SHIFT PLOT

SHIFT PLOT

Graphics Command

Generates a shift plot.

The shift plot is a plot of
y_q - x_q versus x_q

with y_q and x_q denoting the qth quantiles of y and x, respectively. Although this plot can be generated for any pair of variables, the primary usage of this plot is for the case where x is a control group and y is an experimental method. Alternatively, x could represent an "old" method while y represents a "new" method. The plot measures how much the control group must be shifted so that it is comparable to the expermimental method (relative to the quantiles).

The quantiles of a distribution are the distribution's "percent points" (e.g., the 0.5 quantile = the 50% point = the median).

The plot consists of the following:

Vertical axis = difference in the estimated quantiles between data set 1 and data set 2;

Horizontal axis = estimated quantiles from data set 2.

The shift plot is a variation of the quantile-quantile plot and the Tukey mean-difference plot.

SHIFT PLOT <y> <x>       <SUBSET/EXCEPT/FOR qualification>
where <y> is the response variable for the experimental group;
            <x> is the response variable for the control group;
and where the <SUBSET/EXCEPT/FOR qualification> is optional.

With this syntax, the quantiles correspond to the data points of the <x> variable.

SHIFT PLOT <y> <x> <q>       <SUBSET/EXCEPT/FOR qualification>
where <y> is the response variable for the experimental group;
            <x> is the response variable for the control group;
            <q> is a variable (with values between 0 and 1) that specifies the desired quantiles to plot;
and where the <SUBSET/EXCEPT/FOR qualification> is optional.

With this syntax, you explicitly give the desired quantiles.

SHIFT PLOT Y X
SHIFT PLOT Y X SUBSET X > 1
SHIFT PLOT Y X XQ
SHIFT PLOT Y X XQ SUBSET X > 1

None

None

CHARACTERS	= Sets the type for plot characters.
LINES	= Sets the type for plot lines.
QUANTILE-QUANTILE PLOT	= Generates a q-q plot.
TUKEY MEAN-DIFFERENCE PLOT	= Generates a Tukey mean-difference plot.
BIHISTOGRAM	= Generates a bihistogram.
T-TEST	= Carries out a 2-sample t test.
F-TEST	= Carries out a 2-sample F test.

Rand R. Wilcox (1997), "Introduction to Robust Estimation and Hypothesis Testing", Academic Press.

Exploratory Data Analysis

2003/2

 
SKIP 25
READ AUTO83B.DAT Y1 Y2
.
DELETE Y2 SUBSET Y2 < 0
LABEL CASE ASIS
Y1LABEL Difference between Y1 and Y2 Quantiles
X1LABEL Quantiles for Y2
CHARACTER X
LINE BLANK
SHIFT PLOT Y1 Y2