Graphs and When to Use Them

Graphs and When to Use Them

​

Bar Graphs

• 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)

Clustered Bar Graph

• 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.

Bar Graphs: Changing Scale

• 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!

Bar Graphs: Changing Scale Cont.

• 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​

Pareto Chart

• 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

Circle Graph

• 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

Line Graph (Time Plot)

• Shows data measuremnts in chronological order

• Excellent for viewing change over time

• See page 41 for an example

Summary

• 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.

Don't Forget!

• 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.