You are on page 1of 9

Group Assignment

Statistical Methods for Decision Making

Group 7:
● Budithi Bhargavi
● Ritika Khandelwal
● Shobhit Rao
● Tanu Bhasin
Problem Statement:
To determine the effectiveness the new payment incentive scheme introduced by Titan Insurance
Company.

This study helps to deduce whether the new scheme had made significance difference in sales. To
test the effectiveness of the new scheme a sample sales output of 30 salesperson are measured the
penultimate month before the new scheme was introduced and 4th month after the changeover.

Salesperson Output (£000)


Old Scheme New Scheme
1. 57 62
2. 103 122
3. 59 54
4. 75 82
5. 84 84
6. 73 86
7. 35 32
8. 110 104
9. 44 38
10. 82 107
11. 67 84
12. 64 85
13. 78 99
14. 53 39
15. 41 34
16. 39 58
17. 80 73
18. 87 53
19. 73 66
20. 65 78
21. 28 41
22. 62 71
23. 49 38
24. 84 95
25. 63 81
26. 77 58
27. 67 75
28. 101 94
29. 91 100
30. 50 68
a)

Initial Analysis:

1. Sample size : 30
2. No. of samples : 2
3. Measurement repeated on the same sample after 4 months
4. Basic summary of the data given :

> summary(titaninsurance)
Salesperson old.scheme New.Scheme X output.in..000
Min. : 1.00 Min. : 28.00 Min. : 32.00 Mode:logical Mode:logical
1st Qu.: 8.25 1st Qu.: 54.00 1st Qu.: 55.00 NA's:30 NA's:30
Median :15.50 Median : 67.00 Median : 74.00
Mean :15.50 Mean : 68.03 Mean : 72.03
3rd Qu.:22.75 3rd Qu.: 81.50 3rd Qu.: 85.75
Max. :30.00 Max. :110.00 Max. :122.00

> sd(titaninsurance$old.scheme)
[1] 20.45598
> sd(titaninsurance$New.Scheme)
[1] 24.06239
> var(titaninsurance$old.scheme)
[1] 418.4471
> var(titaninsurance$New.Scheme)
[1] 578.9989
> sd(titaninsurance$New.Scheme - titaninsurance$old.scheme)
[1] 14.08105

Since we are comparing two dependent samples, before and after experiment, we will use Paired t-
Test.
A paired t-test is used to compare two population means where we have two samples in which
observations in one sample can be paired with observations in the other sample. In this
scenario its Before-and-after observations on the new Scheme by Titan Insurance Company.

I) Given:
sample size (n) = 30

II) Hypothesis:

H0 : 𝜇d = 0 (New Scheme did not raise the output significantly)


H1 : 𝜇d < 0 (New Scheme raised the output significantly)

Where, 𝜇d = 𝜇old- 𝜇new

III) Calculating p-value using R:


p=0.0652
Now alpha α = 0.05 (as significance level is 5%)

Since, p > α , We fail to Reject Null Hypothesis.

Based on above hypothesis test, we can say that with 95% confidence level there is no significant
improvement is Sales after the launch of new Incentive Scheme for the given 2 dependent samples.

b)
By plotting some graphs we can see that the Old.Scheme sample and New.Scheme sample have
observations with not much significant increase, some values are increased and some
decreased so the mean difference comes out to be very less.
The mean and median and sd are respectively close.
Also the box plot shows for New Scheme looks left skewed

▪ Here we can see that the mean has increased only a little bit in the New Scheme Plot.
R code:-
boxplot(Old.Scheme,New.Scheme,main="Boxplot of Old & New Schemes",xlab="Old and New
Schemes",ylab="$Sales",col=c("yellow","blue"))
▪ The values above and below the “blue” line shows that New Scheme have more observations
in higher sales region than old scheme but it’s not that significant and observations in the
middle and lower regions are almost same.
R code:-
plot(Old.Scheme,New.Scheme,pch=16)
abline(a=0,b=1,col="blue",lwd=2)

▪ Let’s also plot some graphs for the Difference Sample set i.e. New.Scheme - Old.Scheme
sample set: -
Previously, we failed to reject Null hypothesis. So it can be concluded that, The Data provides
insufficient evidence to conclude that the new scheme has significantly raised the outputs.
c)

o The population may not follow normal distribution. The New and Old Scheme both
seem normally distributed however there is a clear bend towards the end of the old
scheme
o Since we failed to reject Null Hypothesis the probability of Type II error can be high.
o It may be too soon to tell based on mean sales performance whether there is any
change because of the new scheme launch
o The sample size is too small to identify any significant impact
o The higher mean in the case of New scheme can be attributed to the higher variance
compared to Old Scheme
o The variance of the second data set, is 38% more the variance of the first data set.
o We are unsure of the sampling error in the data.
o In order for the company to gauge the complete results, it must wait for another 2
months to get a clear picture of the total 6 months data

d)
To break even, the average output > 5 (in 000 pounds).
i.e. µd <5 considering µd = µold - µnew

H0: µd >= -5

H1: µd < -5

At five per cent significance level, for 29 (30-1) degrees of freedom,

i.e. tstat (critical value)= -1.6991

● We fail to reject the null hypothesis if we get T test stat less than 1.699127 for the sample
size of 30
● For calculating Type 2 error we need 𝐷

𝑡𝑠𝑡𝑎𝑡 = (𝑑 - µd)/𝑆𝑑 )/√𝑛 --------------------------------- (1)

-1.699 = [(𝑑 – 0)/14.081]/ √30

Thus, 𝑑 = (-1.6991*2.5708) = -4.3683

For Ud= -5

𝑡𝑠𝑡𝑎𝑡 = (𝑑 – µd)/𝑆𝑑 )/√𝑛

𝑑 = -4.3683

µd = - 5

n=3

𝑡𝑠𝑡𝑎𝑡 = 0.245713

i) Probability of Type I error = Level of significance = 0.05


ii) Probability of Type II error = Area in Pink

Probability (Dbar>=-4.3683|Ud=-5) = 0.4040 = 40.40%

iii) Power of test = 1-Type II Error = 0.5959 = 59.59%

You might also like