# BUSI 275 Spring 2012 Term Project

## Introduction

The objective of this course is not only that you understand the theory (math) behind statistical analyses, but also that you demonstrate how to apply them appropriately to practical situations in business. As such, the term project is a vital component of the course and should represent the pinnacle of your work in this course (more so even than the final exam).
For the term project, you will
• Choose a data-driven, business-related topic you are interested in from the list below;
• Investigate and do background research on the issue;
• Gather data (or find suitable pre-existing datasets);
• Perform appropriate in-depth statistical analyses on the data (e.g., in Excel);
• Present your results in a 15-min in-class presentation; and
• Present your results in a well-written, clearly organized paper.
In the business world you will often (nearly always) need to work in teams, and so in this project you will be expected to work in teams of 2-4 students. Once the project proposal is submitted, the team cannot be disbanded -- so choose your teammates wisely! Learning how to work in a team is one of the main objectives of this project; it is not easy, but it is important and worth the effort. Giving up on working together is giving up on one of the primary objectives; it is your responsibility to make it work! If you are having difficulty with your teammates, I can offer suggestions on conflict resolution, but I will not be your mediator; you must learn to be peacemakers.
"Be completely humble and gentle; be patient, bearing with one another in love. Make every effort to keep the unity of the Spirit through the bond of peace." (Ephesians 4:2-3, NIV)

## Topics

There is a lot of flexibility for your project topic, but the idea is for you to dive into a dataset to glean new insight relevant to business in some way. Most analyses will revolve around a single variable chosen as the outcome (dependent variable, Y). The outcome is typically a quantitative variable; you may choose a nominal or dichotomous outcome, but the analysis will be more tricky. Examples of interesting outcome variables could include:
• "Retail price of new laptops",
• "Starting salary of MBAs",
• "Unemployment rate in BC",
• "Performance of environmentally friendly mutual funds",
• "Acquisition price for soccer players",
• "Future price of oil on the NYMEX",
• "Same-store sales growth in retail",
• etc.
A typical goal for statistical analysis is to gain insight into what factors drive or influence the outcome variable: e.g., what other variables might influence the unemployment rate? These variables are your predictors (independent variables, X's). They may be quantitative (regression) or qualitative (ANOVA). You typically want at least 3-4 in order for your model not to be too simplistic. If you have many predictors, you need to figure out which ones are most important. A different but related kind of analysis that is very common in business is time-series analysis; this tracks a single variable over time in an attempt to predict its value in the future (e.g., stock prices, sales revenue, etc.). If your team is keen on this, you may do this for your project, but be warned that we don't have enough time to cover this material in-class, so you will need to do a lot of self-learning.

## Project Proposal (due 24hrs before meeting, before 3 Feb)

Your entire team must meet with me to go over your project idea. I am usually only on campus on Tuesdays, but during the two weeks from 23 Jan to 3 Feb I will make an exception, but you need to email me to book a time. The proposal meeting must happen by Fri 3 Feb at the latest. Every member from your team must attend the meeting, no exceptions. At least 24hrs before our scheduled appointment, upload a short summary (one-half to one page) describing your chosen topic and the work your team proposes to do:
• Describe the population of interest and the primary outcome variable.
• Describe the predictors and the nature of the relationships that you think those predictors will have with the outcome, e.g., "we think that increased corporate taxes lead to increased unemployment rate".
• Describe the unit of observation and possible sampling strategy: e.g., the unit of observation might be per-person, or per-company, or per-country, or even per-year (for time-series).
• Describe your plan for where to get the data:
• If you will use existing data, what sources have you already explored?
• If you will be gathering your own data, what is your plan?
• Discuss a preliminary plan to divide the work amongst your team members.

### Deliverables:

• As early as possible, email me to set up an appointment. I will try to block out time for you in the two weeks from 23 Jan to 3 Feb.
• At least 24hrs before our meeting, upload your written proposal to myCourses and/or email it to me.
• All project submissions can be submitted by just one member on behalf of the rest of the team.

## Dataset Description (due 7 Feb)

• Who the "owners" of the dataset are,
• If public data, look for a "Legal" or "Terms" page on the website.
• If private data, ask the owners for permission (you will need their permission letter for your REB application).
• The sample size (number of observations)
• For each variable that you plan on using, describe how it is defined, indicate its level of measurement, and show how many missing cases there are.
• For any quantitative variables (interval or ratio), show means and standard deviations, plus box-plots and histograms.
• For any nominal (categorical) or ordinal variables, show the frequency distribution using appropriate charts (e.g, bar chart, pie chart).

### Deliverables:

• Upload a single document (Word, PDF, or similar) to myCourses by 10pm on the due date. It should be formatted clearly and cleanly (make sure to size figures/charts appropriately) in MLA style or similar.

## REB Forms (due 14 Feb)

By Canadian law, any research involving human subjects must be done according to standards of research ethics. If you are gathering your own data (e.g., questionnaires) or using existing non-public data that involves human subjects, you will need to submit a form to TWU's Research Ethics Board (REB) for approval. You are not permitted to begin recruiting subjects or gathering data until you have received REB approval! Allow 3 weeks for this to happen. REB rules allow for certain studies to be exempt from requiring REB approval. We will discuss in class the rules for REB exemption; generally, if you are using public data (e.g., StatCan or US Census), then you will be exempt. This means you don't need to wait for approval before doing your analysis, however, for class purposes, you will still need to complete and upload an REB form to myCourses (it should be fairly easy for you to do). For more on why research ethics is important and what the REB will be looking for, see the Tri-Council Course on Research Ethics (CORE). If the REB rejects your application with major revisions, you may be required to complete this online tutorial. See TWU's REB page for more details; the forms you need are at the bottom of the page: either "Request for Ethical Review" (if you are gathering new data) or "Analysis of Existing Data" (if you are using data that was originally gathered for another REB-approved study, or public data). On the form, list the instructor as your project's supervisor. Your project's principal investigator ("PI") should be one from your team; you may select one team member arbitrarily.

### Deliverables:

• Upload an electronic copy of your completed REB application to myCourses. Also upload any REB-required attachments (questionnaire, script, permission form, etc.).
• If you are not certain that your project is REB-exempt, you also need to submit to me a completed, signed, printed copy of your REB application. The printed copy is due in my hands by 5pm on the due date.

## Presentation (in-class, 10 Apr)

Deliver a 15-min, in-class presentation on your project. If your analysis is not yet complete by the time you do your presentation, that is okay, but you should have some preliminary results to present.
• The format of your presentation is up to you, but you must keep it under 15 minutes, to ensure we have time for all the teams.
• You will be graded not only on the content of the presentation, but also your clarity, delivery style, professional demeanor, etc. Treat this as if you are presenting to your company's CEO or board of directors; assume they are smart but will be bored by statistical jargon. It is recommended that you dress "business casual"; the rule of thumb is to dress "one step above" your audience.
• Every member of your team must participate in the presentation.
• In addition to your own presentation, you must listen to and complete a short feedback form for the other presentations by your classmates. The feedback forms will be provided to the presenters for their reference.

### Deliverables:

• Fill out feedback forms for in-class presentations by other teams.
• After class, upload your presentation slides (PPT, PDF, ODP, etc.) to myCourses.

## Paper (due Mon 16 Apr)

Your paper must be a complete, well-written exposition of the topic you have chosen, the analysis you performed, and your results and conclusions. You will need to do background research and cite reliable sources. As appropriate, include select tables or figures in the body of the paper to illustrate the points made in your paper. Further tables or charts can go in an appendix. As with the presentation, your target audience is someone like your CEO or board of directors -- they are giving you a chance to communicate your results, but you need to convince them of the relevance and validity of your work in a way that they can understand. Do not assume they are familiar with statistical methods. There is no length limitation on the paper! However, your paper must satisfy all the requirements given, following the outline below. Typically, BUSI275 papers that meet these requirements have averaged around 4000 words. But there is no minimum length; if you can write a clear, concise paper that meets all the requirements in less than 4000 words, so much the better!
• Abstract: a short summary (half a page at most) describing the basic results of your research at a glance.
• Introduction: describe your topic and tell me why you think it's relevant or interesting.
• Related Work: research what other people have done related to your topic and summarize their findings. For example, another researcher might have done the exact same analysis but on a different dataset. Try to keep the introduction and this section short, in favour of more space for your own methods and results.
• Methods: tell me about the variables and the dataset you have chosen, and describe how the data were gathered. Describe your analysis in a way that is understandable to someone who might not know or care about the statistics, but also is sufficiently detailed that an interested party could reproduce your results.
• Results and conclusions: what effects were significant? What effects weren't significant? Do the results agree with what you expected? Interpret the results and tell me what it all means in the context of your original topic.
• Future work: if you or another researcher were to continue in this topic, what would be the next step? How could your dataset or analysis be improved? (This section can be short.)
• References: particularly for the "Related Work" section, you will need to do background research and cite sources. The point of a list of references is to enable the reader to look up your sources.
• Appendix (optional): For tables/figures in the body of the paper, you should be very selective so that they do not take up too much space or overwhelm or bore the reader. For any additional tables/figures you wish to include, you may put them in an appendix. You may also separately upload Excel spreadsheets or other datafiles. The appendix is optional.
Your paper should be in proper, professional English. Avoid colloquialisms. As appropriate, prefer the active voice ("We performed linear regression on the data") over the passive ("linear regression was performed on the data"). Actions have agents, so indicate who they are: if you were the one who performed a step, then the use of the pronoun "We" (or "I") is appropriate ("We performed..."). If you are citing someone else's analysis, then indicate who did it: "McDermott et al. performed a similar analysis...". I will not be strict on formatting, as long as your paper is clear and readable. A suggested format is the MLA style: Purdue OWL has a helpful guide. I highly recommend that you email me a rough draft early on to get feedback. This paper constitutes a major portion of your final grade, and you don't want to be surprised that you were heading down the wrong path!

### Deliverables:

• Upload your complete paper, as a single document, to myCourses by 10pm on the due date.
• If you wish, you may upload any datafiles, spreadsheets, etc. as separate files, however the paper should be complete without these.

## Marking

Proposal: 8% 8% 4% 20% 60%