RMIT Classification: Trusted

Business Statistics 1

Tutorial covering Topic 10: Multiple Linear regression

Questions for in-class tutorial:

Q1. A traffic engineer wants to predict the number of road accidents at intersections in a major urban area. He believes the main determinants are
volume of traffic and whether or not the intersection has traffic lights. Traffic lights have the dummy value 1 and no traffic lights the value 0.
Traffic volume is recorded as (‘000/day). A sample of 15 intersections is placed into a table.

SUMMARY OUTPUT

Regression Statistics
Multiple R 0.658
R Square 0.433
Adjusted R Square 0.338
Standard Error 4.904
Observations 15

ANOVA
df SS MS F Significance F
Regression 2 220.310 110.155 4.580 0.033
Residual 12 288.624 24.052
Total 14 508.933

Coefficients
Standard Error t Stat P-value Lower 95%Upper 95%
Lower 95.0%
Upper 95.0%
Intercept 16.561 4.101 4.038 0.002 7.625 25.497 7.625 25.497
Traffi c volume 0.153 0.145 1.054 0.313 -0.163 0.469 -0.163 0.469
Traffi c lights -6.909 2.616 -2.641 0.022 -12.608 -1.209 -12.608 -1.209

a. Use the dataset Accidents.xls to estimate the above regression. State the multiple regression equation.

^
No. of Road Accidents = 16.561 + 0.153 Traffic Volume - 6.909 Traffic Lights at intersections.
 RMIT Classification: Trusted
Business Statistics 1

b. Interpret the meaning of the slopes in this problem.

For every additional 1000 vehicles, the number of accidents increases by 0.153, given the same intersections facilities (or “keeping
everything else constant”).

For every intersection with traffic lights, number of accidents drop by 6.909 when compared with the intersections with no traffic
lights, given the same traffic volume (or “keeping everything else constant”).

Relative to an intersection with no TL, number of accidents drop by 6.909 at an intersection with traffic lights, keeping everything
else constant.

c. Predict the number of accidents for an intersection with lights and 18,000 cars/day.

16.561 + 0.153 x 18 - 6.909 x 1= 12.406 no. of Road accidents on an average.

d. Is there an overall significant relationship between the number of accidents and the two independent variables (traffic volume and
traffic lights) at the 0.05 level of significance? Use the p-value approach.

H0: β1 = β2 = 0,
H1: At least one βj ≠ 0,
α =0.05
F test stat=4.58
As the p-value of 0.033 < 0.05, there is an overall significant relationship between the number of accidents and the two independent
variables (traffic volume and traffic lights).

e. At the 0.05 level of significance, determine whether each independent variable makes a contribution to the regression model. Use
the p-value approach.

For traffic volume:

P-value approach:
H 0 : β 1=0
H 1 : β1 ≠ 0
α =0.05
Since the p-value = 0.313> 0.05, we fail to reject H0. Hence there is no significant relationship between traffic volume and the number of
accidents, therefore, the variable traffic volume doesn’t contribute to the regression model.
Critical t-value approach:
H 0 : β 1=0
H 1 : β1 ≠ 0
α =0.05
The sampling distribution will be from the student-t distribution with 12 degrees of freedom.

For a two-sided test at the 5% level of significance and 12 degrees of freedom (12 = the number of observations – the number of independent
variable -1 = 15-2-1), the critical t values are 2.1788 and -2.1788.

We will reject H0 if t > 2.1788 or t < -2.1788 and fail to reject H0 otherwise.

b1−β 1 0.153−0
t= = =1.054 (You can also read this number from the output table)
Sb
1
0.145

Since your t value =1.054< 2.1788, we fail to reject H0. Hence there is no significant relationship between traffic volume and the
number of accidents, therefore, the variable traffic volume doesn’t contribute to the regression model.

For traffic lights:

P-value approach:
H 0 : β 2=0
 RMIT Classification: Trusted
Business Statistics 1

H 1 : β2 ≠ 0
α =0.05
Since the p-value = 0.022 < 0.05, we reject H0. Hence there is significant relationship between traffic lights and the number of accidents,
therefore, the variable traffic lights does contribute to the regression model.
Critical t-value approach:
H 0 : β 2=0
H 1 : β2 ≠ 0
α =0.05
The sampling distribution will be from the student-t distribution with 12 degrees of freedom.

For a two-sided test at the 5% level of significance and 12 degrees of freedom (12 = the number of observations – the number of independent
variable -1 = 15-2-1), the critical t values are 2.1788 and -2.1788.

We will reject H0 if t > 2.1788 or t < -2.1788 and fail to reject H0 otherwise.

bs −β s −6.909−0
t= = =−2.641 (You can also read this number from the output table)
Sb
2
2.616

Since your t value =-2.641< -2.1788, we reject H0. Hence there is significant relationship between traffic lights and the number of
accidents, therefore, the variable traffic lights does contribute to the regression model.

f. Interpret the adjusted r-squared value.

Approximately 33.8% of the variation in number of accidents is jointly explained by the Traffic Volume and Traffic lights at the
intersection, adjusting for number of observations and independent variables.
Q2. Sales of motor vehicles for a sample of 15 salespeople are compared over the period of one month in a highly intense, large used-car
business in Sydney. The two independent variables are the number of years’ selling experience and the number of different products sold
in the past.

Summary Output          
 RMIT Classification: Trusted
Business Statistics 1

Regression Statistics
Multiple R 0.64
R Square 0.41
Adjusted R
Square 0.31
Standard Error 9.92
Observations 15
 
ANOVA
Significanc
  df SS MS F eF
Regression 2 816.66 408.328797 4.15 0.04
Residual 12 1180.28 98.3563115
Total 14 1996.93      
P- Uppe
Coeffici Standar valu r
  ents d Error t Stat e Lower 95% 95%
Intercept 5.50 6.38 0.86 0.41 -8.39 19.40
years 1.38 0.67 2.05 0.06 -0.09 2.84
products 0.363 0.38 0.93 0.37 -0.48 1.19

a. State the multiple regression equation.

^
Sales = 5.50 + 1.38 Years + 0.363 Products
 RMIT Classification: Trusted
Business Statistics 1

b. Interpret the meaning of the slopes in this problem.

With every additional year of sales experience, sale of motor vehicles increases by 1.38, given the same number of products
sold in the past.

For every extra product sold in the past, sale of motor vehicles increases by 0.363, given the same years of sales experience.

c. Predict the number of sales for a salesperson with 14 years of selling experience and 10 different types of products sold in the past.

Sales = 5.50 + 1.38 Years + 0.363 Products

d. (self-study) Is there a significant relationship between sales and the two independent variables (selling years and products sold) at the
0.05 level of significance? Use the p-value approach.
H0: β1 = β2 = 0,
H1: At least one βj ≠ 0,
α =0.05
As p-value of 0.04 < 0.05 there is a significant relationship between sales and the two independent variables (selling years and
products sold)

e. At the 0.05 level of significance, determine whether each independent variable makes a contribution to the regression model. Use the
critical value approach.

For variable: years of experience: H 0 : β 1=0

The sampling distribution will be from the student-t distribution with 12 degrees of freedom.

For a two-sided test at the 5% level of significance and 12 degrees of freedom (12 = the number of observations – the number of
independent variable -1 = 15-2-1), the critical t values are 2.1788 and -2.1788.
 RMIT Classification: Trusted
Business Statistics 1

We will reject H0 if t > 2.1788 or t < -2.1788 and fail to reject H0 otherwise.

b1−β 1 1.38−0
t= = =2.05 (You can also read this number from the output table)
Sb
1
0.67

Since the t-statistic is not in the rejection region, so we cannot reject the H0. We cannot conclude at the 5% level of significance that
the number of years of experience makes a contribution to the regression model.

For variable: number of products H 0 : β 2=0

The sampling distribution will be from the student-t distribution with 12 degrees of freedom.

For a two-sided test at the 5% level of significance and 12 degrees of freedom, the critical t values are 2.1788 and -2.1788.
We will reject H0 if t > 2.1788 or t < -2.1788 and fail to reject H0 otherwise.

b1−β 1 0.363−0
t= = =0.93
Sb
1
0.38

Since the t-statistic is not in the rejection region, so we cannot reject the H0. We cannot conclude at the 5% level of significance that
the number of products makes a contribution to the regression model