Reading 4.

Creating charts
In this article we dive into different types of charts, situations in which they are applicable,
and their strengths and weaknesses. Use the table below to jump right into specific areas:

Section Description Quick Tip

What Chart Type Should I Use? A look at different chart types Don’t use just one kind of chart,
according to different kinds of each chart type has specific
data. applications.

Column and Bar Charts An overview of different things A bar chart flips your x,y axis and
that can happen with Column and presents the data with your
Bar charts. categories listed vertically.

Line and Scatter Charts A few things to keep in mind It’s generally easier to read line
when building up line and scatter charts when you add markers.
charts.

Pies and Donuts Strengths and weaknesses of Pie Only use pie or donut chart when
and Donut charts. aspects add up to 100%

Stacked Charts An overview of the possibilities Avoid stacked line charts, they
and weaknesses of stacked are difficult to read and often
charts. misrepresent data.

4.1.1 Chart types

From the wide assortment of chart types available in the ribbon, the most useful and the
most common are column and bar charts, line and area charts, pie and donut charts, and XY
scatter charts. These are shown below in edited screen caps of the ribbon controls. I’ve
omitted the 3D versions of these charts, since there is no compelling reason to use 3D
charts; they might look fancier, but 3D charts can distort the data and mislead viewers. I’ve
also left out the less common chart types, but you can explore those after you’ve gained
some expertise.

READING 4.1 | CREATING CHARTS | PAGE 1 OF 11

Chart Axes
Much of the difference between chart types can be related to differences in their axes. Here
are examples of each made using a very simple data set.

Line and area charts are related. A line chart plots its data points, often with markers, and
connects the points with a line. The area chart does not use markers, instead it fills the area
below the line.

Column and bar charts are related: both use the bars that indicate a value with their lengths.
The column chart uses vertical bars or columns, and the bar chart uses horizontal bars.

A pie chart has no axes, but each of the other charts has two axes. The X values in the first
column of data are plotted along the X axis, also called the category axis. The Y values in the
second column of data as plotted along the Y axis, also called the value axis; these are the
blue numerical values. Note that the bar chart switches the category X axis and the value Y
axis.

READING 4.1 | CREATING CHARTS | PAGE 2 OF 11

4.1.2 What Chart Type Should I Use?
Each chart type has strengths and weaknesses. Each chart style is suited for the display of
certain types of data.

Categorical Data
If the X values are non-numerical labels, without an inherent sequence, they are best plotted
with a column (or bar) chart than a line chart. The column chart simply presents each value
with the corresponding category label.

By connecting points with lines, the line chart (and XY chart) wrongly implies a trending
relationship, and even a continuous relationship, between categories.
An XY chart is a poor choice: since it can’t interpret the text labels as numbers, it replaces
them with the counting numbers 1, 2, 3, etc.

The most logical way to sort categorical data is in increasing or decreasing order of the Y
values. In some cases, though, it may make sense to sort the categories alphabetically to
make it easier to find the value for a particular category; however, this is what a table of data
is good for.

Sequential Data
When X values are sequential, that is, they are in a particular logical order, you can use a line
chart or a column chart. The X values can be a sequence of numbers, or of dates, or of non-
numeric labels which still maintain a sequence, like the month names below.

READING 4.1 | CREATING CHARTS | PAGE 3 OF 11

A line chart accentuates the trend of the data from left to right. The lines imply a continuous
trend in Y values along the X axis, in which case you might interpolate to determine an
intermediate value. A column chart emphasizes the values themselves, and there is no hint
of continuity between columns.

The most logical way to sort sequential data is according to the sequence of X values. It
doesn’t make sense to sort months alphabetically, and if you sort by value, months may get
out of order.

Numerical Data
When the X values are numerical, the best chart type is an XY Scatter chart. The first chart
below shows that the data actually loops around and closes on itself. The line chart doesn’t
show us this at all. If you look at the X axis, you can see why. All of the numbers are listed on
the axis, in the order they appear in the worksheet. Also the numbers are equally spaced,
even though the differences between them are not equal. This is because the line chart
treats the numerical X values as non-numerical text labels, as in the categorical example
above.

Part of this confusion comes from Excel’s own interface. We might know that we want our
data points to be connected by lines. We look at the ribbon (below center) and decide to
click on the line chart icon, which shows lines connecting the data, and reject the XY chart
icon, which only shows unconnected points. It’s only when we click the dropdowns that we
see that XY charts (below right) can be formatted with lines like a line chart (below left). By
this point, though, the wrong chart has been used, and the data has been plotted incorrectly.

If you plot a lot of technical or scientific data, you need to be aware of this problem, and you
should get comfortable with using XY charts.

READING 4.1 | CREATING CHARTS | PAGE 4 OF 11

Dates
When Excel recognizes X values are dates, the X axis of a line chart changes from being a
dumb category axis to a smart date axis. Excel must be able to recognize the dates, so just
labels showing month or year are not sufficient.

The line chart and XY chart below both plotted the unevenly spaced dates proportionally (15
January is closer to 1 January than to 1 March), but the line chart (right) picks nicer axis
scales than the XY chart (left). By default, the line chart starts on the first date and ends on
the last, with labels on the first of each month. XY chart can’t do this, because it can only
have equally spaced labels, and the months have different lengths.

The line chart is even smarter than that: if the dates in the worksheet are out of order, Excel
sorts the data internally, so the points are connected from left to right.

Column and bar charts are not suited for dates. Each bar (or column) must fit within the
space for a day, so the bars end up being very narrow. In addition, the bar chart would plot
dates on the vertical axis, and people generally understand the passage of time best when
plotted left to right on a horizontal axis.

Parts of a Whole
A pie chart is the obvious choice when the data consists of parts of a whole, since its
wedges add up to make a whole circle. Alternatives to a pie chart ate 100% stacked column
and bar charts.

READING 4.1 | CREATING CHARTS | PAGE 5 OF 11

There are further discussions of pie charts and stacked charts later in this chapter.

3D Charts
People naturally like to use 3D charts, because they seem to make charts look nicer.
However, parallax effects, strange perspective views, and data in front obscuring other data
in back, 3D charts distort and hide data. Because of these ill effects, 3D charts should never
be used.

4.1.3 Column and Bar Charts

There are some special considerations you should keep in mind regarding column and bar
charts.

Orientation
Except for the bar chart, the category axis (X axis) is the horizontal axis, and the value axis
(Y axis) is the vertical axis. The bar chart flips the axes around, plotting the categories on the
vertical axis and the values on the horizontal axis. We learned in algebra class that the X axis
is the horizontal axis, but we also learned that the X axis contains the independent variable
and the Y axis the dependent variable, and this second rule takes precedence. Therefore, the
X axis in a bar chart is vertical and the Y axis in a bar chart is horizontal.

In general, it is a matter of preference whether you use a horizontal bar chart or a vertical
column chart. However, when the category axis labels are relatively long, a bar chart is the
better option.

The next few charts compare data for categories with long labels. When a column chart is
selected, Excel tries to fit all labels along the axis, but it rotates the labels and shortens them
(left chart below). People find it difficult to read inclined labels, and when they are shortened,
they lose much of their meaning (North what?).

You can override the label orientation (right chart below), but Excel may decide to omit some
labels to prevent overlap.

READING 4.1 | CREATING CHARTS | PAGE 6 OF 11

You can further override the skipping of labels, but then the overlap makes them illegible
(below left).

This is where a bar chart, below right, has a clear advantage. The labels are horizontal, so
people can read them easily, and they fit in their entirety without overlapping.

Chart Aspect Ratio

The default shape of an Excel chart, that is, its relative width and height, are well suited for
line and column charts. But you may find a bar chart works better when it is taller and not as
wide.

My Bar Chart is Upside Down

When you create a bar chart, the first thing you may notice is that the categories along the
chart’s vertical axis are in the opposite order to those in the worksheet (see data and first
chart below). In addition, the bars are sorted bottom to top rather than top to bottom. There
is a long and tedious explanation for why this happens, but the important thing is that it’s
easy to correct.

READING 4.1 | CREATING CHARTS | PAGE 7 OF 11

Select the vertical axis and press the Ctrl+1 shortcut, which formats whatever you have
selected in Excel. In the task pane, check the box labeled Categories in Reverse Order, which
puts the category axis in the same order as the worksheet categories, but also puts the
horizontal axis at the top of the chart (middle chart above). In the same task pane, find the
Horizontal Axis Crosses setting, and choose At Maximum Category, which brings the
horizontal axis back to the bottom of the chart (right hand chart above).

4.1.4 Line and Scatter Charts

Here are a few things to keep in mind about line and XY charts.

Should I Use Markers?

The default line chart has lines but no markers. The problem with this is that the lines can’t
tell you where each individual data point is located. In the first chart below, there are
obviously points at both ends of the line, at X=1 and X=7. But we don’t know if these are the
only points, or if there are other point along the line. If we use markers, we know where all
the data points are.

Smoothed vs. Straight Lines

Smoothed lines may make a chart look nice, but they can misrepresent your data. Suppose I
am collecting data that must have a value between zero and 100. The chart below left has
smoothed lines, and these lined extend from below zero to above 100. This violates the
actual behavior of the data. The chart below right with straight lines connecting points
shows the actual extent of the data.

READING 4.1 | CREATING CHARTS | PAGE 8 OF 11

Using markers with these charts would help to show where the actual data lies, but the
curved lines still are misleading.

4.1.5 Pies and Donuts

Pie charts are popular, but they’ve gotten a bad rap. Many experts have admonished against
their use: claiming that people can’t judge areas accurately and have other problems with
interpreting pie charts. Recent research has shown this to not be true, when certain
guidelines are followed.

In the charts below, we can easily determine values in the column and bar chart, while we
can only judge relative sizes in the pie chart. And in the pie chart we can only tell that the
dark blue wedge (Bill) is largest, and the light blue wedge (Eve) is smallest, but we can’t say
much about the others.

But column and bar charts provide no insight into how much each bar contributes to the
total, while the pie chart is uniquely qualified for this: we know the dark blue wedge is almost
one-fourth of the total.

If you follow a few simple rules, your pie charts can be successful:

• Don’t use too many pie charts.

• Only use data sets that add up to a meaningful total.
• Sort the values from largest to smallest.
• Avoid using a pie with many points, unless one or two make up a large fraction of
the total.

READING 4.1 | CREATING CHARTS | PAGE 9 OF 11

Donut (or doughnut) charts are like pie charts, with a hole cut out of the middle. There are a
couple more rules for Donut charts:

• Don’t make the donut hole too large: the rings must be large enough to be judged.
• Limit the chart to one series (one ring), because it’s impossible to distinguish one
ring from another without excessive labels, and because the outer ring’s values
appear larger due to a larger diameter.

4.1.6 Stacked Charts

Stacked charts plot each series on top of the previous series, resulting in a cumulative
subtotal in the chart. 100% stacked charts normalize the values to percentages, so each
category adds up to 100%.

Stacked Line Charts

Stacked line charts should be avoided at all costs, because they are hard to read, and in fact,
they are usually used by accident. In the middle and right-hand charts below, there are no
visual cues that the values of the second series have been add onto those of the first. The
second series in the stacked line chart in the middle seems to be twice as large as the first,
when the values (shown in the first chart) are similar in magnitude. The 100% stacked chart
is similarly misleading.

4.1.7 Stacked Area, Column, and Bar Charts

Even unstacked area charts appear to be stacked, except that the left-hand chart below has
some values in the first series that are hidden by the second series. For this reason, it is
probably best to avoid unstacked area charts, while stacked area charts are perfectly
acceptable.

READING 4.1 | CREATING CHARTS | PAGE 10 OF 11

Clustered column and bar charts don’t have any issues with hidden data, and stacked
column and bar charts are fine to use.

If it is important to be able to compare the lengths of bars from different series, one should
use unstacked charts. In a stacked chart, only the first series and the totals can be easily
compared, since the first series and the totals start from the same baseline (the category
axis).

100% stacked charts are good for comparing parts of a whole, but not as good as pie or
donut charts. If you want to compare the constituents of several categories, however, it’s
better to use one stacked column or bar chart, rather than multiple pie charts.

READING 4.1 | CREATING CHARTS | PAGE 11 OF 11