Quantitative Analysis （stata）

Homework 1: Telling a Story Using Descriptive Statistics and GraphicsThis assignment will help you begin thinking about how to tell a story using
descriptive statistics such as means, medians, standard deviations, percentiles,
and frequencies. The work you do in this assignment will be directly applicable to
the analysis memo and the final policy report. You can begin to think about what
topics you would like to explore for those assignments.
Assignment Requirements
● Read the documentation for the course dataset.
● Read the “Introduction to Data Structure & Stata” document
● Read the data into Stata. Pick two or three variables on a related theme and find
the appropriate descriptive statistics for them (sample means, standard deviations,
medians, frequencies, histograms or bar charts). Write a two-page memo (singlespaced) describing your findings. Incorporate your graphics or tables into the
memo as you tell your story. (This is steps 6-8 below. Be sure to read “Writing Tips
for Quantoids” on Canvas.)
Upload your memo as a Word document. Additionally, upload the .do file that you
used to run your analysis.
Get Acquainted with Stata and the Data Set
1. Make a plan
Before you begin in Stata, you should spend some time thinking about the data
and your research question. Decide on a few variables that interest you and how
you want to explore them (which descriptive statistics and graphics). To learn about
the dataset, read the description (ACS 2019 description.doc) and look at the
variable list (ACS 2019 variables.xlsx). You will have to check the IPUMS website
(linked in variable list) for their meaning and codes!
2. Get the data and start your software
From your CSDE remote server session, open Stata, and start a new do-file with
the task (e.g., “Homework Assignment #1”), your name, and the date at the top of
the file. (Refer to the ”Introduction to Data Structure & Stata” guide if you need help
with this.) Set your working directory to the file in which you are storing your class
work and read the 2019 ACS dataset into Stata. Make sure to save your .do file
(filename format: [last name]_[task]_[date].do) to your working directory as well .
3. Get some statistics
Reminder: When you are using a new data set, it’s good to get in the habit of
summarizing each variable. This includes: looking at the number of observations,
tabulating categorical variables, and generating means, standard deviations, and
ranges for continuous variables. That way, you can determine if you have missing
values for variables or assess whether there are any unexpected data values or
distributions (sometimes known as “outliers”).While you are encouraged to revisit
the “Introduction to Data Structure & Stata” document, the table below provides a
quick summary of some of the more common summarizing commands. Note:
varname is the generic Stata placeholder for any variable with which you are
working.
4. Developing Graphics
Reminder: Bar charts are appropriate for discrete variables. Use a histogram or
scatterplot for continuous variables. Pie or bar charts show each value separately
(disastrous with hundreds of different variable values).
Recall that once you have generated the graph from your code, it will be
displayed in a new window. You can click on the “Graph Editor” to modify its
appearance, and you can save your graph to use it later. Note that by default
Stata saves graphs in its own format (“.gph”), and make sure to select a different
file extension (for example, “.png”) if you want to include the graph in the memo.
Find more graph-related ideas
5. Saving your data
To save your data after you create new variables, pull down the FILE menu from
the data editor, select SAVE AS, and type in the name of your new dataset in an
appropriate drive – such as the H:\ drive
6. Saving Stata Output
To save your output, make sure you are saving your .log file. (Remember to save
your do-file, as well.)
7. Write the Story
Write a two-page memo (single-spaced including graphics) describing your
findings. Incorporate your graphics or tables into the memo as you tell your story.
(Be sure to read “Writing Tips for Quantoids” on the course website.)
When you’re done with your initial computing, study the results and start writing.
Writing about statistics is an iterative process because, as you write, you’ll find you
need a different graphic or statistic to make your argument. The goal of your memo
is to tell a story supported by data to a policy maker who is interested in the issue
you’ve just looked at (not in how you manipulated Stata). You will need to interpret
your findings for a non-technical audience — NOT explain how you got them.
As you write your memo, make sure you touch on the following points:
● What research questions are you exploring? Explain your theories about how
and why factors are related to each other.
● What is the overall story that you can tell with your data? Discuss the evidence
concerning the research question provided by your results.
● Explain what you’ve learned clearly and concisely without resorting to statistical
jargon. Be sure to think about the constraints of the data, issues that need further
exploration, and the needs of your audience.
● Make sure that all graphics and tables would be understandable to someone who
is unfamiliar with the dataset. This could mean including a title, use your own labels
(not variable names), increasing the font size, creating horizontal and vertical axis
labels, or any other features (e.g. a legend for categorical variables) that help
clearly represent the data.
Writing Tips for Quantoids
Empirical information can be compelling and interesting when combined with a
model and understanding of the world. However, it’s your job to weave the statistics
into a narrative. Data, statistics. and graphics will help you “tell your story”; they are
not the story.
DO:
Use numbers to support your argument, not to make it. Don’t write about the numbers;
write about ideas and hypotheses.
Use pretty pictures to keep your readers’ attention. Graphics (histograms, bar charts, etc.)
can help tell your story visually. Some people can most easily use info presented graphically.
Describe every statistic or graphic you use in the text. If you can’t find room in the text for
the explanation, ask yourself if the numbers belong there.
Provide enough information so a statistician can evaluate your work. Use appendices and
footnotes for technical details of the survey or analysis. This gives other researchers clues
about the details of your work and how they can further study the issue.
Provide a clear summary of your conclusions drawn from empirical info. It’s your job to
synthesize all of the data and information. Your one page executive summary (at front of
memo) may be the only thing your client reads.
Acknowledge the shortcomings of your data and methods. No data or study is perfect for the
purpose; yours will be no exception. Part of your job is to explain the uncertainties and
caveats associated with your study. What can’t we know from this study that we care? What
data, samples, or methodologies could provide better information for our purposes?
DON’T:
Don’t try to teach statistics. Busy policy-makers can’t wade through a treatise on statistics.
Your job is to provide the “translation”, not a lecture on the definition of a confidence
interval or a p value.
Don’t use the text to describe how you created the statistics. Most clients will not care about
your process only your results. Data collection and your analysis methods should be
described in an appendix, with only the important information in the text.
Don’t use a cookie-cutter writing style to describe your outcomes. (E.g., For the eastern WA
households, the mean was 10 and the standard deviation was 20. For the western WA
households, the mean was 13 and the standard deviation was 15.) Vary your sentence
construction and style to create interest. Concentrate on “telling your story” to facilitate this
process.
General tips for memo writing:
Make your sentences short and direct. Policy makers are busy people–don’t make them wade
through endless clauses and qualifying statements. Make your sentences active, clear, and
concise. This is a professional paper and there is no extra credit for flowery phrases or
complex clauses. It’s your job to make the paper readable and easy to understand.
Headings and short sections give direction to reader. Bold or underlined headings allow the
reader to skim and will emphasize important points. By creating short sections you will keep
the reader interested and keep your analysis on track.
Put important points first. Don’t hide the information your client needs by prefacing it with
details. This is not a mystery–you don’t have to build your argument for 3 paragraphs before
presenting it. Also, extended quotations and repeated citations and references are
unnecessary. Put sources in footnotes or appendices, unless important to client.
Always include an executive summary. This is the take-home message for your analysis. It
should include the research question, data source, and the key results. A policy-maker
should be able to throw away the rest of the memo and still perfectly understand the quality,
results, and implications of your analysis.
Don’t forget your audience. Only provide information the client needs. Interesting facts don’t
belong in the analysis unless they will be useful to the client. Don’t forget who your client is
and what info they need.
Use appendices for tangential or extended materials and information. Your analysis should
be lean–just the facts. Use appendices for information you think client could skip, but may
want or need. That way, the flow of your narrative is not disturbed and the client can choose
to read appendices now, later, or not at all.
Appearance counts. Make your paper look good by using sections, headings, a readable font,
and useful graphics. And, don’t forget to spell-check. It may seem trivial, but misspelled
words won’t engender confidence in your analytic abilities.
Writing takes practice. Don’t expect your writing to be perfect on the first draft. It takes time
to make words serve your purpose. Good rewriting is the key.