You work for a consulting firm and have been asked to use regression analysis to address the
topic that you have chosen. Your data set will be found on Google Drive, with information I will
give you separately. It is a subset of the US Bureau of the Census’s American Community
Survey (ACS) 2008 – 2012. You can find more information about the survey HERE if you’re
interested, but outside research is not required for this project.
For the project, you need to take this data and decide what regression or regressions you should
run to answer your question. You can also include other statistical analyses in addition to these
analyses such as pivot tables, histograms or other graphs, etc.
You are likely to need to make some new variables from old ones… For instance, most of the
variables are categorical, and you will want to make some dummy/indicator variables from these
categorical variables.
Part I, due Monday November 17 at 6pm to 520F
Question: Do immigrants of age between 25 to 45, with a bachelor’s degree or above who work
as an accountant or auditor (0800) earn more than comparable non-immigrants who work in this
occupation?
Occupational Codes: (0800) Accountants & Auditor
Ages: 25-45
Set limit to working full time? Yes – full time
Education: bachelor’s degree or above
Part I includes:
(1) a Powerpoint presentation handed in in hard copy, with 2 slides on a single page.
(2) an Excel spreadsheet uploaded to SMGTools (Assignments, Project Data Set.)
The PowerPoint presentation should include a maximum of 5 slides.
? Slide 1: Introduce the question you are addressing (and anything else you want to
put here)
? Slide 2 (or more): Your regression output put into a table like the one in problem set 4 page
6 (with either standard errors or t-stats in parentheses under the coefficients and this fact
included at the bottom of the table).
Somewhere on this slide you should write the main facts we should be learning from your
regression that addresses your point. You will probably want to do this in a text box. If you
have many results to discuss, you can duplicate the regression on more than one slide, with
different test boxes on the different slides.
SM222 Modeling Business Decisions Fall 2014
? If you have a second regression or some other analysis that is key to your result, you can
include this also as an additional slide, once again explaining in words what you found.
? Final slide: Conclusions
Not only will we will grade this presentation, but we will choose projects to be presented in
class the week after Thanksgiving break. These presentations will be 10 minutes, and you
and your partner will present corrected versions of your slides to the class. Your slides should
be such that they could be presented in 10 minutes. The presentation will go towards
classroom participation points but will not affect your grade on part I. However, the
feedback you get from presenting might help you in part II.
We will grade part I (these slides) using the same criteria for the analysis as we list below for
Part II. Therefore we suggest that you should have all of your analyses completed for Part I,
although they don’t have to be discussed as thoroughly. You may change your analysis
between Parts I and II as necessary based on feedback you get on your slides and/or
presentation.
The Excel data set should include:
? A worksheet labeled “data� which includes your original data plus and any new variables
that you have made (with their formulas remaining).
? Separate tabs for the regressions in your presentation and other analyses in your
presentation.
We will not grade the Excel spreadsheet (except for completeness). However, we may refer
to it as we grade your slides and take off if your data creation or your analysis was done
incorrectly.
Part II due December 8 at 6pm to 520F
Part II is a 2-3 page report in hardcopy and uploaded to SMGTools (Assignments – Project 2
Part II), including all of the following:
1. Introduction: What question you have addressed, what data you are using and whatever
else you believe belongs there.
2. Analysis and methodology: Explaining your statistical analysis, interpreting the results
that address your question, etc.
3. A conclusion about what you found (and anything else you believe belongs in your
conclusion, including any caveats, limitations or reservations you have about your
analysis.)
PLUS
Exhibits including the table with your regression or regressions plus any additional statistical
analysis and graphs you refer to in your paper.
SM222 Modeling Business Decisions Fall 2014
Please either single space your report (or use 1.15 line spacing) and put spaces between the
paragraphs. Use 11 or 12 point type and 1� margins.
Also, you need to re-post your dataset to SMG tools including any new variables you have created
and used (with their formulas still there) and tabs with the analyses included in your project. As
before, we will not directly grade this Excel spreadsheet (except for completeness). However, we
may refer to it as we grade your project and take off if your data creation or your analysis was
done incorrectly.
Your project will be judged on the following criteria:
? Do your report and slides use statistical analyses (and particularly multiple regression) that are
most appropriate to answer your question? More specifically, have you thought through what
variables should be in the regression and do you understand their role in the regression?
? Do you deeply understand what the results of the multiple regression tells us? Does your
discussion of its results demonstrate a correct and deep understanding of the statistics taught in
the course? This really is where you will get or lose most of the points!
? Does the writing develop the ideas in logical order and clearly?
? Does your project explain the statistics in words an executive/client could understand?
? Are regressions presented in easy to read tables and/or as equations? Are any graphics you use
appropriate and graphically do they clearly convey information to the reader?
? Is your report professionally and clearly written? We realize that those who are not native-
speakers may not be able to write the most elegant English, but your writing can be clear, your
spelling can be checked, etc.
? The best projects will have a deep understanding of the interpretation of the coefficients holding
specific other X variables constant (and know when it is appropriate to hold other variables
constant.)
? The best projects will have a deep understanding of testing and interpreting statistical
significance of coefficients, as well as making predictions from your model (with confidence
intervals) if appropriate.
? The best projects will use nonlinear terms, multiple dummy variable for a single kind of
categorical dummy and/or interaction terms, etc. as appropriate and interpret them correctly.
? The best projects will compare the simple regression or single key X (RHS) variable(s) with one
or more multiple regressions to show how simple correlation doesn’t (or does) reflect the true
causal direct impact of the key variables.
? More generally, the best projects will address omitted variable bias, how you’ve dealt with it,
what variables you are still missing, and how those affect the results that you have.
? We will also be posting and updating a FAQ (frequently asked questions) on SMGTools
about this project.