Vous êtes sur la page 1sur 20

1

Statistics
Statistics

Correlation
Correlationand
andRegression
RegressionAnalysis
Analysis

Alan
Alan D.
D. Smith

Correlation
Correlation and
and
Regression
Regression Analysis
Analysis

GOALS
GOALS

TO
TO DRAW
DRAWSCATTER
SCATTER DIAGRAMS.
DIAGRAMS.

TO
TO CALCULATE
CALCULATE AND
AND DISCUSS
DISCUSS PEARSONS
PEARSONS
CORRELATION
CORRELATION COEFFICIENT.
COEFFICIENT.

TO
TO CALCULATE
CALCULATE AND
AND DISCUSS
DISCUSS THE
THE
COEFFICIENT
COEFFICIENT OF
OFDETERMINATION.
DETERMINATION.

TO
TO USE
USE THE
THE LEAST
LEAST SQUARES
SQUARES METHOD
METHOD
TO
TO DETERMINE
DETERMINE THE
THE REGRESSION
REGRESSION
EQUATION.
EQUATION.

USING
USING EXCEL
EXCEL FOR
FOR REGRESSION
REGRESSION
ANALYSIS
ANALYSIS

CORRELATION
CORRELATION ANALYSIS
ANALYSIS

Correlation
Correlation Analysis
Analysis -- A
Agroup
group of
of statistical
statistical
techniques
techniques used
used to
to measure
measure the
the strength
strength of
of the
the
relationship
relationship (correlation)
(correlation) between
between two
two variables.
variables.

Scatter
Scatter Diagram
Diagram -- aa chart
chart that
that portrays
portrays the
the
relationship
relationship between
between the
the two
two variables
variables of
of
interest.
interest.

Dependent
Dependent Variable
Variable -- the
the variable
variable that
that isis being
being
predicted
predicted or
or estimated.
estimated.

Independent
Independent Variable
Variable -- the
the variable
variable that
that provides
provides
the
the basis
basis for
for estimation.
estimation. It
It isis the
the predictor
predictor
variable.
variable.

THE
THE COEFFICIENT
COEFFICIENT OF
OF CORRELATION,
CORRELATION, rr

The
The Coefficient
Coefficient of
of Correlation,
Correlation, rr -- isis aa measure
measure of
of
the
the strength
strength of
of the
the linear
linear relationship
relationship between
between
two
two variables.
variables.

It
It can
can range
range from
from -1.00
-1.00 to
to +1.00.
+1.00.

Values
Values of
of -1.00
-1.00 or
or +1.00
+1.00 indicate
indicate perfect
perfect and
and
strong
strong correlation.
correlation.

Values
Values close
close to
to 0.0
0.0 indicate
indicate weak
weak correlation.
correlation.

Negative
Negative values
values indicate
indicate an
an inverse
inverse relationship
relationship
and
and positive
positive values
values indicate
indicate aa direct
direct relationship.
relationship.

PERFECT
PERFECT NEGATIVE
NEGATIVE CORRELATION
CORRELATION

PERFECT
PERFECT POSITIVE
POSITIVE CORRELATION
CORRELATION

ZERO
ZERO CORRELATION
CORRELATION

STRONG
STRONG POSITIVE
POSITIVE CORRELATION
CORRELATION

FORMULA
FORMULA FOR
FOR rr

10

nn
XY

XX
YY

XY

rr

2
2

2
2

2
2
2
2
n

X
X
Y
Y
n

n
X

X
n
Y

22
COEFFICIENT
OF
DETERMINATION,
r
COEFFICIENT OF DETERMINATION, r

11

The
The Coefficient
Coefficient of
of Determination,
Determination, rr22 -- the
the
proportion
proportion of
of the
the total
total variation
variation in
in the
the dependent
dependent
variable
variable YY that
that isis explained
explained or
or accounted
accounted for
for by
by
the
the variation
variation in
in the
the independent
independent variable
variable X.
X.

The
The coefficient
coefficient of
of determination
determination isis the
the square
square of
of
the
the coefficient
coefficient of
of correlation.
correlation.

It
It can
can range
range from
from 00 to
to 1.0.
1.0.

EXAMPLE
EXAMPLE

Lance
Lance Engle
Engle isis president
president of
of the
the student
student body
body at
at
The
The Computer
Computer University
University (CU).
(CU). He
He isis concerned
concerned
about
about the
the cost
cost of
of textbooks
textbooks at
at CU.
CU. To
Toprovide
provide
insight
insight into
into the
the problem
problem he
he selects
selects aa sample
sample of
of
eight
eight textbooks
textbooks currently
currently on
on sale
sale in
in the
the
bookstore.
bookstore. He
He decides
decides to
to study
study the
the relationship
relationship
between
between the
the number
number of
of pages
pages in
in the
the text
text and
and
cost.
cost. The
The collected
collected data
data isis given
given on
on the
the next
next
slide.
slide.

Compute
Compute the
the correlation
correlation coefficient.
coefficient.

Answer:
Answer: rr == 0.614
0.614

12

EXAMPLE
EXAMPLE

(continued)
(continued)

13

EXAMPLE
EXAMPLE (continued)
(continued)

14

REGRESSION
REGRESSION ANALYSIS
ANALYSIS

Purpose
Purpose -- to
to determine
determine the
the regression
regression equation.
equation.
It
It isis used
used to
to predict
predict the
the value
value of
of one
one variable
variable
(Y,
(Y, called
called the
the dependent
dependent variable)
variable) based
based on
on
another
another variable
variable (X,
(X, called
called the
the independent
independent
variable).
variable).

Procedure:
Procedure:

Select
Select aa sample
sample from
from the
the population,
population, and
and list
list the
the
paired
paired data
data (X,
(X, Y)
Y) for
for each
each observation.
observation.

Draw
Draw aa scatter
scatter diagram
diagram to
to give
give aa visual
visual
portrayal
portrayal of
of the
the relationship.
relationship.

Determine
Determine the
the regression
regression equation
equation YY== aa ++ bX.
bX.

15

REGRESSION
REGRESSION ANALYSIS
ANALYSIS

16

YY isis the
the average
average predicted
predicted value
value of
of YY for
for any
any X.
X.

aa isis the
the Y-intercept,
Y-intercept, or
or the
the estimated
estimated YY value
value
when
when XX == 0.
0.

bb isis called
called the
the slope
slope of
of the
the line.
line. It
It isis the
the average
average
change
change in
in YY for
for each
each change
change of
of one
one unit
unit in
in X.
X.

The
The least
least squares
squares principle
principle isis used
used to
to obtain
obtain aa
and
and bb and
and are
are given
given by:
by:
nn
XY

X
YY

XY
X

bb
2

2
2
2
nn
X

X X

Y
X

aa n bb nX
n
n

EXAMPLE
EXAMPLE (continued)
(continued)

17

Develop
Develop aa regression
regression equation
equation for
for the
the
information
information given
given in
in the
the EXAMPLE
EXAMPLE that
that can
can
be
be used
used to
to estimate
estimate the
the selling
selling price
price based
based on
on the
the
number
number of
of pages.
pages.

bb == 0.01714,
0.01714, aa == 16.00175.
16.00175.

YY == 16.00175
16.00175 ++ 0.01714X
0.01714X ..

What
What isis the
the estimated
estimated selling
selling price
price of
of aa 650-page
650-page
book?
book?

YY == 16.00175
16.00175 ++ 0.01714(650)
0.01714(650) == $27.14.
$27.14.

Chapter
Chapter 12
12 Homework
Homework

Chapter
Chapter 12:
12: CD-ROM
CD-ROM

18

EXCEL
EXCEL

Tools
Tools
Data
DataAnalysis
Analysis
Regression
Regression

19

20
FOR NEXT TIME
Please read Chapter 13 : ANOVA

Vous aimerez peut-être aussi