Graphs and When to Use Them

Used to compare categorical data

Categories on the x - axis

Frequency, percentage or amount on the y - axis

Always include title, axes labels and scale (for y-axis)

Allows for an additional level of comparison (another categorical variable can be compared)

For example: pg. 38 displays decade born on the x - axis and life expectancy on the y - axis. Thus we can compare the life expectancy across decades and gender.

Sometimes, to emphasize the differences in values on graphs, we compress the scale. We alert the reader of the graph to this by inserting a “squiggle” or “break” on the scale.

Ideally we want to avoid this. However, certain situations make this a necessity. The important aspect is to make it clear to the graph reader you are changing the scale.​

Be very careful to do this with bar graphs! This is how data can become misleading! Don’t do this unless you have to!

Times we might change the scale...

• ​If zero doesn’t make sense as a value, we might change the scale

• If we have an extreme data value, we might edit the scale.

  • We can even do this with a singular bar.

Better alternatives to changing the scale...

• Use a dot plot instead

• Change your y-axis to percentages (relative frequency)

Use good judgement and be aware that changing scales can be misleading! Always use a squiggle if you must change the scale​

Specific type of bar graphs

Categorical variables on the x - axis are organized from greatest frequency to least

Great for quality control (what is the product sold the least, what is the number one reason you are late, etc.)

See visual and read the example on page 39

Good for showing the division of something into percentages or component parts

To make a circle graph: take different categories, sum how often each occurs, divide into fractional parts.

Think parts of a whole

Just because a variable is discussed in terms of percentages, does NOT mean a circle graph is appropriate. Must be a total broken into parts.

See page 40 for an example

Shows data measuremnts in chronological order

Excellent for viewing change over time

See page 41 for an example

Bar graphs - Great for comparing categories of data

Pareto charts - Frequency of events or categories in order of decreasing frequency. Excellent for quality control testing.

Pie charts- show how a total is broken up into several categories.

Time plots - Visualizing how data changes over time.

Always title your graph.

Always label your axes.

Always create a legend or scale to help interpret your graph.

Make your graph truthful, clear, concise and to the point.