Members Click Here for Additional Tools
Get Instant Access to 200+ Cheat Sheets, Videos, and Other Immediately Usable Tools for Busy Managers – Try Out a Wily Manager Membership Today!
The Scatterplot Graph is a simple technique that is not often used to help managers make better decisions. Below we talk about the following aspects of the Scatterplot Graph:
- Why would I use a Scatterplot Graph?
- Types of correlations
- Two Examples of how to use the Scatterplot Graph.
- How to do a Scatterplot Graph
Why Would I Use a Scatterplot Graph?
- The Scatterplot Graph can test for possible Cause & Effect Relationships. There are a variety ways to do Cause & Effect analyses, but the Scatterplot Graph is a good place to start if you’ve got some data.
- A Scatterplot Graph can be used for predictive action when the is a strong correlation between variables. This is explained in further detail below.
Types of Correlations in a Scatterplot Graph
When gathering data points, a pattern may or may not emerge. Here are how we label those patterns:
- Positive correlation. If two things are positively related, there will be a visible pattern on the Scatterplot Graph, that moves from the Southwest quadrant to the Northeast quadrant of the graph. If you drew a line between the Scatter plots, most of the data points would be very close to the line.
- Possible positive correlation. This looks much like the pattern above, but the data points are a bit further from the trend line, and it is not as clear as to whether the variables are correlated.
- No correlation. The pattern of the data points on your Scatterplot Graph appear to be random.
- Possible negative correlation. If two things are possibly negatively correlated (ie: more of this is a cause of less of that), then the pattern on the Scatterplot Graph will generally move from the Northwest quadrant to the Southeast quadrant. However, the distance from the trend line may make the pattern less distinct.
- Negative correlation. If two things are negatively correlated, the pattern will be the same as the one above, but will be much more easily recognized, and tightly connected to the trend line.
The R-Squared
Excel makes it easy to determine a trend line for any Scatterplot Graph. Excel will also provide an equation for the line, and an R-squared statistic. The R-squared stat measures the collective distance from the trend line of all the data points. If something is highly correlated, the R-squared number will approach 1. If the data points are not at all correlated, the R-squared number will approach 0.
Example 1
This is actual data from an industrial operation that was testing a theory that their production was largely based on the output of one machine. After tracking their overall output, and the availability of the machine for 30 days, this Scatterplot Graph was produced, showing very little correlation between overall production, and the availability of the machine in question:
Example 2
This is actual data from a retailer that was trying to predict soft drink sales based on outside temperature. As you can see, the Scatterplot Graph shows a very tight, positive correlation between the outdoor temperature, and the volume of soft drink sold. As a result, the retailer could use the equation on the trend line to predict volume for inventory control purposes.
How to do a Scatterplot Graph
Some people will avoid doing a Scatterplot Graph because they think it is time consuming or difficult. It is neither. Here’s how to do it:
- Determine what you are trying to test. What two variables do you want to test a correlation for? The examples above should provide some ideas.
- Gather data (the more points, the better). Ideally, you will want to track 30 – 50 data points as a minimum.
- Put it into a spreadsheet.
- Create a scatter graph. This can be done with the “charts” function in Excel.
- Ask for a trend line. This is in the “tools” menu.
- Ask for the equation. You can do this by right-clicking on the trend line.
Three Things to Remember About the Scatterplot Graph
- Lies, damn lies and statistics. You can probably find data to support anything, so make sure you leave your mind open to what the data you have is telling you.
- This is easy – do it once. It is easy to dismiss this if you aren’t comfortable in Excel or with statistics. It is actually very easy.
- Find someone you work with who is good with Excel if that is what it will take to get this done quickly.
Watch the ‘3-Minute Crash Course’ about the Scatterplot Graph (CLICK THE ARROW TO START THE VIDEO):
Looking for the Full-Length Podcast/Video? …
Get Instant Access to 200+ Cheat Sheets, Videos, and Other Immediately Usable Tools for Busy Managers – Try Out a Wily Manager Membership Today!