Scribd
Upload
55 views

Regression Diagnostics in Minitab

Minitab provides tools for regression diagnostics including partial regression plots, residuals, leverage and influence measures, and variance inflation factors. It produces three types of residuals - residuals, standardized residuals, and deleted residuals. Leverage and influence measures included are leverages (Hi), Cook's distance, and DFITS. While Minitab does not directly produce DFBETAS statistics, they can be calculated manually using outputs from regressions with and without specific observations. Variance inflation factors can be displayed by checking the appropriate option.

Uploaded by

jehana_beth
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
55 views

Regression Diagnostics in Minitab

Minitab provides tools for regression diagnostics including partial regression plots, residuals, leverage and influence measures, and variance inflation factors. It produces three types of residuals - residuals, standardized residuals, and deleted residuals. Leverage and influence measures included are leverages (Hi), Cook's distance, and DFITS. While Minitab does not directly produce DFBETAS statistics, they can be calculated manually using outputs from regressions with and without specific observations. Variance inflation factors can be displayed by checking the appropriate option.

Uploaded by

jehana_beth
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Regression Diagnostics in Minitab

Partial Regression Plots


Minitab does not explicitly produce partial regression plots. Fortunately, they can be
created easily (if tediously, for large models):
(1) regress Y on all Xs except Xk, saving the residuals
(2) regress Xk on all the other Xs, saving the residuals
(3) plot the residuals from (1) against those from (2)
(4) repeat (1)(3) for each X.
Nothing is needed from these regressions but the residuals, so you can use the Results window
to tell Minitab not to print any output.
Residuals
Minitab produces three forms of residuals (all obtained on the Storage window):
Minitab name
residuals

text name
residuals

definition

standardized residuals

studentized residuals

ei divided by its standard error

deleted t residuals

studentized deleted
residuals

di divided by its standard error,


where di is the deleted residual

di = Yi Yi ( i )

ei = Yi Yi

In the Graphs window in the regression procedure, these three kinds of residuals are called
Regular, Standardized, and Deleted, respectively. The standardized residuals are what Minitab
uses to flag unusually large residuals (any observations with standardized residual greater than 2
in absolute value).
Leverage and Influence
The Storage window of the regression procedure provides three measures of leverage
and influence:
Hi (leverages) [hii in the texts notation]
Cooks distance
DFITS

[Di]

[ (DFFITS)i ]

DFBETAS
Minitab does not explicitly produce DFBETAS statistics of influence on particular coefficients. It can be calculated for a particular suspect observation i (perhaps flagged by the preceding
measures), and coefficient k, as follows:
(1) from the regression on the full data set, obtain
bk
(X'X)-1 (this is X'X inverse in the Storage window)
(2) from the regression without observation i, obtain
bk(i)
MSE(i)
(3) find ckk, the kth diagonal element of (X'X)-1
(4) calculate

( DFBETAS ) k ( i ) = ( b k b k ( i ) ) ( MSE ( i ) c kk )

Multicollinearity Diagnostics Variance Inflation Factor


Variance Inflation Factors can be requested on the Options window of the regression
procedure (check box for Variance inflation factors, under Display).

Regression Diagnostics in Minitab

You might also like