the reader.
Number Size Plays Important Part
Qualitative Table 1 (Attachment A)does not provide any information in numerical
form. Table 2 (Attachment B), on the other side, was forced to deal with
large numbers. To break the numbers down for ease of comparison, the table
expresses large aounts in smaller figures. This breakdown is communicated to
the reader at the top of the column. "Millions" is stated in parentheses.
This labeling is a common method used to solve the problem of large multiples. The
numbers contained within Figure 1 (Attachment C)are broken into a scale. The
scale had to be wide enough to cover the entire temperature spectrum, and in
intervals small enough to render the chart useful. The numerical multiples used
in Figure 2 (Attachment D)work sufficiently well for the purpose of illus-
tration. Precise detail is not necessary for this graph to be successful in its
intentions.
Dr. G. Jay Christensen
Page 4
Current Date
Tables and Charts Need Examining
Many details of a chart or table must be examined before one can accept the
contents as valid. Specific layouts may intentionally or non-intentionally
confuse a reader. The maker of a table may use certain setups to purposefully
lead the reader to interpret the chart in a specific way. Written the wrong
way, an illustration may sway a reader in the wrong direction for the creator.
All of these details must be considered before one should believe the contents
of any table or chart.
Attachments: A. Table 1, ". . . ."
B. Table 2, ". . . . "
C. Figure 1, ". . . ."
D. Figure 2, ". . . . ."
Own Criteria Questions for Tables
Own Criteria Questions for Charts
Note: Did you spot the writer needed a way to end the memo? The writer chose to
comment on the importance of visuals and their place of interpretation. The
writer drew from class lectures and readings how important tables and charts
have become. You must have some kind of summary paragraph.
Is the day coming when we will talk in our language only in digits? Will the keypad replace the keyboard? What will the conversation of the future sound like? The computer has achieved the triumph of numbers. These questions are worth pondering.
I am indebted to Gene Zelazny for the previous problems as explained in his Say It with Charts. If you ever want to look at this excellent reference, you will probably find the book on library reserve.
Zelazny is famous for how he classifies the kind of charts to use:
- Component: Try a pie chart. How much of the pie or percentage are you considering? Always look at the 100 percent issue.
- Item: Now, you are looking at a bar chart (a horizontal bar chart). You want to make some comparisons. Maybe you are looking at students or employees; those are considered items.
- Time Series: Here you are considering forecasts or trends. You want to know the changing dynamics of some kind of data. Therefore, a columnar bar chart or a line chart seem most appropriate.
- Frequency Distribution: You look for the normal curve. Actually, you can superimpose the normal curve on a tightly designed columnar bar chart.
- Relationships: You want to consider a dot or correlation chart. You want to know the relationship between one variable and another. You are interested in the line of best fit. Correlation charts are more rare in ordinary business charting, but they can be useful.
