Multiple linear regression is a linear relationship between two or more independent variables (x1,

x2,… xn) with the dependent variable (y)

This analysis is used to determine the relationship direction between the independent variable with
the dependent variable whether each independent variable is positively or negatively related and to
predict the value of the dependent variable if the value of the independent variable has increased or
decreased. The data used is usually interval or ratio scale.

Multiple regression analysis is used by researchers, if researches intend to predict how the state (ups
and downs) of independent variables (criteria), if two or more independent variables as a predictor
factor is manipulated (dinaik turunkan nilainya). So the multiple regression analysis will be used if
the number of independent variables is at least 2.

The formula of multiple linear regression analysis is :

Y’ = a + b 1 X 1 +b 2 X 2 + ⋯ bn X n

Explanation :

Y’ : Dependent variable (predicted value)

X : Independent Variable

a : a Constant

b : regression coefficient (increase or decrease value)

 Multiple Correlation Analysis (R)

This analysis is used to determine the relationship between 2 or more independent variables
on the dependent variable simultaneously. R values range is between 0 and 1. The closer it
get to the number 1, the stronger the relationship is, otherwise, the closer it get to 0, the
weaker relationship is.

The multiple correlation formula with 2 independent variables is :

2 2
( ry x 1 ) + ( ry x 2 ) −2 ⋅ ( ry x1 ) ⋅ ( ry x 2 ) ⋅ ( r x 1 x2 )
Ry ⋅ x 1 x 2=
√ 1−( r x 1 x 2 )
2

Explanation

Ry ⋅ x 1 x 2 =The correlation between variables X1 with X2 to Y

ry x 1 = simple correlation (pearson product moment) between X1 and Y

ry x 2 = Simple correlation between X2 with Y

r x1 x2 = Simple correlation between X1with X2

According to Sugiyono (2007) guidelines for providing interpretation of correlation coefficients are :
 value Equivalent

0,00 –
Very low
0,199

0,20 –
low
0,399

0,40 –
Average
0,599

0,60 –
Strong
0,799

0,80 –
Very strong
1,000

 Determination Analysis (R2)

This analysis is used to determine the percentage contribution of independent variables
influence simultaneously on the dependent variable. If the R2 is 0 then there is no slightest
influence of contribution percentage given by the independent variable to the dependent
variable, but if the R2 is 1 then the contribution of influence that is given by the independent
variable to the dependent variable is perfect.

The formula to find the coefficient of determination with 2 independent variables as :

2 2
2( ry x 1 ) + ( r y x 2 ) −2 ⋅ ( ry x 1 ) ⋅ ( ry x 2 ) ⋅ ( r x1 x 2 )
R=
1−( r x 1 x 2 )2

Explanation :
R2 = Coefficient of determination

ry x 1 = simple correlation (pearson product moment) between X1 and Y

ry x 2 = Simple correlation between X2 with Y

r x 1 x 2 = Simple correlation between X1with X2

 Joint Regression Coefficient Test (F Test)

This test is used to determine whether the independent variables together significantly
influence the dependent variable, or used to determine whether the regression model can
be used to predict the dependent variable or not. Significant means the relationship that
occurs can apply to the population (can be generalized)

The formula to find for F arithmetic is :

 R2 ∕ k
F value=
( 1−R2 ) ∕ ( n−k−1 )
Explanation :

R2 = Coefficient of determination
n = The number of data or case
k = The number of independent variable

The steps in testing the regression coefficient together as are :

a. Hypothesis
H0 : there is no significant effect
H1 : there is significant effect
b. Significance level
Usually using level 5%
c. Test Method
F test statistics
d. Rejection Regions
Rejecting the null hypothesis when F value ≥ F a ;ⅆ f ;ⅆ f
1 2

Df1 : the number of independent variable

Df2 : n-k-1
e. Decision
f. Conclusion

 Partial Regression coefficient test (t test)

This test is used to determine whether in the regression model the independent variable
partially has a significant effect on the dependent variable.

t value formula in this regression is :

bi r √ n−k −1
t= or t=
Sbⅈ √ 1−r 2
Explanation :
bi = Variable regression coefficient i
Sbⅈ = Standard error variable i
r = Partial correlation coefficient
k = The number of independent variable
n = The number of data or case

The steps for testing the partial regression coefficient are as follows :
a. Hypothesis
H0 = Partially there is no significant effect
H1 = Partially there is significant effect
b. Significance level
Usually using level 5%
c. Test Method / Test Statistics
t Test Statistics
d. Rejection Regions
 Rejecting the null hypothesis when |t |≥t α ;n−k−1
2

e. Decision
f. Conclusion

Sample Case

A capital market observer named Eiden wants to do research on the factors that affecting stock
prices at companies on the Indonesia Stock Exchange (Bursa Efek Indonesia). Eiden in his research
wanted to find out the relationship between the financial ratios of PER and ROI (Return on
Investment) against stock prices. From this statement, we obtained, the dependent variable (Y) is
the stock price, and the independent variables (X1 and X2) are PER and ROI.

NO Year Y (stock price) X1 (PER%) X2 (ROI%)

1 2000 7500 3.28 3.14
2 2001 8950 5.05 5.00
3 2002 8250 4.00 4.75
4 2003 9000 5.97 6.23
5 2004 8750 4.24 6.03
6 2005 10000 8.00 8.75
7 2006 8200 7.45 7.72
8 2007 8300 7.47 8.00
9 2008 10900 12.68 10.40
10 2009 12800 14.45 12.42
11 2010 9450 10,50 8,62
12 2011 13000 17,24 12,07
13 2012 8000 15,56 5,83
14 2013 6500 10,85 5,20
15 2014 9000 16,56 8,53
16 2015 7600 13,24 7,37
17 2016 10200 16,98 9,38
18 2017 10600 16,57 9,20
19 2018 9270 14,83 8,82
20 2019 11430 16,93 10,25

On the data view, type the data on SPPS. On the 1 st column enter the stock price data, on the 2 nd
column enter the PER data, and on the 3 rd column enter the ROI data.
Click the “variable view”, then in the “name” column, replace the 1 st row with Y, the 2nd row with X1
and the 3rd row replace with X2.

For the decimals column, changed the Y variable to 0 while the others leave it filled with 2.

 Click analyse -> click regression -> click linear

 Transfer the independent variable X1 and X2 (PER and ROI) to the independent box
 Transfer the dependent variable Y (stock price) to the dependent box
 Click statistics : choose estimates, model fit, descriptive

 Click continue
 Click plots on standardized residual plot, choose histogram and normal probability
  Click continue and click OK

The output can be seen as :

Coefficientsa
Standardized
Unstandardized Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 4604,424 661,443 6,961 ,000
X1 -64,991 56,322 -,190 -1,154 ,264
X2 697,671 114,543 1,005 6,091 ,000
a. Dependent Variable: Y

 A constant of 4604,424; means that if X1 (PER) and X2 (ROI) value is 0, then the Y (stock
price) value is Rp. 4604,424
 X1 (PER) variable regression coefficient of -64,991; means if X1 (PER) increases by 1% then
the Y (Stock Price) will decrease by Rp. 64,991 assuming other independent variables have a
fixed value. Negative coefficient means that there is a negative relationship between PER
and stock price, the higher PER get then the lower the stock price get.
 X2 (ROI) variable regression coefficient of 697,671; means that if ROI increases 1%, then the
stock price will increase by Rp. 697,671 assuming other independent variables have a fixed
value. Positive coefficient means that there is a positive relationship between ROI and stock
price. The higher ROI then the stock price also will get higher.

To present the regression equation as :

Y ' =a+ b1 X 1 +b2 X 2

Y ' =4604,424+ (−64 , 991 ) X 1 +697,671 X 2

Y ' =4604,424−64,991 X 1 +697,671 X 2
 Multiple Correlation Analysis (R)

Model Summaryb
Adjusted R Std. Error of the
Model R R Square Square Estimate
a
1 ,879 ,773 ,746 860,648
a. Predictors: (Constant), X2, X1
b. Dependent Variable: Y

 Based on the output, R value is 0.879

 Because the multiple correlation value is between 0.80 – 1,000, it can be concluded that
there is a very strong relationship between PER (X1) and ROI (X2) on stock price (Y)

 Determination Analysis ( R2)

Model Summaryb
Adjusted R Std. Error of the
Model R R Square Square Estimate
a
1 ,879 ,773 ,746 860,648
a. Predictors: (Constant), X2, X1
b. Dependent Variable: Y

 Based on the output, The R2 (R Square) value is 0,773 or 77,3%

 This shows that the percentage contribution of independent variables influence (PER and
ROI) to the dependent variable (Stock Price) is 77,3%
 Or the independent variables variation used in the model (PER and ROI) are able to explain
77,3% of dependent variable (stock price) variation. While the remaining 22,7% is influenced
or explained by other variables that is not included in this research model.

 Joint Regression Coefficient Test (F Test)

a. Hypothesis
H0 : there is no significant effect between PER and ROI together on stock price
H1 : there is significant effect between PER and ROI together on stock price

b. Significance level
Usually using level 5%

c. Test Method/ Test Statistics

F test Statistics

ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression 42891144,751 2 21445572,376 28,953 ,000b
Residual 12592155,249 17 740715,015
Total 55483300,000 19
a. Dependent Variable: Y
b. Predictors: (Constant), X2, X1

 Based on the table the F value is 28,953

d. Rejection Regions
Rejecting the null hypothesis when F value ≥ F a ;ⅆ f 1 ;ⅆ f 2
F value ≥ F 0,05 ;2 ;17
F value ≥3,5 9 2

α : 0.05
df 1 : 2 (the number of independent variables)
df 2 : n-k-1
: 20-2-1
: 17

e. Decision
Since the F value (28,953) is more than the F table (3,592) then we can’t accept the null
hypothesis
f. Conclusion
We can conclude that there is significant effect between PER and ROI together on stock
price

 Partial Regression coefficient test (t test)

1. Testing the PER variable regression coefficient
 a. Hypothesis
H0 = Partially there is no significant effect between PER with stock price
H1 = Partially there is significant effect between PER with stock price
b. Significance Leve
α : 5% = 0,05
c. Test Method/Test Statistic
t test statistics

Coefficientsa
Standardized
Unstandardized Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 4604,424 661,443 6,961 ,000
X1 -64,991 56,322 -,190 -1,154 ,264
X2 697,671 114,543 1,005 6,091 ,000
a. Dependent Variable: Y
 Based on the table the t value of X1 (PER) is -1,154

d. Rejection Regions
Rejecting the null hypothesis when |t |≥t α ;n−k−1
2
|t|≥t 0,05
;20−2−1
2
|t|≥t 0,025 ;17
|t |≥2,110
e. Decision
Since the t value (-1,154) is less than the t table t 0,025;17 (2,110) then we can accept the
null hypothesis
f. Conclusion
We can conclude that, Partially there is no significant effect between PER with stock
price

2. Testing the ROI variable regression coefficient

a. Hypothesis
H0 = Partially there is no significant effect between ROI with stock price
H1 = Partially there is significant effect between ROI with stock price
b. Significance Level
α : 5% = 0,05
c. Test Method
t Test Statistics

Coefficientsa
Standardized
Unstandardized Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 4604,424 661,443 6,961 ,000
X1 -64,991 56,322 -,190 -1,154 ,264
 X2 697,671 114,543 1,005 6,091 ,000
a. Dependent Variable: Y
Based on the table the t value of X2 (ROI) is 6,091
d. Rejection Regions
Rejecting the null hypothesis when |t |≥t α ;n−k−1
2
|t|≥t 0,05
;20−2−1
2
|t|≥t 0,025 ;17
|t |≥2,110
e. Decision
Since the t value (6,091) is more than the t table t 0,025;17 (2,110) then we cannot accept
the null hypothesis
f. Conclusion
We can conclude that, Partially there is significant effect between ROI with stock price