PROBLEM SET 4
Chapter 14
Individual Work
1. Do problem 14.21 (.5 point).
Common causes of variation are variability in a process that is anticipated while special causes of
variation are unusual patterns in the process.
2. Do problem 14.25 (.5 point).
Attribute control charts are used for categorical or discrete variables while the variables control
charts are used when you have numerical data such as time, money and weight.
3. Do problem 14.26 (.5 point).
You must ensure that the process is in control using the R chart and then analyzing the X bar
chart for the process mean. If the process is not in control then the X bar chart will be misleading.
Individual or Group Work
4. Do problem 14.30 (a, b, c, and additional d,e,f below) (2.0 points). Data is in PS 3
RudyBird.
p Chart Summary
0.4
0.35
0.3
0.25
p
Proportion
LCL
0.2
Center
0.15 UCL
0.1
0.05
0
0 5 10 15 20 25 30 35
X
Before the promotion, the process is in control. There are no points above or below the UCL or
LCL. Also none of the points form a pattern in which 8 or more of them are in a row on either side
of the mean or moving in the same direction. Also the amount of points above and below the
mean does not have a difference of 8 or more.
p Chart Summary
0.6
0.5
0.4
p
Proportion
LCL
0.3
Center
UCL
0.2
0.1
0
0 5 10 15 20 25 30 35 40
X
This chart is not in control since there are 6 points that are above or below the UCL or LCL. There
is also over 8 points towards the middle of the chart that are in a row which indicates that the
process is not in control. However even though the chart is not in control, the promotion seems to
be working at increasing sales of the RudyBird disks.
(d) Explain why it is necessary to generate two p-charts and compare rather than just
one p-chart for the 37 days.
It is necessary to generate two p charts so that you are able to see the effect the
promotion had on the sales. If I were to only look at the pattern from the chart with
days 1-37, I would assume that the process was out of control anyways and that the
promotion did not have the desired effect but by being able to compare the two, I was
able to notice that the promotion did in fact work at increasing sales but the additional
caused the chart to seem like it was out of control the entire time.
(e) Explain also why it would be insufficient to compare the chart with days 1-30 with
a chart for days 31-37.
You wouldn't have been able to see the previous sales data and promotion sales in
their entirety if you had split the charts into days 1-30 and days 31-37. By analyzing
the chart prior to the promotion you were able to see it as a sort of control (pun
intended) and could clearly see how the promotion impacted the sales altogether.
(f) In this particular problem, is the process being out-of-control undesirable or
indicates a problem that needs to be corrected? Explain!
In this particular problem, the process was only out of control after the promotion but
the promotion did have a good impact on sales by increasing them so this out of
control process was expected. However if it was determined that the sales prior to the
promotion were out of control, it would have been wise to correct the problem by
figuring out what days in particular were out of control and then determining factors
that would affect the process.
5. Do problem 14.31 (a and b) (1.5 points). Data is in PS 3 Sealant.
a) You will create the R and X-bar charts at the same time using PHStat. Just be careful to cooy
the R chart here!
Boston
0.4
0.35
0.3
0.25 Range
LCL-R
Sample Range
Center-R
0.2
UCL-R
0.15
0.1
0.05
0
0 5 10 15 20 25 30
Vermont
0.35
0.3
0.25
Range
LCL-R
Sample Range
0.2
Center-R
UCL-R
0.15
0.1
0.05
0
0 2 4 6 8 10 12 14 16 18 20
b) Copy the X-bar chart here.
Boston
1.6
1.4
1.2
1 XBar
LCL-X
Sample Mean
Center-X
0.8
UCL-X
0.6
0.4
0.2
0
0 5 10 15 20 25 30
Vermont
1.6
1.4
1.2
1 XBar
LCL-X
Sample Mean
Center-X
0.8
UCL-X
0.6
0.4
0.2
0
0 2 4 6 8 10 12 14 16 18 20
All of the charts seem to be in control except the x bar chart for Boston where one point is above
the UCL and one point is below the LCL. Otherwise it doesn't seem like any of the charts have
any out of control processes. For both the Boston and Vermont x bar chart, it seems as though it
is trending upwards and improving towards the middle of the chart so we should take note of
these improvements and apply them to the other days. They also seem to dip lower towards far
left and right of all the charts for both Boston and Vermont so we should check to see what could
possibly be causing the decline and work to implement the improvements that we found caused
other days to trend upwards.
Chapter 8: Individual
1. This problem is based on the file Absent PS4 S14 on Laulima. (1.5 points)
The data represents a sample of 100 organizations and gives the Wages for each organization.
Unstack Wage by Union. Use the data in the Unstacked sheet.
Construct 95% Confidence intervals for the population mean for Wages for each of these 3 union
levels and write a one sentence interpretation for each of the 3 confidence intervals you obtained.
Also for each interval assess whether the assumption that the data comes from a Normal shaped
population is acceptable for each of the 3 sets of data. Then comment on whether the 3
confidence intervals should be valid or not using the fact that they should be valid whenever the
data is normal or the sample size is large (n > 30)
Low
Low
Data Low
Sample Standard Deviation 10238.6202 Mean 37087.03448
Sample Mean 37087.03448 Median 35026.5
Sample Size 29 Mode #N/A
Confidence Level 95% Minimum 20404.5
Maximum 64479
Intermediate Calculations Range 44074.5
Standard Error of the Mean 1901.26404 Variance 104829343.5702
Degrees of Freedom 28 Standard Deviation 10238.6202
t Value 2.0484 Coeff. of Variation 27.61%
Interval Half Width 3894.5628 Skewness 0.8155
Kurtosis 0.6059
Count 29
Confidence Interval
Standard Error 1901.2640
Interval Lower Limit 33192.47 We can be 95% confident that population
Interval Upper Limit 40981.60 mean of wages for Low union companies
falls between $33,192.47 and $40,981.60.
The confidence interval estimate is valid since the skewness is 0.8155 and the kurtosis is 0.6059,
which is between -1 and 1. However according to the central limit theory, since there are only 29
samples not 30, the validity is not certain.
Medium Medium
Data
Sample Standard Deviation 10385.19016Mean 34605.60526
Sample Mean 34605.60526Median 33487.5
Sample Size 57Mode #N/A
Confidence Level 95%Minimum 18034.5
Maximum 71513.5
Intermediate Calculations Range 53479
Standard Error of the Mean 1375.55204 Variance 107852174.5869
Standard
Degrees of Freedom 56 Deviation 10385.1902
t Value 2.0032 Coeff. of Variation 30.01%
Interval Half Width 2755.5619 Skewness 1.2678
Kurtosis 2.4795
Confidence Interval Count 57
Interval Lower Limit 31850.04 Standard Error 1375.5520
Interval Upper Limit 37361.17
We can be 95% confident that population mean of wages for Medium union companies falls
between $31,850.04 and $37,361.17. The confidence interval estimate is not normal since even
though the sample size is 57 which is over 30, the skewness is 1.2678 and the kurtosis is 2.4785
which is not between -1 and 1 so it is not normally distributed.
High
High
Data
High
Sample Standard Deviation 12982.63228
Mean 37152.53571
Sample Mean 37152.53571
We Median 33231.75
Sample Size 14 can Mode #N/A
Confidence Level 95% be
Minimum 22617
Intermediate Calculations Maximum 61792.5
Standard Error of the Mean 3469.754425 Range 39175.5
Degrees of Freedom 13 Variance 168548740.8255
t Value 2.1604 Standard Deviation 12982.6323
Interval Half Width 7495.9487 Coeff. of Variation 34.94%
Skewness 0.9284
Confidence Interval Kurtosis -0.2483
Interval Lower Limit 29656.59 Count 14
Interval Upper Limit 44648.48 Standard Error 3469.7544
95% confident that population mean of wages for Low union companies falls between $29,656.59
and $44,648.48. The confidence interval estimate is not valid because the data only has 14
samples, which is less than 30 so according to the central limit theory, the data is not normal. But
since the skewness is 0.9284 and the kurtosis is -0.2483, which is between -1 and 1, the data is
actually normally distributed.
2. Scott Gerber of Young Entrepreneur Council teamed up with Buzz Marketing Group (and
sponsor [Link]) to put together a survey of 1,000+ college students and recent
graduates. [Link]
They asked questions about entrepreneurship, the economy and business aspirations.
Here are the major findings:
36% are "sidepreneurs," i.e. they've started businesses in addition to getting their
degrees
21% Started businesses because they are unemployed.
69% would like to work for an entrepreneur
89% feel entrepreneurship education is important given the new economy and job
market
73% were not offered classes in entrepreneurship
70% who took entrepreneurship classes said they were not adequate
From the survey results, we can gather that the demand for entrepreneurship is high, and our
education system isn't adapting quickly enough.
The survey was based on 1635 respondents.
Answer the following questions based only on the first finding (36% figure) above (2.0 points).
a. Write a one sentence interpretation of your 95% confidence interval.
We can be 95% confident that between the percentage of students surveryed that are
sidepreneurers or have started businesses in addition to getting their degrees is between 33.7%
and 38.4%.
b. Calculate both a 95% confidence interval and a 90% confidence interval. See
directly below for how to do this. The 90% CI is the same except to change the
95 to 90. What is the difference in the width (upper – lower limits) of the 2
confidence intervals? Which confidence interval gives you a more precise
estimate (smaller width)?
We can be 95% confident that between the percentage of students surveryed that are
sidepreneurers or have started businesses in addition to getting their degrees is between 33.7%
and 38.35%. The difference is 4.65%.
We can be 90% confident that between the percentage of students surveryed that are
sidepreneurers or have started businesses in addition to getting their degrees is between 34.07%
and 37.98%. The difference is 3.91%.
The 95% confidence interval gives us a more accurate estimate since the interval is larger
however the 90% confidence interval gives us a more precise interval but since it is more precise,
the interval is narrower since your certainty is lower.
Question 1
Data
Sample Size 1635
Number of Successes 589
Confidence Level 95%
Intermediate Calculations
Sample Proportion 0.360244648
Z Value -1.9600
Standard Error of the Proportion 0.0119
Interval Half Width 0.0233
Confidence Interval
Interval Lower Limit 0.3370
Interval Upper Limit 0.3835
Question 2
Data
Sample Size 1635
Number of Successes 589
Confidence Level 90%
Intermediate Calculations
Sample Proportion 0.360244648
Z Value -1.6449
Standard Error of the Proportion 0.0119
Interval Half Width 0.0195
Confidence Interval
Interval Lower Limit 0.3407
Interval Upper Limit 0.3798
Excel example results are also available on Laulima.
3. Problem 8.50 on P.283 (.5 point)
You can never really have 100% confidence when estimating the population
characteristic of interest because it uses normal distribution and the only numbers that
you can be 100% confident that your characteristic falls between is plus or minus infinity
and then its no longer an estimate.
4. Problem 8.52 on P.283 (.5 point)
An increase in confidence is achieved by widening the confidence interval and making
less precise because we widen it to increase the number of possible values that are
plausible. But since there is only 1 exact value for the population characteristic, wddening
the range of possibilities makes the estimate less precise.
5. Problem 8.53 on P.283 (.5 point)
According to the formula for estimating the sample size needed for estimating a
population proportion contains a part where the proportion is multiplied by (1-
poprportion). So if we were to solve it for 20% and 50% we get 0.2*(1-0.2) = (0.2)(0.8) =
0.16 and 0.5*(1-0.5) = (0.5)(0.5) = 0.25. So since 0.25 is larger than 0.16 the sample size
needed to determine the population proportion is smaller when it is 20% as given in the
formula.