Scribd
Upload
45 views4 pages

Spss Case

The document analyzes data from the Springville Herald newspaper to predict new subscriptions based on telemarketing hours. A linear regression model found that telemarketing hours explained 86.2% of the variation in new subscriptions and significantly impacted new subscriptions. However, the assumptions of normality and homoscedasticity required for linear regression were not met based on the residual analysis, so predictions may not be reliable.

Uploaded by

Aastha
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
45 views4 pages

Spss Case

The document analyzes data from the Springville Herald newspaper to predict new subscriptions based on telemarketing hours. A linear regression model found that telemarketing hours explained 86.2% of the variation in new subscriptions and significantly impacted new subscriptions. However, the assumptions of normality and homoscedasticity required for linear regression were not met based on the residual analysis, so predictions may not be reliable.

Uploaded by

Aastha
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 4

THE SPRINGVILLE HERALD CASE

Qns 9.1
- As in the last 3 months new subscription has sometimes decreased and
increased so the method of forecasting is not consistent.
Qns 9.2
- The factor other than the number of telemarketing hours spent which might
be useful in predicting the number of new subscriptions is as follows:
- Trend
- Price
- Training etc
Qns 9.3
→ In this case, the dependent variable is New Subscriptions and Independent
Variable is Telemarketing Hours.

Model Summary
Std. Error of the
Model R R Square Adjusted R Square Estimate Durbin-Watson
1 .929 a
.862 .856 409.199 1.703
a. Predictors: (Constant), Telemarketing Hours
b. Dependent Variable: New Subscription

(r2) =0.862 i.e 86.2% of variation in new subscriptions is explained by


telemarketing hours.
Adjusted r2 =0.856 i.e 85.6 % of variation in new subscriptions is explained by
telemarketing hours after adjusting by degree of freedom.
Standard Error of the Estimate (S.E) = 409.199 i.e The variability of observed
value of new subscriptions around regression line is 409.199
According to Durbin –Waston Test it shows there is no autocorrelation among
residuals.
ANOVA TABLE
ANOVAa

Model Sum of Squares df Mean Square F Sig.


1 Regression 23032883.654 1 23032883.654 137.556 .000b
Residual 3683772.305 22 167444.196
Total 26716655.958 23
a. Dependent Variable: New Subscription
b. Predictors: (Constant), Telemarketing Hours

From the regression ANOVA table, we can see that the significance value is 0.00
(i.e p=0.00) which is below 0.05 therefore there is significant difference in new
subscriptions and telemarketing hours.
The slope of regression (b1) is 4.409 i.e when the number of telemarketing hours
increased by 1 unit then a new subscription is expected to increase by 4.409 units.
And y-intercept (b0) is -326.710 i.e the new subscription due to factors other than
tele marketing hours is -326.710 units.
Coefficientsa

Standardized
Unstandardized Coefficients Coefficients Collinearity Statistics
Model B Std. Error Beta t Sig. Tolerance VIF
1 (Constant) -326.710 416.941 -.784 .442
Telemarketing Hours 4.409 .376 .929 11.728 .000 1.000 1.000
a. Dependent Variable: New Subscription

H0 : There is not significant impact of new subscription on telemarketing hours.


H1 : There is significant impact of new subscription on telemarketing hours.
Decision: p-value (0.00) < α (0.05) H0 is rejected.
Hence, there is significant impact of new subscription on telemarketing hours.
According to the above diagram, the residual analysis for normality is non normal.

According to the scatter diagram, there exist heteroscedasticity.


b) Soln
When X = 1000 hours then,

Ŷ = -326.710 + 4.409X
= -326.710 + 4.409*1000
= 4082.29 units
N=24 K=1
α = 0.05
D-W Value=1.703 (tab)
From the Durbin Watson table
DL=1.273 Du=1.446
Hence, there is no evidence of autocorrelation among residuals.
The assumptions of linear regression upon which prediction is based are listed
below:
 Normality
 Homoscedasticity
 Independence of error

The assumption isn’t valid as the assumption of linear regression does not match
the results we obtained. According to the above histogram, the data seems to non -
normal and also the scatter diagram shows heteroscedasticity nature.

c) Soln
Telemarketing hour (X) = 2000
According to the question maximum value of X is 1498
Let’s suppose that 10% variation in 1498 = 149.8
+ 1498
1647.8
So, the value greater than it wouldn’t be reliable.

You might also like