Practicum 1

Due Week 8 · Friday, October 16

Author

Prof. Jack Reilly

Published

Fall 2026

Rules

  • This Practicum is due Friday, October 16, by midnight. Late work is accepted up until Tuesday, October 20 at 2:00 PM, with a one-point penalty. As usual, we will cover the practicum in class, so work submitted after Tuesday at 2:00 is not accepted.

  • You are permitted as much time as you like between when the exam is handed out and when it is due. The practicum IS open-book, open-note, and open-internet, but it IS NOT open-human or open-ai. In short: you can use any resource you want, so long as that resource does not involve asking another intelligence, whether human or artificial, a question.

  • Exceptions:

    • You can ask the professor clarification questions (including about AI use).

    • While AI use is not open, you can use AI for certain limited cases. If you use AI on this practicum, you must keep a record of the prompts you use in a plain-text file or Word document and submit it with your assignment, and your use must fit within a category of acceptable use.

      • You MAY NOT feed the assignment directly into an AI system.
      • You MAY NOT ask it about how to create particular code in Stata, ask it to create interpretation language for you, or ask it substantive questions like “what is an ordinal variable?”
      • You MAY NOT ask it to evaluate your practicum after you have written it for content or substance, nor ask it to assess your entire code file for accuracy.
      • You MAY ask it for help with troubleshooting individual lines of code when you receive an error.
      • You MAY ask it to copyedit sections of writing once you have created it.
NoteAcceptable and Unacceptable AI Use

Another way of thinking about this guideline (AI for creation and knowledge queries = not OK; AI for editing = OK):

  • Unacceptable: “Here is my practicum; please provide me with draft code and answers for all questions.”
  • Acceptable: “This line of code gives me an error; what is the mistake?”
  • Unacceptable: “How do I recode this variable in Stata?”
  • Acceptable: “Do I have any typos in this paragraph?”
  • Unacceptable: “What does this regression output mean?” or “Double check to make sure that my interpretation matches this output and is correct.”

Directions

Please type your answers and turn your exam in to Blackboard. You will need to turn in at least two things: a plain-text Stata .do file and a write-up interpretive document as a .pdf. (If you use AI, you must also turn in a record of your prompts.) In the header of your write-up, make sure to include your name, the date, and the course. Make sure your write-up is cleanly formatted.

In your Stata .do file, follow all the style guidelines we have discussed in class. Your .do file, once I change the file path, should properly execute all commands needed to reproduce the results of your practicum, and should do so without any errors. If your .do file does not do this, the relevant answers will be considered wrong. Furthermore, make sure to include an appropriate header and comments throughout your .do file.

The Data

The American National Election Studies are surveys run every two years during the United States presidential and midterm elections, and represent a nationally representative sample of all people in the United States at the time they are sampled. Use the 2016 National Election Study to answer these questions. (This data file can be found in the course drive.) Also, please note there is a codebook for this data file that you may want to use called Frequencies.pdf.

Data Work (.do file)

  1. Use the following variables: V162034A (vote) and V161158X (party identification).
    1. Recode both variables to remove missing data. (Leave in third party votes). Title both cleaned variables as vote_clean and pid7_clean.
    2. Cross-tabulate vote and seven-category party identification.
    3. Recode the seven category party identification measure to a three category measure. Your new variable should code strong and weak Democrats as “Democrats”, strong and weak Republicans as “Republicans”, and independent “leaners” - those originally in categories 3 and 5 - as independents alongside true independents. Title this variable pid3_clean.
    4. Cross tabulate vote and three category party identification.
  2. A “feeling thermometer” is a variable where respondents are asked to share how warmly or coolly they feel towards an individual, group, policy, etc. Use the codebook for the dataset or the search function in Stata to locate the feeling thermometers for Donald Trump, Hillary Clinton, Scientists, Big Business, the Black Lives Matter movement, and the Pope.
    1. Clean the variables to remove any missing data values, and title them trump_therm, clinton_therm, scientists_therm, business_therm, blm_therm, and pope_therm.
    2. Calculate summary statistics for all thermometers.
    3. Correlate each of these thermometers.
  3. Use the following variables: V161361X (income), V161267 (age), and V161270 (education).
    1. Clean these variables such that missing data is removed and each variable is, at minimum, ordinal (has categories that clearly increase or decrease as you move along their scale) if not interval. Recode all levels of education less than a High School degree to be in the same category, just under “high school” degree itself. Title the variables income_clean, age_clean, and educ_clean.
    2. Regress trump_therm on income.
    3. Regress trump_therm on age.
    4. Regress trump_therm on education.
  4. Come up with the most compelling graphical way of showing the relationship between education and feelings towards Trump that you can. Be creative: explore the graphics controls in Stata, make sure your graph is clean and presentable, etc.

Write-up Document (.pdf)

In your write-up, please answer the following questions, based on results from your .do file.

  1. Vote and PID
    1. How many Trump voters are also strong Democrats?
    2. What third party candidate did the best among strong Republicans?
    3. Using the three category PID measure, did Democrats or Republicans, overall, have an identification advantage in the 2016 election?
  2. Feeling Thermometers
    1. Of the six feeling thermometers, which two are most strongly correlated?
    2. Of the six feeling thermometers, which two are most weakly correlated?
    3. Which of the six people or groups is most warmly received by the American people?
  3. Regression
    1. Paste in your regression results for “age”. Does “age” predict feelings toward Trump? Interpret the regression coefficients. Is it a strong relationship?
    2. Paste in your regression results for “income”. Does “income” predict feelings toward Trump? Interpret the regression coefficients. Is it a strong relationship?
    3. Paste in your regression results for “education”. Does “education” predict feelings toward Trump? Interpret the regression coefficients. Is it a strong relationship?
    4. Based on the above work, what can you tell me about the relationship between feelings towards Trump and education? Income? Age? (short paragraph, 3–5 sentences)
  4. Graphics
    1. Include your figure in your write-up document.
    2. Does this figure suggest anything different to you than the regression from (#3)?

Evaluation

You will be evaluated for both the accuracy of your results as well as the clarity of your code file.

Hints

Don’t forget that you have Stata’s help files as well as the optional Acock book to help you. You will need, at minimum, some or all of the following commands: use, cd, tab, sum, gen, recode, corr, reg. Don’t forget to use options: for example, generating a new variable can be specified after a recode like this:

recode pid7_clean 1/2=-1 3/5=0 6/7=1, gen(pid3_clean)