Stats Lab 4- Mattea Krug

1. a)

Confidence Significance Margin of Failed Failure

Interval Level error Hypothesis Frequency
0.9 0.1 0.480492 10 0.1
0.95 0.05 0.572542 6 0.06
0.99 0.01 0.7524479 1 0.01

The relationship between confidence interval and margin of error is that the wider the
Confidence Interval is, the larger the margin of error is. This is because the confidence interval
takes a wider and wider range of samples as it increases, meaning it includes values that are
further and further from the mean, therefore increasing the margin of error

b) If we compare the failure frequency, or more commonly known as significance level, we can
see that the theoretical and experimental values are very close or exactly the same, with only
the confidence level of 0.95 deviating by 0.01.
2. a)

# of
Significance Rejected Frequency
level Samples of Rejection
0.1 10 0.1
0.05 6 0.06
0.01 1 0.01

The relationship between the reduction of significance level and the number of rejected samples
is that as the significance level decreases the number of rejected samples decreases. By
comparing the frequency and the significance level we can see that they are relatively close or
the same showing that the theoretical and experimental values agree. This is similar to question
one in that as the significance level decrease the number of samples that can reject is decreased
as well. Decreasing the sample size would decrease the number of rejections

b) If we compare the failure frequency from question 1 and the Frequency of Rejection from
question two, we can see that they are the exact same value, indicating that there is a direct
relationship between two sided test and this two sided confidence interval.
We can see that the test at alpha=0.1 with 90% confidence interval values are exactly the same
We can see that the test at alpha=0.01 with 99% confidence interval values are exactly the same
However the test at alpha=0.05 with 95% confidence interval values are the same but differ
from the theory.

This makes sense as confidence level and significance level are directly related. Using the t-
distribution would not affect this conclusion because it directly uses the alpha value, or the
significance level to calculate the probabilities of the sample.
3. a)

Ornithopter Current

Mean 71.324567 Mean 70.95203

Standard Error 0.1849598 Standard Error 0.3907
Median 71.094 Median 70.6565
Mode 70.875 Mode 69.781
Standard Deviation 1.0130667 Standard Deviation 2.139954
Sample Variance 1.0263042 Sample Variance 4.579404
Kurtosis 10.0444 Kurtosis -0.11005
Skewness 2.9745033 Skewness 0.561959
Range 5.141 Range 8.103
Minimum 70.328 Minimum 67.484
Maximum 75.469 Maximum 75.587
Sum 2139.737 Sum 2128.561
Count 30 Count 30
Confidence Confidence
Level(99.0%) 0.5098207 Level(99.0%) 1.076921

Lower 70.814746 69.87511

Upper 71.834387 72.02895

If we look at the upper and lower bounds of the 99% confidence interval we can see that the
ornithopter has a confidence interval is highly above 70 m/s meaning that with 99% confidence
the mean speed of the ornithopter is between 70.8 and 71.8 meters per second. This is
completely above the minimum necessary speed of 70 m/s and there is sufficient evidence to
suggest that the mean value is above 70 m/s.
In comparison with the current helicopter, there is insufficient evidence to suggest that the
mean value of the current helicopter is above 70 m/s as the minimum mean value falls within
the interval.
Tis means that the ornithopter is the better helicopter as it has a higher mean speed.

b) The mean speed of the ornithopter has sufficient evidence that the mean value is above 70
m/s. However, there is insufficient evidence to suggest that the mean value of the current
helicopter is above 70 m/s as the minimum mean value falls within the interval.
4. a)

H0: mu =70 Ha: mu >70

Alpha 0.005 0.005

T-stats 7.1613747 2.436735
p-value 3.493E-08 0.010596

Using the significance level approach, we can se that the p-value is much smaller than alpha,
than we reject H0 as there is sufficient evidence to reject for the ornithopter, meaning that
there is significant evidence to assume that the mean speed is higher than 70.
We can also see that for the current helicopter p-value is larger than alpha meaning that we do
not reject Ho. Therefore, there is significant evidence to not reject Ho for the current helicopter.

b) Assumptions: 1) Random Sample? The given data specifically says that the sample is a random
sample, therefore this assumption is satisfied. 2) independent observations? Because these
samples are random, one helicopter does not directly affect another, however if there is an
issue with the production process, we could assume that the overall mean would be affected. 3)
n >= 30? Yes, each sample is 30 therefore this assumption is satisfied, however the value is
exactly 30 meaning that it is just on the “edge” of rejection, that could cause issues. Therefore,
all assumptions hold.
5. a)

Ornithopter Current
Mean 71.32456667 70.95203333
Variance 1.026304185 4.579404033
Observations 30 30
Pooled Variance 2.802854109
Hypothesized Mean Difference 0
df 58
t Stat 0.861808
P(T<=t) one-tail 0.19617117
t Critical one-tail 2.392377475
P(T<=t) two-tail 0.392342339
t Critical two-tail 2.663286954

Ho= mu1-mu2 Ha= mu1-

=0 mu2 /= 0

test statistic 0.861808

p-value 0.3923 two-tailed

b) Because this is a paired test the following assumptions must hold

1. samples are random and independent.

It is explicitly stated that the helicopters are randomly selected, and are therefore independent,
this assumption holds.

2. Both populations are normal

Both samples are large (n>=30) therefore we can assume normality of the samples, however
upon inspection we can see that the sample size is exactly 30 and could cause issues by
assuming total normality. However, numerically this assumption holds

3.Both population variances are known and Smax/Smin <2

Using the descriptive statistics, we can see that the variances of the current is √4.579/√1.026 =
2.112

This is not smaller than 2 therefore this does not hold and using the paired statistics will not be
valid

Using the significance level approach we can see that the p-value is significantly smaller than
alpha this gives us strong to convincing evidence against Ho, this means that there is strong
evidence to conclude that the mean speed increases after redesign.
6.a)
t-Test: Paired Two Sample for
Means

Emergency Current
Mean 73.08806667 70.95203333
Variance 0.42123372 4.579404033
Observations 30 30
Pearson Correlation 0.595881944
Hypothesized Mean Difference 0
df 29
t Stat 6.396517736
P(T<=t) one-tail 2.69975E-07
t Critical one-tail 1.699127027
P(T<=t) two-tail 5.39949E-07
t Critical two-tail 2.045229642

H0: mu_D = 0 Ha: mu_D>0

t-stats 6.396517736
tdist with df of
29
P-value 2.69975E-07 one tailed
a) the p-value is much smaller than alpha of 0.01, this means that there is little to no evidence
to reject Ho, indicating that there is strong evidence to suggest that the mean speed
increase after the redesign.

Mean 2.136033333
Standard Error 0.333936905
Median 2.2425
Mode 3.172
Standard Deviation 1.829047755
Sample Variance 3.345415689
-
Kurtosis 0.205948836
-
Skewness 0.229239313
Range 7.557
Minimum -1.978
Maximum 5.579
Sum 64.081
Count 30
Confidence
Level(98.0%) 0.822159792

lower 1.802096429
upper inf
(1.802,inf)
b) this interval agrees with part a) because the interval does not include the value of zero, and
sits entirely above zero, indicating that there is strong evidence to conclude that the mean
increases after the redesign, this agrees with part a)
c) 1.Random sample? This is explicitly stated in the question statement, this assumption holds
2. independent observations? This is stated in the question statement, this assumption
holds
3. Normality? n=30>=30, this assumption holds
Because all of these assumptions hold this test is valid

Change Vs Current
7
6
5
4
Change (m/s)

3
2
1
0
-1
-2
-3
66 68 70 72 74 76

d) Mean Speed (m/s)

We can see that the change in speed of the redesign decreases as the mean speed of each
sample increases. This means that these values are not independent, we can see that there
is a dependency of the two parameters. This allows us to use the paired distribution tests.