Week 01 - Intro

Prof. C. Austin Davis

Spring 2026

Introductions

Welcome! Roadmap for today

• Introductions
• Syllabus and what to expect
• What is data
• Intro to R
• Preview of next time

Introductions

• What is your name?
• What is your quest?
• A mundane little quirk OR a hobby you’ll never quit

A little more about me

A little more about me

A little more about me

A little more about me

Syllabus

Why are we here? Learning objectives:

• translating important questions into quantitative frameworks,
• selecting and conducting appropriate methods of quantitative analysis,
• drawing substantive insights from your quantitative analysis, and
• communicating your methods and conclusions so that others can understand and interrogate your quantitative reasoning

What will we learn?

Descriptive statistics

How can we synthesize large quantities of information to establish general, actionable facts about the word?

Regression

What is the nature of relationships between key measures of environments, behavior, or well-being?

• What do we know about mental health from surveys?
• How do traumatic experiences relate?

What will we learn?

Inference

Do these facts or relationships generalize?

Causality

How can we identify causal relationships in data?

• What can we say about the pop?
• What programs lead to improvements

Why R is our primary tool

• Free and open source
• Powerful but approachable
• Makes it easy to share and reproduce results

Why R is our primary tool

Job relevant

• “Many positions I’ve been applying for request quantitative research experience and frequently request specific experience with R …”
• “There is a sentiment among students that [SIS 600] required too much coding that ‘they will never use this in the real world’ … Not only have I seen evidence to the contrary in my short time since graduation, but I have also come to understand that R is a far more powerful tool than I had previously understood as a student.”

Our guides

[Text]book chapters

• Wheelan Naked Statistics
• Bueno de Mesquita and Fowler Thinking Clearly with Data

Our guides

Course-specific materials we provide

• SIS 600 R Survival Guide
• Problems and exercises with solutions
• Written feedback on presentation, paper
• Lecture outlines with code examples

Your notes are key learning materials from this class

• Most classes I ban electronics.
• Research suggests laptops bad, handwriting good
• I will facilitate that by providing outlines

How will I be evaluated?

Quizzes (7) (10 percent)

• First set (next meeting) covers syllabus
• Approx. three basic questions based on reading
• Nine of these with two drops no questions asked
• Anticipates material covered in same class

How will I be evaluated?

Homework (5) (30 percent)

• First set covers basic math skills
• Generally one or two multipart questions
• Reinforce material covered in prior class

Midterm exam (20 percent)

How will I be evaluated?

Project planning worksheet (individual) (5 percent)

• Structures basic elements of your project
• Opportunity for early feedback

Project presentation (individual) (10 percent)

• Present preliminary analysis
• Pitch to your classmates for group formation

Project final paper (group) (25 percent)

How will I be evaluated?

Assignment	Due date	Weight (pct)
In-class exercises (7)	See Syllabus / Canvas	10
At-home problems (5)	See Syllabus / Canvas	30
Project planning worksheet	Mon Mar 2 / Tue Feb 24	5
Midterm exam	Mon Mar 23 / Tue Mar 17	20
Project presentation	Mon Apr 13 / Tue Apr 7	10
Project final paper	Mon May 4 / Tue May 5	25

• Canvas is the authoritative source for due dates

Grading*

Assignnment grades

For quizzes, homework, and exams, each question (or part) will be worth one point and scored as follow:

• 1 point: a full and correct answer
• 0.5 points: any reasonable attempt, i.e. evidence of effort
• 0 points: no reasonable attempt, i.e. no evidence of effort

Course grades are curved

Attendance and evaluation

• I expect you to be here absent extenuating circumstances
• Quizzes, exam, presentation are all in class
• Quiz drops to account for absences
• If something serious comes up, contact me ASAP

Group work

• I encourage you to collaborate
• But all submissions must be original & individual
  • OK: Discuss concepts, programming errors, results
  • Not OK: sharing electronic files, copying and pasting
  • Exception: group final project
• Any questions or concerns, ask me

AI tools like ChatGPT

• Things are moving fast
• Promise
  • Annotated reference code
  • Something between Google Searching and Office Hours
• Peril
  • Can be inaccurate, incomplete, biased
  • Rely on it so much you can’t tell when it leads you astray

AI tools like ChatGPT

• Policies and principles for this class
  • You may use AI tools to assist with at-home problems, but must make good-faith effort first
  • Anytime you use an AI tool, submit inputs and outputs from the tool as an appendix to the assignment
  • You are responsible for the quality and integrity of your submissions
• Recommend you read output carefully

Late work*

How can you do well in this class?

• Be consistent and budget plenty of time
• Review feedback on graded work
• Ask for help
  • cadavis@american.edu
  • My office hours: M 10-11am; W 2-3pm; F 4-5pm
    • Please use calendly
  • TA: Pietro Zecca, OH TBD
• If needed, do more

Typical meeting

• In-class exercises + solutions
• Review at-home problem(s)
• Prof-led presentation / working examples
• Intermission
• Prof-led presentation / working examples
• Preview of coming week

Scheduling notes

• No class on Jan 19 (MLK)
• No class on Mar 9 / 10 (Spring Break)

Intermission

Randomness is everwhere

A process that can be repeated with different outcomes each time, i.e. the results are not perfectly predictable.

• Roll of a die, flip of a coin, draw of a card
• Gas prices, rainfall, vote totals
• SAT/GRE score, income, number of children

Our basic goals

• Assessing the likelihoods of various outcomes
• Uncovering systematic relationships

Our basic tools

• Probability theory
• Repeated, systematic measurement, i.e. data
• Statistical methods

Probability: characterizing uncertainty

Frequentist

Over a large number of repeated, identical trials, in what fraction of those trials would some event occur?

Bayesian

Given all the available evidence, what is a reasonable expectation about the likelihood some event will occur or some hypothesis is true?

Shared conceptualization and basic properties

Probabilities range from zero to one inclusive

• Probability of zero (\(p(A)=0\)) means the event is impossible
• Probability of one (\(p(A)=1\)) means it is certain, or
• it is possible but not certain (\(0 < p(A) < 1\)).

Concrete examples*

• \(p(\text{Heads})\) for a fair coin?
• \(p(6)\) for a fair six-sided die?
• \(p(\text{Rainfall 20016} > 0)\)?

Flipping a fair coin

I asked the computer to simulate 1,000 coin flips and plot the results. A few things to notice here - Graphical representation of probabilities: horizontal axis represents the possible outcomes, vertical axis represents their frequency. This is called a histogram. - The bar heights are both near 0.5, meaning we get heads about half the time and tails about half the time. - Even though the coin is fair, i.e. there’s an equal change of landing on heads or tails, we do not get exactly 500 heads and 500 tails. - We understand the process pretty well. We know enough to make good decisions. We can make informed bets. Imagine a casino or carnival game where you paid a dollar to participate. If the coin comes up heads you win ten dollars. If the coin comes up tails, you get nothing and the casino keeps your dollar. Would you play that game? - But there’s still fundamental uncertainty about the outcome. You could lose any one coin flip. Or two. Or more.

Rolling a fair die

Here, I asked the computer to simulate 6,000 coin flips and plot the results. Basically the same story as with the die, except we have more possible outcomes.

Rainfall in 20016

Here I’ve plotted daily precipitation in our zip code from 1948 to 2023. We learn something about general weather patterns in DC. How often does it rain? But we still have limited ability to forecast the weather tomorrow or next week.

What is data?

Examples

Key definitions

• Data
• Unit of analysis
• Variable

What does data look like?

Definitions

• Systematic observation of many “units” / Pieces of information structured to facilitate analysis and comparison.
• The type of entity about which observation is gathered.
• A property or characteristic that can take on different values for different units.

Examples

• Surveys (individual, household, firm)
• Remote sensing
• Administrative
• Texts and media

What does data look like?

Roll Result
4
6
1
4
2
3
1

What does data look like?

Flip Result	Heads?
Heads	1
Heads	1
Tails	0
Heads	1
Tails	0
Tails	0
Tails	0

What does data look like?

Date	Precip (in)	Temp (min)	Temp (max)
1948-08-01	2.68	68	82
1948-08-02	0.26	67	86
1948-08-03	0.00	71	87
1948-08-04	1.71	69	87
1948-08-05	0.00	66	76
1948-08-06	0.00	54	77
1948-08-07	0.00	56	80

Why use data?

• A powerful way to answer questions and draw conclusions, particularly about repeated behaviors or large groups
• But to do that we need to map abstract concepts to measurable quantities

Exercise: operationalization

• Military power / capability
• Freedom of speech
• Poverty
• Trauma
• Mental health

• Again, the great strength and limitation of data is that, because things are measured in a consistent way, we can aggregate or compare withing large groups.
• Just to give a personal example, we have relied a lot on data in making decisions related pregnancy and parenting. There are a million questions that come up, like (and this is true), is it safe for a pregnant person to eat a ham sandwich? Is it safe to take ibuprofen? How can we get our baby to sleep through the night? These are all questions about which data is highly informative. We relied a lot on an economist named Emily Oster, whose books and blog provide such vivid examples of the usefulness and limitations of data. Even if you’re not a parent, she will train you well!

How does the data capture the concept? And how well?

Measurement metrics

• Categorical
  • Binary
• Continuous
• (We will mostly sidestep count variables)

• variable assumes a finite set of named categories.
• variable assumes values from a continuous or nearly continuous range of numbers with equal unit difference

Reliability and validity

Random variables

• A measure of probabilistic or uncertain process
• Specific realizations of an event captured on real number line
• E.g. daily max temp at BWI

Data generating process

• Underlying process that yields specific realized vaules of RV
• E.g., combined meteorological forces that determine temperature like
  • solar radiation
  • surface albedo
  • atmopheric composition

Probability distribution

• Precise specification of probability of event or range of events
• Often expressed graphically to show likelihood of possible realizations of random variable
• Appropriate representation depends on values variable can assume
  • Continuous vs. categorical
  • Bounded vs. unbounded
• Can take any shape such that the total area sums to 1
• Area of subregions measures probability of ranges of events

Probability distribution

Total incomes of working people in 2019 ACS

Probability distribution

Total incomes by gender

A data generating process

• F indicates female gender
• D indicates discrimination
• A indicates “ability” or productivity
• O indicates occupation
• Y indicates income

Let’s get our hands dirty!

• Stop talking in abstract, get familiar with our tools
• Walk through the different panels
• Do arithmetic in the console

Thinking like R

• Scripts
• Objects
• Functions
• Packages / libraries

Scripts in R

• Canvas to download template
• Save a series of instructions
• Make things transparent and reproducible
• Ctrl + Enter runs one line
• Ctrl + Shift + Enter runs entire script
• Options also available via dropdown

Objects

• Different kinds of structured information
• Have names
  • starting with a letter or a “.”
  • made of letters, numbers, “_“,”.”
  • Can’t use FALSE and other special ones
• Defined using the assignment operator

Functions

• Self-contained bits of code
• Always have parentheses, sometimes arguments
• Perform operations or manipulations on arguments
• Return an output
• Use c() to combine or concatenate objects

Libraries

• User-contributed add-on to R
• New functions, new types of objects, new syntax
• We will use tidyverse a lot, so I’ll show you how to install it

Preview of next time

Describing distributions using real data

Questions to ponder

• How could you visually summarize the mental health of CBPS respondents?
• How could you visually compare the mental health between refugees and members of the host community? Men and women?
• If you had to guess, what do you think that comparison would show?
• Assume your guess is correct, what are some policy implications?

See you next time!