TOPICS
Search

Anscombe's Quartet


AnscombesQuartet

Anscombe's quartet is a collection of four data sets, each consisting of 11 pairs (x,y), with nearly identical summary statistics but very different scatter plots (Anscombe 1973).

In each data set, the sample mean of x is 9 and the sample variance of x is 11. The sample mean of y is approximately 7.50, the sample variance of y is about 4.12 to 4.13, and the Pearson correlation coefficient ranges from about 0.81624 to 0.81652. Each data set gives nearly the same least squares fitting line, y^^=3+0.5x to the displayed precision.

The first data set has points scattered around a straight line, while the second follows a curved pattern. In the third, one outlier pulls the fitted line away from the pattern of the other points. In the fourth, ten observations have x=8, and the remaining observation at (19,12.50) determines the slope together with the mean of the ten responses at x=8. Without the observation at x=19, the slope cannot be estimated.


See also

Correlation Coefficient, Linear Regression, Outlier, Scatter Diagram

Explore with Wolfram|Alpha

References

Anscombe, F. J. "Graphs in Statistical Analysis." Amer. Statist. 27, 17-21, 1973. https://doi.org/10.1080/00031305.1973.10478966.

Cite this as:

Weisstein, Eric W. "Anscombe's Quartet." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/AnscombesQuartet.html

Subject classifications