Order

49
47
73
17
46
51
75
19
13
40
74
18
11
3
24
10
50
28
56
31
2
71
60
14
33
26
9
70
76
20
6
12
29
67
32
45
15
69
36
7
42
5
66
63
52
61
34
4
65
62
43
68
41
27
38

Deliverer PrepTimeTravelTime
1
14.6
1.9
1
11.7
3.3
1
9.0
3.5
1
12.1
5.0
1
14.0
5.3
1
7.1
7.2
1
11.0
11.6
1
7.9
9.0
1
14.9
8.7
1
10.3
9.5
1
13.1
11.9
1
9.9
9.9
1
12.0
12.9
1
9.2
9.2
1
11.0
10.5
1
14.3
13.4
1
9.5
12.6
1
9.2
13.4
1
13.0
12.6
1
9.3
22.8
1
10.9
17.8
1
17.3
16.6
1
9.5
21.6
1
9.7
29.9
1
11.4
26.1
2
10.1
2.6
2
9.8
4.9
2
8.7
7.0
2
12.0
7.5
2
17.2
6.5
2
9.5
8.1
2
13.0
8.8
2
18.7
8.9
2
10.9
17.8
2
10.1
14.5
2
9.7
13.8
2
10.0
18.2
2
13.4
18.9
2
15.4
21.2
2
7.1
31.6
2
9.4
23.5
2
14.5
21.9
2
12.9
29.5
3
17.0
6.0
3
8.5
7.3
3
13.1
10.1
3
18.0
10.8
3
7.2
14.7
3
8.8
13.6
3
8.4
24.1
3
6.6
17.0
3
9.7
23.8
3
14.4
29.7
3
8.2
19.0
3
13.9
21.9

Distance
0.6
1.4
1.5
1.7
2.5
3.3
3.6
3.7
4.2
4.3
4.4
4.5
4.5
4.7
5.0
5.3
6.5
7.0
7.2
8.2
8.6
9.3
11.6
11.7
12.9
1.4
2.6
2.9
3.3
3.4
3.6
4.1
5.6
6.9
7.7
9.2
9.5
10.1
10.7
11.3
12.2
12.2
14.0
2.8
3.0
3.1
3.8
4.3
4.6
5.8
6.3
6.4
6.9
7.5
8.4

Total
16.5
15.0
12.5
17.1
19.3
14.3
22.6
16.9
23.6
19.8
25.0
19.8
24.9
18.4
21.5
27.7
22.1
22.6
25.6
32.1
28.7
33.9
31.1
39.6
37.5
12.7
14.7
15.7
19.5
23.7
17.6
21.8
27.6
28.7
24.6
23.5
28.2
32.3
36.6
38.7
32.9
36.4
42.4
23.0
15.8
23.2
28.8
21.9
22.4
32.5
23.6
33.5
44.1
27.2
35.8

Speed
18.95
25.45
25.71
20.40
28.30
27.50
18.62
24.67
28.97
27.16
22.18
27.27
20.93
30.65
28.57
23.73
30.95
31.34
34.29
21.58
28.99
33.61
32.22
23.48
29.66
32.31
31.84
24.86
26.40
31.38
26.67
27.95
37.75
23.26
31.86
40.00
31.32
32.06
30.28
21.46
31.15
33.42
28.47
28.00
24.66
18.42
21.11
17.55
20.29
14.44
22.24
16.13
13.94
23.68
23.01

72
16
22
8
1
48
30
64
37
21
58
39
53
23
44
55
25
57
54
35
59

3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4

7.0
10.8
15.0
8.5
13.0
12.7
11.0
14.2
6.1
10.2
10.3
9.5
9.9
12.5
15.6
12.0
15.7
16.2
14.3
14.0
8.5

31.3
22.9
24.0
29.0
30.0
0.7
9.3
18.6
6.9
29.3
9.4
17.8
26.2
19.5
18.5
23.1
29.0
28.3
28.7
28.8
29.8

8.6
8.8
9.8
10.3
13.3
0.2
2.2
2.4
2.8
3.4
4.1
4.6
5.2
5.4
5.9
7.7
7.9
8.2
8.9
11.7
13.0

38.3
33.7
39.0
37.5
43.0
13.4
20.3
32.8
13.0
39.5
19.7
27.3
36.1
32.0
34.1
35.1
44.7
44.5
43.0
42.8
38.3

16.49
23.06
24.50
21.31
26.60
17.14
14.19
7.74
24.35
6.96
26.17
15.51
11.91
16.62
19.14
20.00
16.34
17.39
18.61
24.38
26.17

Range
Minimum
Maximum
Sum
Count
Confidence Level(95.0%)
Confidence Interval

Deliverer 1

Mean
Standard Error
Median
Mode
Standard Deviation
Sample Variance
Kurtosis
Skewness
Range
Minimum
Maximum
Sum
Count
Confidence Level(95.0%)
Confidence Interval

TravelTime

TotalTime

12.248
1.410923102
11.6
13.4
7.05461551
49.7676
0.587031829
0.892993951
28
1.9
29.9
306.2
25
2.912001562

23.524 26.60763351
1.436666048 0.901911893
22.6 27.27272727
22.6
#N/A
7.18333024 4.509559463
51.60023333 20.33612655
-0.177503025 -0.901690944
0.650202141 -0.178943293
27.1 15.66502463
12.5 18.62068966
39.6 34.28571429
588.1 665.1908378
25
25
2.96513238 1.861454275

9.34
15.16

20.56
26.49

Speed

24.75
28.47

Deliverer 2

Mean
Standard Error
Median
Mode
Standard Deviation
Sample Variance
Kurtosis
Skewness
Range
Minimum
Maximum
Sum
Count
Confidence Level(95.0%)
Confidence Interval

TravelTime

TotalTime

Speed

14.73333333
2.02059656
14.15
#N/A
8.572665177
73.49058824
-0.73233619
0.487086577
29
2.6
31.6
265.2
18
4.263092052

26.53333333
2.062947317
26.1
#N/A
8.752344224
76.60352941
-0.937011048
0.142293166
29.7
12.7
42.4
477.6
18
4.352444465

30.13609782
1.097332355
31.23380874
#N/A
4.655586896
21.67448935
0.325378021
0.153385263
18.5443038
21.4556962
40
542.4497607
18
2.31517213

10.47
19.00

22.18
30.89

27.82
32.45

Deliverer 3

Mean
Standard Error
Median
Mode
Standard Deviation
Sample Variance
Kurtosis

TravelTime

TotalTime

19.71764706
1.992342608
21.9
#N/A
8.214639013
67.48029412
-1.1813084

30.78235294
2.010875491
32.5
#N/A
8.291052051
68.74154412
-1.042563113

Speed
20.90760273
1.01274957
21.31034483
#N/A
4.175673448
17.43624874
-0.89441226

Deliverer 1
Deliverer 2
Deliverer 3
Deliverer 4

Skewness
Range
Minimum
Maximum
Sum
Count
Confidence Level(95.0%)
Confidence Interval

-0.22513287
25.3
6
31.3
335.2
17
4.223576698
15.49
23.94

-0.042181458 -0.169517395
28.3 14.06060606
15.8 13.93939394
44.1
28
523.3 355.4292464
17
17
4.262864648 2.146932695
26.52
35.05

18.76
23.05

Deliverer 4

Mean
Standard Error
Median
Mode
Standard Deviation
Sample Variance
Kurtosis
Skewness
Range
Minimum
Maximum
Sum
Count
Confidence Level(95.0%)
Confidence Interval

TravelTime

TotalTime

20.24375
2.338891631
21.3
#N/A
9.355566525
87.526625
-0.551398
-0.76072745
29.1
0.7
29.8
323.9
16
4.985232569

32.2875 17.66307043
2.66027685 1.456298681
34.6 17.26400808
#N/A
#N/A
10.6411074 5.825194724
113.2331667 33.93289357
-0.69620224 -0.367517939
-0.692576889 -0.23034761
31.7 19.21203931
13 6.962457338
44.7 26.17449664
516.6 282.6091269
16
16
5.670249368
3.10402907

15.26
25.23

26.62
37.96

Speed

14.56
20.77

CI for speed
24.75
28.47
27.82
32.45
18.76
23.05
14.56
20.77

Interpretation:
Total time is not a good measure of the efficiency of the drivers
as it includes food preparation time which is out of their control.
Travel time is not a reliable measure of driver efficiency because
it does not take into account the distance travelled by each driver.
Average speed is the best indicator as it normalises the travelling
time to distance. i.e. how many miles each driver can complete
in one hour.

sure of the efficiency of the drivers

n time which is out of their control.

easure of driver efficiency because

he distance travelled by each driver.

dicator as it normalises the travelling

ny miles each driver can complete

Order
73
26
37
48
51
9
47
70
52
49
19
17
6
3
46

Deliverer PrepTimeTravelTime
1
9.0
3.5
2
10.1
2.6
4
6.1
6.9
4
12.7
0.7
1
7.1
7.2
2
9.8
4.9
1
11.7
3.3
2
8.7
7.0
3
8.5
7.3
1
14.6
1.9
1
7.9
9.0
1
12.1
5.0
2
9.5
8.1
1
9.2
9.2
1
14.0
5.3

Distance
1.5
1.4
2.8
0.2
3.3
2.6
1.4
2.9
3.0
0.6
3.7
1.7
3.6
4.7
2.5

Total
12.5
12.7
13.0
13.4
14.3
14.7
15.0
15.7
15.8
16.5
16.9
17.1
17.6
18.4
19.3

76
58
40
18

2
4
1
1

12.0
10.3
10.3
9.9

7.5
9.4
9.5
9.9

3.3
4.1
4.3
4.5

19.5
19.7
19.8
19.8

30

11.0

9.3

2.2

20.3

24

11.0

10.5

5.0

21.5

12
4
50
65
75
28
63
61
45
13

2
3
1
3
1
1
3
3
2
1

13.0
7.2
9.5
8.8
11.0
9.2
17.0
13.1
9.7
14.9

8.8
14.7
12.6
13.6
11.6
13.4
6.0
10.1
13.8
8.7

4.1
4.3
6.5
4.6
3.6
7.0
2.8
3.1
9.2
4.2

21.8
21.9
22.1
22.4
22.6
22.6
23.0
23.2
23.5
23.6

Proportion
(p)

Deliveries within 10 mile radius (n)

63

Deliveries that took 25 mins or less

36

0.57

Deliveries that took 30 mins or less

45

0.71

Deliveries that took 35 mins or less

53

0.84

Z=

Z x p(1

95% Confidence interval of the expected vouchers amount

Delivery promise

Upper level

Lower level

Not over 25 minutes

5,533

3,039

Not over 30 minutes

3,995

1,719

Not over 35 minutes

2,508

667

3
2
2
1
1
1
3
4
2
1
2
2
1
3
4
1
3
4
3
3
1
4
4
3
4
3
3
4
4
3
4
4

6.6
17.2
10.1
12.0
13.1
13.0
8.2
9.5
18.7
14.3
10.0
10.9
10.9
18.0
12.5
9.3
8.4
14.2
9.7
10.8
17.3
15.6
12.0
13.9
9.9
7.0
15.0
10.2
14.3
14.4
16.2
15.7

17.0
6.5
14.5
12.9
11.9
12.6
19.0
17.8
8.9
13.4
18.2
17.8
17.8
10.8
19.5
22.8
24.1
18.6
23.8
22.9
16.6
18.5
23.1
21.9
26.2
31.3
24.0
29.3
28.7
29.7
28.3
29.0

6.3
3.4
7.7
4.5
4.4
7.2
7.5
4.6
5.6
5.3
9.5
6.9
8.6
3.8
5.4
8.2
5.8
2.4
6.4
8.8
9.3
5.9
7.7
8.4
5.2
8.6
9.8
3.4
8.9
6.9
8.2
7.9

23.6
23.7
24.6
24.9
25.0
25.6
27.2
27.3
27.6
27.7
28.2
28.7
28.7
28.8
32.0
32.1
32.5
32.8
33.5
33.7
33.9
34.1
35.1
35.8
36.1
38.3
39.0
39.5
43.0
44.1
44.5
44.7

Histogram

Bins
Frequency
10.00
0
15.00
7
20.00
12
25.00
17
30.00
9
35.00
8
40.00
6
45.00
4
More
0

Frequency

43
20
32
11
74
56
27
39
29
10
15
67
2
34
23
31
62
64
68
16
71
44
55
38
53
72
22
21
54
41
57
25

20
15
10
5
0

Total time

Z x p(1-p)
n

CI = p Z x p(1-p)
n

0.12

44.7%

69.6%

0.11

60.0%

82.8%

0.09

74.9%

93.3%

vouchers amount

istogram

Frequency

Order
49
47
73
17
46
51
75
19
13
40
74
18
11
3
24
10
50
28
56
31
2
71
60
14
33
26
9
70
76
20
6
12
29
67
32
45
15
69
36
7
42
5
66
63
52
61
34
4
65
62

Deliverer PrepTimeTravelTime
1
14.6
1.9
1
11.7
3.3
1
9.0
3.5
1
12.1
5.0
1
14.0
5.3
1
7.1
7.2
1
11.0
11.6
1
7.9
9.0
1
14.9
8.7
1
10.3
9.5
1
13.1
11.9
1
9.9
9.9
1
12.0
12.9
1
9.2
9.2
1
11.0
10.5
1
14.3
13.4
1
9.5
12.6
1
9.2
13.4
1
13.0
12.6
1
9.3
22.8
1
10.9
17.8
1
17.3
16.6
1
9.5
21.6
1
9.7
29.9
1
11.4
26.1
2
10.1
2.6
2
9.8
4.9
2
8.7
7.0
2
12.0
7.5
2
17.2
6.5
2
9.5
8.1
2
13.0
8.8
2
18.7
8.9
2
10.9
17.8
2
10.1
14.5
2
9.7
13.8
2
10.0
18.2
2
13.4
18.9
2
15.4
21.2
2
7.1
31.6
2
9.4
23.5
2
14.5
21.9
2
12.9
29.5
3
17.0
6.0
3
8.5
7.3
3
13.1
10.1
3
18.0
10.8
3
7.2
14.7
3
8.8
13.6
3
8.4
24.1

Distance
0.6
1.4
1.5
1.7
2.5
3.3
3.6
3.7
4.2
4.3
4.4
4.5
4.5
4.7
5.0
5.3
6.5
7.0
7.2
8.2
8.6
9.3
11.6
11.7
12.9
1.4
2.6
2.9
3.3
3.4
3.6
4.1
5.6
6.9
7.7
9.2
9.5
10.1
10.7
11.3
12.2
12.2
14.0
2.8
3.0
3.1
3.8
4.3
4.6
5.8

Total
16.5
15.0
12.5
17.1
19.3
14.3
22.6
16.9
23.6
19.8
25.0
19.8
24.9
18.4
21.5
27.7
22.1
22.6
25.6
32.1
28.7
33.9
31.1
39.6
37.5
12.7
14.7
15.7
19.5
23.7
17.6
21.8
27.6
28.7
24.6
23.5
28.2
32.3
36.6
38.7
32.9
36.4
42.4
23.0
15.8
23.2
28.8
21.9
22.4
32.5

Speed
18.95
25.45
25.71
20.40
28.30
27.50
18.62
24.67
28.97
27.16
22.18
27.27
20.93
30.65
28.57
23.73
30.95
31.34
34.29
21.58
28.99
33.61
32.22
23.48
29.66
32.31
31.84
24.86
26.40
31.38
26.67
27.95
37.75
23.26
31.86
40.00
31.32
32.06
30.28
21.46
31.15
33.42
28.47
28.00
24.66
18.42
21.11
17.55
20.29
14.44

Deliverers can be compared by c

a significant difference between
details continue reading...

However, we can also answer the

between each pair of deliverers at
squared) of the two underlying pop
Anlysis

Excel gives us the outcome for bot

faster/slower than deliverer 2"). Ho

43
68
41
27
38
72
16
22
8
1
48
30
64
37
21
58
39
53
23
44
55
25
57
54
35
59

3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4

6.6
9.7
14.4
8.2
13.9
7.0
10.8
15.0
8.5
13.0
12.7
11.0
14.2
6.1
10.2
10.3
9.5
9.9
12.5
15.6
12.0
15.7
16.2
14.3
14.0
8.5

17.0
23.8
29.7
19.0
21.9
31.3
22.9
24.0
29.0
30.0
0.7
9.3
18.6
6.9
29.3
9.4
17.8
26.2
19.5
18.5
23.1
29.0
28.3
28.7
28.8
29.8

6.3
6.4
6.9
7.5
8.4
8.6
8.8
9.8
10.3
13.3
0.2
2.2
2.4
2.8
3.4
4.1
4.6
5.2
5.4
5.9
7.7
7.9
8.2
8.9
11.7
13.0

23.6
33.5
44.1
27.2
35.8
38.3
33.7
39.0
37.5
43.0
13.4
20.3
32.8
13.0
39.5
19.7
27.3
36.1
32.0
34.1
35.1
44.7
44.5
43.0
42.8
38.3

22.24
16.13
13.94
23.68
23.01
16.49
23.06
24.50
21.31
26.60
17.14
14.19
7.74
24.35
6.96
26.17
15.51
11.91
16.62
19.14
20.00
16.34
17.39
18.61
24.38
26.17

t Critical two-tail

Comparison of Deliverer 2 and 4

t-Test: Two-Sample Assuming Une

Mean
Variance
Observations
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Comparison of Deliverer 3 and 4

t-Test: Two-Sample Assuming Une

Mean
Variance
Observations
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Comparison of Deliverers

Deliverers can be compared by checking if the confidence intervals for their speed overlap or not. If the intervals over
a significant difference between the two deliverers. If they don't overlap we conclude that there is a significant differen
details continue reading...

However, we can also answer the same question is a more accurate and sophisticasted way. Below is the outcome and interpr
between each pair of deliverers at the Red Dragon restaurant. Two types of tests are performed. In the first one we assume th
squared) of the two underlying populations are unequal and in the second one we assume the variances are equal. Both tests
Anlysis
Excel gives us the outcome for both a two tailed hypothesis test ("Has deliverer 1 the same average speed as deliverer 2 or n
faster/slower than deliverer 2"). However, we need to carefully interprete the output from Excel. See interpretation for the

Comparison of Deliverer 1 and 2

We could be only interested in checking if there is a significant differ

the following two-tailed hypothesis test with a = 0.05:

t-Test: Two-Sample Assuming Unequal Variances

Mean
Variance
Observations
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Deliverer 1 Deliverer 2
26.60763 30.1361
20.33613 21.67449
25
18
0
36
-2.48411
0.008888
1.688297
0.017777
2.028091

H0: mdel1 - mdel2 = 0

HA: mdel1 - mdel2 not equal to 0

The "t Stat" tells us how many standard errors the sample mean diff
the sample mean difference is 2.48 standard errors to the left of 0. In
mean difference needs to be more than 2.03 standard errors ("t Criti
that the mean speed of deliverer 1 is the same as deliverer 2 at 5%
p-values. "P(T<=t) two-tail" tells us the p-value for the two
below 0.05, our significance level, we reject the null hypothesis.
It is probably more reasonable to check if deliverer 2 is significantly
what we would like to prove in the alternative hypothesis):
H0: mdel1 - mdel2 >= 0
HA: mdel1 - mdel2 < 0

Comparison of Deliverer 1 and 3

t-Test: Two-Sample Assuming Unequal Variances

Mean
Variance
Observations
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Deliverer 1 Deliverer 3
26.60763 20.9076
20.33613 17.43625
25
17
0
36
4.203139
8.3E-05
1.688297
0.000166
2.028091

Intuitively we would reject H0 if the sample mean difference is many

1). Here the sample mean difference is 2.48 standard errors to the le
standard errors to the left of 0. Furthermore, the p-value is 0.009, wh
be quite convinced that deliverer 2 is faster.

In general: The more standard errors the "t Stat" is away from 0 (the
we are about rejecting the null hypothesis.
Note: Excel doesn't tell us where the rejection area is in the one
we are far away, in either direction, from the hypothesized mean we
which tail the rejection region is. In the test above the rejection regio
H0). Hence, we reject H0 since "t Stat" = - 2.48, is further away to the

Deliverer 1 is significantly faster than deliverer 3 at the 5% significa

Comparison of Deliverer 1 and 4

t-Test: Two-Sample Assuming Unequal Variances

Mean
Variance
Observations
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Deliverer 1 Deliverer 4
26.60763 17.66307
20.33613 33.93289
25
16
0
26
5.221683
9.37E-06
1.705616
1.87E-05
2.055531

Deliverer 1 is significantly faster than deliverer 4 at the 5% significa

Comparison of Deliverer 2 and 3

t-Test: Two-Sample Assuming Unequal Variances
Deliverer 2 Deliverer 3
Mean
30.1361 20.9076
Variance
21.67449 17.43625
Observations
18
17
Hypothesized Mean Difference
0
df
33
t Stat
6.180135
P(T<=t) one-tail
2.84E-07
t Critical one-tail
1.69236
P(T<=t) two-tail
5.68E-07
t Critical two-tail
2.034517

Deliverer 2 is significantly faster than deliverer 3 at the 5% significa

Comparison of Deliverer 2 and 4

t-Test: Two-Sample Assuming Unequal Variances

Mean
Variance
Observations
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Deliverer 2 Deliverer 4
30.1361 17.66307
21.67449 33.93289
18
16
0
29
6.840372
8.18E-08
1.699127
1.64E-07
2.045231

Comparison of Deliverer 3 and 4

Deliverer 2 is significantly faster than deliverer 4 at the 5% significan

t-Test: Two-Sample Assuming Unequal Variances

Mean
Variance
Observations
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Deliverer 3 Deliverer 4
20.9076 17.66307
17.43625 33.93289
17
16
0
27
1.829113
0.039224
1.703288
0.078449
2.051829

Deliverer 3 is significantly faster than deliverer 4 at the 5% significanc

However, we are not very convinced since the p value is not very low
Note that we cannot reject the two-tailed test at the 5% level. Howeve
do that at about 8% level. This might seem contradicting, but the thing
in mind is that since the p-value is not very low in both cases we are
border of accepting and rejecting both the two-tail and one

r not. If the intervals overlap we conclude that there is not

re is a significant difference. If you are interested in more

is the outcome and interpretation of the hypothesis testing

he first one we assume the variances (standard deviation
nces are equal. Both tests can be found in Data...Data

speed as deliverer 2 or not") and one tailed test ("Is deliverer 1

interpretation for the first test below.

there is a significant difference between the mean speed of deliverer 1 and deliverer 2. We set up
= 0.05:

rrors the sample mean difference is from the hypothesized mean difference (here 0). We can see that
ard errors to the left of 0. In order to reject the null hypothesis in the two tailed test above, the sample
.03 standard errors ("t Critical two-tail") away from the mean. Hence, we reject the null hypothesis
same as deliverer 2 at 5% significance level. We can come to the same conclusion by considering the
value for the two-tail test (the area in the tails beyond +/-2.48 standard errors from 0). Since 0.018 is
ct the null hypothesis.

deliverer 2 is significantly faster that 1. We set up the following one-tailed hypothesis test (we put
ive hypothesis):

e mean difference is many standard errors to the left of 0 (then we would believe del2 is faster than
48 standard errors to the left of 0. In order to reject H 0 we need to be 1.69 ("t Critical one-tail")
value is 0.009, which is quite low and lower than 0.05, our significance level. Hence, we can

"t Stat" is away from 0 (the hypothesized value) and the smaller the p value is, the more convinced

ction area is in the one-tail test. In the two-tail test it is trivial since the rejection area is in both tails (if
he hypothesized mean we reject). In the one-tail test we need to determine from the test we set up in
st above the rejection region is the left tail (if the mean of del1 is much smaller then of del2 we reject
2.48, is further away to the left of 0 that "t Critical one-tail" = 1.69.

liverer 3 at the 5% significance level.

iverer 4 at the 5% significance level.

liverer 3 at the 5% significance level.

verer 4 at the 5% significance level.

erer 4 at the 5% significance level.

the p value is not very low, about 4%.
st at the 5% level. However, we can
contradicting, but the thing to keep
low in both cases we are on the
tail and one-tail hypothesis tests.

If we are certain that the variances of the

two populations are equal we can use the
test below. This test is more powerful,
meaning, that we can reject with a lower
p-value (we are further away from the null
hypothesis). However, if we are not certain
that the variances are the same this test
does not apply.

Comparison of Deliverer 1 and 2

t-Test: Two-Sample Assuming Equal Variances

Mean
Variance
Observations
Pooled Variance
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Deliverer 1 Deliverer 2
26.60763 30.1361
20.33613 21.67449
25
18
20.89106
0
41
-2.49734
0.00831
1.682879
0.01662
2.019542

Comparison of Deliverer 1 and 3

t-Test: Two-Sample Assuming Equal Variances

Mean
Variance
Observations
Pooled Variance
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Deliverer 1 Deliverer 3
26.60763 20.9076
20.33613 17.43625
25
17
19.17618
0
40
4.140626
8.68E-05
1.683852
0.000174
2.021075

Comparison of Deliverer 1 and 4

t-Test: Two-Sample Assuming Equal Variances

Mean
Variance
Observations
Pooled Variance
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Deliverer 1 Deliverer 4
26.60763 17.66307
20.33613 33.93289
25
16
25.56565
0
39
5.525464
1.18E-06
1.684875
2.36E-06
2.022689

Comparison of Deliverer 2 and 3

t-Test: Two-Sample Assuming Equal Variances

Mean
Variance
Observations
Pooled Variance
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Deliverer 2 Deliverer 3
30.1361 20.9076
21.67449 17.43625
18
17
19.61958
0
33
6.160453
3.01E-07
1.69236
6.02E-07
2.034517

Comparison of Deliverer 2 and 4

t-Test: Two-Sample Assuming Equal Variances

Mean
Variance
Observations
Pooled Variance
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Deliverer 2 Deliverer 4
30.1361 17.66307
21.67449 33.93289
18
16
27.42062
0
32
6.932503
3.76E-08
1.693888
7.52E-08
2.036932

Comparison of Deliverer 3 and 4

t-Test: Two-Sample Assuming Equal Variances

Mean
Variance
Observations
Pooled Variance
Hypothesized Mean Difference
df
t Stat
P(T<=t) one-tail
t Critical one-tail
P(T<=t) two-tail
t Critical two-tail

Deliverer 3 Deliverer 4
20.9076 17.66307
17.43625 33.93289
17
16
25.4185
0
31
1.847586
0.037112
1.695519
0.074225
2.039515

An example of a recommendation

ample of a recommendation

Red Dragon - Recommendations to Management

Efficiency of Delivery People
(i) Total Time and (ii) Travel Time are not fair measures for the efficiency of the delivery
people because (i) includes food preparation time while (ii) does not take into account the
different route lengths. From the Mean Speed of Delivery we can infer that there are
significant differences in efficiency between delivery people. The significance tests
performed for each pair of delivery people provide further evidence in favour of this
argument. Deliverers 3 and 4 appear to be significantly slower than Deliverers 1 and 2.
(iii) Management could use these statistical tools to measure performance of delivery people
and set incentives and take appropriate action. One way to reduce efficiency gaps is to share
best practice between delivery people. Further, Management could invest in the training of
new hires, i.e. familiarity with destinations within the 10 mile radius. To increase overall
efficiency, Management should develop a tool for the planning of delivery routes, so that one
deliverer can take care of multiple orders for destinations close to each other.
Certificates
From analysing the certificate plans for the different delivery times, we can infer that more
than 50% and more than 30% of deliveries would be late in the case of the 25 minutes plan
and the 30 minutes plan respectively. Hence, assuming an average purchase value of 10 per
delivery, on average these plans would cost more than 50% and 30% of revenues
respectively. Therefore, Management should refrain from introducing these plans. The risk
associated with the 35 minutes plan is limited. We could state with 95% confidence that the
35 minutes plan would cost between 5% and 22% of revenues on average. Management could
further reduce this risk by measuring and improving the performance of the delivery people
as described above. As eager delivery service customers, we also feel that 35 minutes is a
Notes on the Data

1) The data samples for each deliverer are relatively small. To enable a more reliable
approximation to the Normal Distribution it would be helpful to have a larger sample size
approximately 30 data points for each deliverer.

2) The data does not include other important information such as rush hours. For example,
3) In the sample for Deliverer 4, two extraordinarily slow deliveries were spotted. We may
need to investigate the causes for these data points and exclude them from the analyses to
obtain more reliable results.