Anscombe's Quartet: Understanding Different Stories Told by Residuals, Study notes of Descriptive statistics

Anscombe's quartet is a famous example in statistics that demonstrates how identical fitted lines and equal residual sums of squares can hide significant differences in the distribution of residuals. an overview of Anscombe's quartet, explains how to calculate residuals, and discusses the importance of analyzing residuals to understand the underlying data.

Typology: Study notes

2021/2022

Uploaded on 09/12/2022

gerrard
gerrard 🇮🇹

3.9

(7)

213 documents

1 / 5

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
Residuals: Anscombe’s quartet
5 10 15 20
2 4 6 8 10 12 14
x
y
5 10 15 20
2 4 6 8 10 12 14
x
y
5 10 15 20
2 4 6 8 10 12 14
x
y
5 10 15 20
2 4 6 8 10 12 14
x
y
pf3
pf4
pf5

Partial preview of the text

Download Anscombe's Quartet: Understanding Different Stories Told by Residuals and more Study notes Descriptive statistics in PDF only on Docsity!

Residuals: Anscombe’s quartet

x y 5 10 15 20

x y 5 10 15 20

x y 5 10 15 20

x y

Residuals: Anscombe’s quartet

x y 5 10 15 20

x y 5 10 15 20

x y 5 10 15 20

x y

Why are the four sets of residuals different?

obvious if you have only one explanatory variable

 - res1 res2 res3 res 
  • 1 -0.740 -1.901 0.389 0.
    • 2 0.179 -0.761 0.229 -0.
    • 3 1.239 0.129 0.079 -1.
  • 4 -1.681 0.759 -0.081 0.
  • 5 -0.051 1.139 -0.230 -1.
    • 6 1.309 1.269 -0.390 1.
    • 7 0.039 1.139 -0.540 -0.
  • 8 -0.171 0.759 -0.689 1.
    • 9 1.839 0.129 -0.849 -1.
  • 10 -1.921 -0.761 3.241 0.
  • 11 -0.041 -1.901 -1.159 0.

Analysing residuals

check that their average value is (close to) zero

plot them against fitted values

plot them against explanatory variables in the model

plot them against explanatory variables not in the model

(for example, residuals against time-order)