FREE ESSAY ON BUSINESS AND APPLIED STATISTICS RESEARCH

BUSINESS AND APPLIED STATISTICS RESEARCH

Table of Contents:
Executive Summary
1
Problem/Opportunity
2-4
Objectives/Goals
5
Research Design
5-6
Data Collection
6-9
Data Analysis
9-10
Data Interpretation
10-11
Appendix:
Appendix A(College Pro Overview)
12
Figure A(Questionnaire-rough draft)
13
Crosstabulation Table
14
Appendix B(Calculations/Equations)
15-16
Appendix C(Manager Success Model)
17
Figure B(Graphs)
18
References
19
Resume
20
Keeping Quality Under Control
A Proposal 
For 
xxxxxxxxxx
1999
Executive Summary
I am coming to xxxxxxxxx with a research proposal that could quite possibly revitalize
the spirit in which it was founded. Since xxxxxx first began in 1978, quality workmanship
has been one of the principles around which its foundation was constructed. It is that
very same principle that established a clientele which has supported xxx over the years
with repeat business and word of mouth recommendations. 
After having the privilege of working for xxx this past summer I was able to gain some
insight on a problem that could potentially crack that foundation. After years of high
quality performance it appears to me that, like many other opportunistic companies, xxx
has let targets and numbers get in the way of the thing that made them the largest
residential painting company in the world. That thing is quality.
It is my assertion that the production target driven structure which xxx has grown to
embrace is the culprit in an ongoing decrease in overall quality. If given the chance I
am capable of doing pertinent business research that can determine the existence or lack
there of of an association between production targets and quality ratings.
This research will not only allow management to understand the connection between these
targets and quality but also enable them to pick any desired level of overall quality by
choosing the corresponding production target.
Such understanding will inherently lead to and increase in overall quality. However,
there are countless indirect results of such an understanding. To name a few: increased
profits, increased demand, lowered stress, less turnover and on and on. 
The majority of this proposal details specifically how I would go about obtaining and
analyzing this data as well as what it could do for you. Thank you for your time and
consideration.
NOTE: If the reader is not familiar with xxx please read the brief company description in
Appendix A. 
Problem/Opportunity Description:
Overview
xxxx has three main principles by which it stands. These principles are intended to guide
administrators in the decision making process and lead managers on a daily basis as they
work with customers and painters. As I understand it, these three statements are the
commandments of the xxx business. Do as they say and you can do no wrong. 
In everything you do quality should be the motivating factor.
This is one of the
three principles and today this commandment is under scrutiny. Recently I had the
privilege of working for The Franchise Company(founder of xxx), so I have some insight as
to how the company operates. One of the things that I noticed was a discrepancy between
quality and production targets(aka Designated Target DT). 
Larger production targets do not inherently imply that quality will go down. If an
emphasis is placed on quality above all things then a well run business can maintain the
same level of quality no matter what the DT. However, I believe that one can find some
drastic inconsistencies between what xxx preaches and what xxx teaches. 
From personal experience, I can attest to the fact that hitting your DT is the number one
priority. As I would expect, xxx administration maintains that quality(one of the
founding principles) is their number on priority no matter what the DT. The problem here
is, through no fault of their own, xxx is unaware of the impact that higher DTs have on
young managers interest in quality.
In an insert from a xxx field manual called "Manager Success Model," the contradiction is
blatantly obvious. On one side of the sheet points are given for quality, profit and
volume.(see Appendix C) The points awarded at the "star" level are twice as much(twice as
significant) as both profit and volume which are equal in points. However, on the other
side of the sheet the exact wording is "xxx and your General Manager will deem your
summer a success provided that you hit your sales target, make between $6,000 and $10,000
and deliver quality service to your customers..." In my opinion the order of the latter
is more indicative of the true culture. 
Quality Contradiction
There are three factors that undermine the quality commandment. 
1) When a manager hits his DT he pays less royalty on each job produced thereafter. 
2) The general managers who set the DTs for the managers receive a bonus when each
manager hits the designated DT. 
3) The company as a whole will naturally make more money as more DTs are met. 
It is this structure that undermines the culture and specifically the quality that xxx's
three principles try and instill. It is not necessarily intentional but rather
unavoidable.
Benefit of Quality
In my opinion, xxx has the right idea in trying to emphasize quality. Quality is the
cornerstone of the most powerful means of getting more contracts and that is word of
mouth. It is the number one goal of most companies to provide a service or product at
such a level that new business is generated by word of mouth.
According to Gitlow the benefits of improving quality are similar. He puts them in this
order:
"1) Productivity rises
2) Quality Improves
4) Price can be cut
5) Cost per good unit or service is lowered
6) Worker morale goes up"(Gitlow, 1990) 
Downfall of Low Quality
Poor quality of work damages the business more than large DTs help it. It is hard to
calculate the effects of low quality on a business because the repercussions are
intangible. 
Here are three of the main outcomes from a low level of quality. For one, there are an
inflated number of complaints that have to be dealt with on a daily basis by employees
whose sole job it is to listen to complaints. These are employees that could be used
elsewhere or not at all. 
Secondly, every year there are several huge lawsuits that are the direct result of
neglect. The majority of these settlement fees and lawyer fees could be avoided with a
consistent emphasis on quality. 
Finally, one of the toughest things for xxx to do year in and year out is bring in
qualified managers and train them. This requires a huge amount of resources. Quality is
one of the main factors that ultimately drives managers away which leads to xxx having to
rehire and retrain. When quality is high there are fewer job-site problems and an overall
lowered level of stress for the manager. This means a happy manager at the end of the
season and usually a returning manager (CPPs most valuable asset). 
Opportunity
By continuing to emphasize quality and restructuring certain aspects of the DT program,
xxx may initially show a drop in revenue. On the other hand, the ultimate benefits of
such an adjustment bode increased profits, smoother operation, and long term stability.
As Brown puts it "The goal of research is not to measure past performance, but to drive
future behavior."(Brown, 1991)
With the right data and the proper research I feel that I can provide xxx with the
information necessary to create a marriage between production targets(DTs) and the
quality of the work. Such a marriage would allow for a maximum level of production at a
desired level of quality.
Research Objectives/Goals
Hypothesis: An increase in production targets will lead to an overall decrease in quality
ratings.
If you contract with me:
1) I will attempt to demonstrate to you that there is a relationship between production
targets and the quality ratings from customers.
2) I will dissever the nature of the relationship determined in (1).
3) I will provide you with the necessary information for determining the level to adjust
your DTs to in order to achieve a desired average quality rating.(ie you want and average
quality rating of x out of y-I will tell you how to set a DT to achieve that level of
success)
Plainly put, the goal of the research which I would conduct is to provide xxx with the
know-how to dictate their own quality level. The information from this research will
allow me to demonstrate to you how xxx can maintain a desired level of excellence(as per
its principles) while maximizing production.
Research Design
In order to achieve the former objectives we must go through all the steps that are
involved in a research project. The following is a practical order of events from start
to finish. 
 Identify the problem/opportunity (this has already been done-see above)
 Data Collection Method
-  Design a questionnaire/survey for the target population
-  Define population
-  Determine method of sampling
-  Distribute questionnaires to sample
 Data Analysis
-  Decide which method of displaying data makes analysis most efficient
-  Figure out which test will best suit the objectives 
 Interpret Results
-  Work with management to determine appropriate levels of certainty 
 Present results in layman's terms to management
To do this research for xxx I would need information first on quality ratings and then
information on the corresponding DTs. (ie customer x gave a rating of 3 and he was the
customer of a manager who had DT y)
Data Collection
Questionnaire Design
Because xxx does not have existing data on customer satisfaction as it pertains to
quality, I would be designing the questionnaires from scratch. When designing a
questionnaire there are a few things to keep in mind.
- questionnaire should be relatively short (about 1 page)
- questions should move from impersonal to more personal
- questions should be simple 
- specifically, this questionnaire should cover all aspects of quality(idea of quality
may vary from person to person) 
(personal communication, Dickson, BPA 402, 1999)
See Figure A in appendix(rough draft of a questionnaire) 
This particular questionnaire asks about quality of a xxx job. Possible answers are
provided on an ordinal scale which provides for distinct differences between responses.
Specifically, it was designed around the Likert scale model. In this model a statement is
made about quality and respondents are asked to respond based on how much or how little
they agree with the statement. (ie the quality of work was good: 1)strongly disagree
2)disagree 3)neutral 4)agree 5)strongly agree) By making a statement and making the end
points of responses polar opposites, I can determine the degree to which the customer
feels one way or another about quality. In this case, all statements about quality are
positive so that a high score(5) represents high quality and vice versa.
These types of questions are also known as closed questions. "Closed questions have
become the mainstay of survey researchers."(Patricia Labaw, 1980) However there are
several drawbacks to the closed question, the most notable being "that researchers may
not know what the answers really meant to the respondent."(Ibid) In this case, I feel
like the closed question will be easier to analyze and more efficient. 
Defining Population
In this case we want our population to consist of people who have already had their
houses painted by a xxx manager. In order to minimize variables that might affect quality
other than DTs, we also should limit the population to customers in the state of
Washington and Oregon from the 1999 season. 
Sampling
Sample size represents the number of people from the target population that are going to
receive the questionnaire. Sample sizes will vary depending on what we want to do with
the data and how confident management wants to be that the information wasn't tainted by
sampling error. 
The sample for our purposes will be a probability sample. This is generally more time
consuming and costly, but it will allow us to project our results onto the entire
population and make a more confident conclusion since it is a random sample. (each member
has an equal chance of being selected) The probability of being selected is equal to 1/n
where n is the number of people in the population.
I would choose the stratified sampling method which would allow me to split the
population into two mutually exclusive groups.(ie income or geographic location) Next,
samples are taken from each group at random. This can help to minimize random sampling
error. For instance, we don't want people from only one level of income. In this case, I
would split my population by geographic location. In the xxx business it is common
knowledge that customers vary drastically from area to area. Another appropriate reason
for using the stratified method is that is requires smaller sample sizes. In this
situation, the total population is not that big. Because the total sample size should not
excede 5% of the total population a smaller sample is adequate.
One of the strengths and weaknesses inherent in the stratified sampling technique is the
relatively smaller size of the sample. This means less time and cost but potentially
greater random error. Another weakness is that the information necessary to properly
stratify the sample is not always available.(Gates, McDaniel; 1999) In our situation this
is not the case. In addition, stratification can be time consuming because of the time
required to obtain the necessary information. Again this doesn't affect our research
because the segmenting information is internal and readily available.
Once the means of sampling has been established you can determine the size of the sample
by working with management and the designated objectives.
(See Appendix Ba for equation and calculation) 
What this example data is telling us is with a sample size of 89 we can be 95% confident
that the true mean of the quality ratings will be within 5% of the true population mean.

The standard deviation of this equation will have to be estimated. Usually we would use
the results of a pasts study but there is no such study. Another option would be to
conduct a pilot study and calculate the standard deviation for those results. In this
case we can be extremely positive that any answer will be within 6 standard deviations of
the answer. Thus, 6/5=1.25 is the equivalent of our standard deviation.
Once the sample size is determined we are ready to send the questionnaires out to the
customers. Ultimately we want to find average quality ratings for different DTs. This
means that we need to separate or distinguish the questionnaires that come back by their
respective DTs. One way to do this is to make a mark on the questionnaire itself. (ie
DT=$60,000 gets a red dot; DT=$100,000 gets a yellow dot; DT=$120,000 gets a black dot)
This will allow the person receiving the returned questionnaires to separate them by DT
and average them separately. 
Distribution
Customers will receive the questionnaires on quality via ground mail. Although personal
interviews would be preferable, they would not be cost effective or efficient. Instead,
the respondents will receive a questionnaire including an envelope that is self-addressed
and stamped so that it is easy to return. 89 questionnaires will go out to customers of
managers with a DT equal to $60,000. 89 questionnaires will go out to customers of
managers with a DT equal to $100,000 and the same for the $120,000 DT level. 
One of the things that mail surveys always have to deal with is non-response bias. In a
smaller sample, it could drastically affect our results if people do not respond for one
reason or another. To minimize this possibility, we will hand stamp and hand address
envelopes in order to personalize the exchange. If this doesn't have the desired affect
and people don't respond then we will send out a second wave of questionnaires.
Data Analysis
Testing Results
Initially, we want to test the results of the data we receive for a direct relationship.
In other words, we want to make sure that the differences that are apparent between what
is observed and what is expected are not due to chance. But rather, we want to be able to
show that average quality ratings vary significantly enough with changes in DTs to
warrant our continued analysis. One positive aspect of initially determining whether or
not DTs affect quality is that if DTs do not affect quality, we can stop the research and
save money and time.
This is not as complicated as it sounds, but rather one of the simpler ways to conduct
research analysis. The normal way to set up a crosstabulation table is to make a table
where the rows(horizontal) consist of categories that influence the data in the
columns(vertical).(ie the production targets influence the quality ratings) See
Crosstabulation Table in Appendix.
Once the table has all the data filled in, percentages can be easily calculated on the
basis of row totals. The results lend themselves to easy comparison of the degree of
correlation between DTs and quality ratings.
In addition, we will be able to look at a comparison between the observed(O) value and
the expected(E) value in each cell. The expected value is representative of the value
that would be observed if there was no difference between the variables.(then O=E) The
observed value is taken from the data off the questionnaires. The comparison of the
observed and the expected is done using a test called Chi Squared or X2.
"The X2 test enables a research analyst to determine whether an observed pattern of
frequencies corresponds to or fits and "expected" pattern...many marketing research
studies, possibly most, go no further than crosstabulation in terms of analysis."(Gates,
McDaniel; 1999) 
When conducting a Chi Squared test the first step is to establish the null and
alternative hypothesis. In the case of X^2, the null hypothesis (Ho) is always an
association of no relationship between the two variables(DT and quality). The alternative
hypothesis (Ha) is a significant relationship between the two variables. By looking at
the answer to the equation, we can compare it to the Chi Square table. From that we
determine whether to accept or reject a difference between the two variables.
From the crosstabulation table we can easily apply our data to the X2 equation. See
Figure Bb for equation and calculations.
We know the observed value from our data collection. The expected value can be calculated
using row sums, column sums and totals. See Figure Bd. And K is equal to the number of
categories. This is all the necessary information to calculate Chi Squared. See Figure
Bc. 
Data Interpretation
Note: the data that was used was completely fictitious and is used for demonstration
purposes only.
To calculate the significance of this information two more things are needed. The first
is degrees of freedom and the second is confidence level. With these two pieces of
information the results from the Chi Squared test are used to determine a corresponding
value from the Chi Squared Table that will relate back to the null hypothesis(Ho). From
this information it can be determined if a significant relationship truly does exist
between production targets and quality.
Degrees of freedom can be calculated from the crosstabulation table by simply subtracting
one from both the number of rows and columns and then multiplying the two. 
(R-1)(C-1) = (3-1)(5-1) = 8 degrees of freedom
After that, the researcher would work with management to determine how confident
management would want to be in the answer in order to move forward and take action.
Because it is better to error on the side of confidence I chose a %95 confidence level
for this example.(industry standard)
At these levels the Null Hypothesis exists at 15.5073 
Consequently we can reject the null hypothesis at every level DT because our calculated
value is much higher than Ho.(54.5 vs 15.5073) From this we conclude that there is a
significant difference between quality and DTs.
Further analysis of the data will allow us to determine a very close estimate of the
exact relationship between DT and quality. To do this type of extended analysis I would
recommend bivariate regression analysis. Although seemingly daunting, this sort of
analysis will clearly show the data and the nature of the relationship in such a way as
to permit the management to determine where to set production targets in order to achieve
desired levels of quality.
We use bivariate regression analysis to determine the strength of the linear relationship
between two variables when one is considered dependent(y) and the other
independent(x).(Gates, McDaniel; 1999) In this research, the DT is the independent(x)
variable while the quality is the dependent(y) variable.
See Figure B
Interpreting the graph of the bivariate regression analysis is not difficult. There is an
obvious inverse or negative relationship. This means that as one variable increases the
other decreases. The benefit of such an analysis to the management is our ability to
extrapolate that data. By taking a straight line graph and making a best fit line through
the data points you can look at a DT and immediately get a sense of what sort of quality
to expect from your managers.
Finally, I feel I have demonstrated that the objectives laid before you in this proposal
are reachable and actionable. The insight that this sort of research could provide xxx
management would be unparalleled. It could provide tremendous leverage for maintaining
and strengthening your customer base into the next millenium. 
Figure A
Questionnaire
The overall quality of the work was high.
1) strongly disagree 2) disagree 3) neutral 4) agree 5) strongly agree
The workers kept the job-site tidy.
1) strongly disagree 2) disagree 3) neutral 4) agree 5) strongly agree
The workers were conscientious about your concerns.
1) strongly disagree 2) disagree 3) neutral 4) agree 5) strongly agree
The work was done in a timely fashion.
1) strongly disagree 2) disagree 3) neutral 4) agree 5) strongly agree
The workers paid attention to detail.
1) strongly disagree 2) disagree 3) neutral 4) agree 5) strongly agree
The workers were friendly.
1) strongly disagree 2) disagree 3) neutral 4) agree 5) strongly agree
The manager kept the lines of communication open.
1) strongly disagree 2) disagree 3) neutral 4) agree 5) strongly agree
The manager delivered on what was promised.
1) strongly disagree 2) disagree 3) neutral 4) agree 5) strongly agree
I am likely to recommend College Pro Painters to friends and family.
1) strongly disagree 2) disagree 3) neutral 4) agree 5) strongly agree
Appendix A:
Overview of xxx
xxxis one of many different companies that operate under The Franchise Company. xxx has
been around since the late seventies when it began in Canada. Now it is mainly in the
U.S. where it operates in over 20 states. 
The basic premise behind xxx is a company that hires and trains college students to run
their own businesses. These businesses are called franchises but unlike many franchises,
the student manager/franchisee does not have to buy the franchise. Rather, he or she pays
royalty on every job produced.
After a intense interviewing season in the Fall the new managers go through a rigorous
training period where they are taught everything from selling to production and all the
comes in between. During the Spring months the managers are expected to book the majority
of the work they will be doing over the summer. As the summer approaches, they begin to
focus on hiring workers(also college students) and training them. Once the summer begins
managers are exposed to everything a young businessman could hope to experience and many
other things they hoped not to experience.
In general, it is a very tight ship. xxx general managers(2 in Washington) are
responsible for hiring the managers they will stick with throughout the year. For this
reason, GMs have weekly checkpoints with their managers counseling and guiding them
through the stressful summer. 
Each General Manager is responsible for about 15-20 managers. In an average summer an
average manager will paint about 20 homes. This works out to roughly 400 customers per GM
and 800 per state.
Figure B: 
Graphs of the inverse relationship between quality and production targets
x-axis is quality rating scale(5 is high)
y-axis is number of respondents
x-axis is DT level
y-axis is total quality rating points
Appendix B
a)The formula for calculating the required sample size for problems that involve the
estimation of a mean is(Gates, McDaniel; 1999):
N=(Z2)(2)
E2
Sample size(N) equals level of confidence(Z) squared times standard deviation(o)squared
all divided by the acceptable amount of sampling error(E)squared.
(1.96)2(1.2)2 = 88.5
(.25)2
b)The equation for the Chi Squared test is as follows:
X2= k (O-E)2
E
O=observed value
E=expected value
K=number of categories
c)Chi Squared Calculations(from crosstabulation table)
At a DT of $60,000 X2 = (3-7.3)2 + (5-15)2 + (14-26.3)2 + (29-20.6)2 + (38-20)2 = 33.5
7.3 15 26.3 20.6 20
At a DT of $100,000 X2 = (8-7.3)2 + (18-15)2 + (33-26.3)2 + (18-20.6)2 + (12-20)2 = 5.8
7.3 15 26.3 20.6 20
At a DT of $120,000 X2 = (11-7.3)2 + (22-15)2 + (32-27)2 + (15-20.6)2 + (9-20)2 = 15.3
7.3 15 26.3 20.6 20
TOTAL SUM = 54.5
Degrees of Freedom = (R-1)(C-1) = (3-1)(5-1) = 8 
Confidence Level = 95%
REJECT NULL HYPOTHESIS AT EVERY LEVEL OF PRODUCTION TARGET
d)The calculation for determining E for any given cell is as follows:
E= (Rsum) x (Csum) x total = (Rsum)(Csum)
total total total
R=row
C=collumn
E=expected
References
Brown, Timothy P.(1991), Internal research helps to define service quality; Marketing
News, Feb. 4 1991 v25 n3 p11(1).
Dickson, J.P. (1999), BPA 402:Business Research; Personal Communication.
Gates and McDaniel(1999), Contemporary Marketing Research. Cincinnati... :South-Western
College Publishing.
Gitlow, Howard S.(1990), Planning for quality Productivity and Competitive Position.
Homewood, Ill.: Dow Jones-Irwin.
Labaw, Patricia(1980), Advance Questionnaire Design. Cambridge, Mass.: Abt Books.