Scatterplots Reference Page

**Purpose:** Scatterplots are used to determine whether
there is some kind of association between two variables.

**Conditions:** None, though scatterplots are be more
informative when both variables are quantitative, or at least ordinal.

**How to Create Them:** Simply set up two axes, as for a
graph, and label each axis with one of the variables. Carefully mark appropriate
scales on your axes, then plot your data points the same way you plotted points
in Algebra class. DO NOT CONNECT THE DOTS. There is no justification for doing
so.

**Interpretation:** Once the plot is constructed, decide
whether there is a pattern in the plot. If so, look for the following things in
it:

Form: If the pattern is roughly straight, call it “linear.” In Math 221, all other patterns can safely be called “nonlinear.” But to be more specific, you can use words and phrases like "bulging" and "fan-shaped."

Direction: If the pattern goes __
up__ as you go left-to-right, say there is a __positive__ association
between the two variables. If the pattern
goes __down__ as you go from left-to-right, say there is a __negative__
association between the two variables. If there is no clear positive or negative
association, or if your perception of a positive or negative association depends
on the presence of one or two points, say there is no association between the
two variables.

Strength: The more closely the points follow the pattern you perceive, the stronger the association is. Weak associations tend to look more like strung-out blobs of points. If the scatterplot looks like a shotgun blast, there probably isn’t any association to speak of.

Deviations: Any point or group of points “far away” from the main body or main pattern should be labeled “unusual” and investigated. There are four types: Outliers are points far above or below the main pattern. Influential points are far away horizontally. Points far away both vertically and horizontally will strongly affect any calculations you make concerning relationships between the variables, as will influential points. Points far away that seem to follow the pattern are of less concern and therefore have no name of their own. In any case, unusual points tend to render conclusions drawn from scatterplots somewhat unreliable.

**SPSS instructions for constructing scatterplots:**
Click here.

**Examples of scatterplots and their interpretation:**
Click here.

**Related topics:**

Pearson’s (linear) correlation coefficient

BYU-Idaho mailto:brownd@byui.edu

232 Ricks Building 208-496-1839 voice

Rexburg, ID 83460 208-496-2005 fax

**Please
do not call me at home.**