Syllabus (public companion)
Orientation and course shape — Blackboard is authoritative for graded specifics
This public page is an orientation companion to the syllabus, not the operational syllabus itself. It describes the shape of the course in qualitative terms. It carries no numeric grade weights, no schedules, and no submission rules. For everything that counts toward your grade — exact weights, deadlines, policies, and submissions — Blackboard (the LMS) is authoritative. If this page and Blackboard ever disagree, follow Blackboard.
Course description
This course studies how statistical evidence is produced. Good analysis begins with a question, a population or process, a unit of analysis, measurements, a design, and assumptions about how the data were generated. We begin with statistical questions, units of analysis, variables, measurement, and operational definitions, then study random sampling versus random assignment, bias, confounding, internal and external validity, completely randomized experiments, blocking, paired designs, factorial experiments, interactions, observational studies, causal diagrams, surveys, sampling frames, stratified and cluster sampling, missing data, nonresponse, and study critique. The course modernizes the traditional design-of-experiments course by placing experiments inside a broader evidence framework: experiments stay central because random assignment is one of the strongest tools for causal evidence, but most real work involves observational data, imperfect measurement, incomplete frames, and missing data. The emphasis throughout is reasoning, interpretation, and communication — recognizing what a study design can and cannot support.
Who it is for
This course assumes a prior introductory statistics course or comparable preparation: you should be comfortable with variables, data tables, summary statistics, graphical summaries, confidence intervals, hypothesis tests, p-values, and basic regression interpretation. Prior regression and prior coding are helpful but not required — the course uses some regression ideas when discussing adjustment, confounding, and interactions, but the focus is design reasoning, not advanced modeling, and software examples are scaffolded. The main expectation is willingness to read studies carefully, write short evidence critiques, and explain design choices in complete sentences.
Learning outcomes
By the end of the course, a successful student will be able to:
- Translate a research question into units of analysis, variables, measurements, comparisons, and a target claim.
- Distinguish populations, samples, treatment and comparison groups, outcomes, covariates, and experimental units.
- Explain the difference between random sampling and random assignment and why each supports a different kind of conclusion.
- Identify common threats to validity: selection bias, measurement bias, confounding, attrition, nonresponse, missing data, and post-treatment adjustment.
- Design and analyze basic experiments with two or more groups, and use blocking, pairing, and factorial designs to improve precision and structure.
- Interpret main effects and interactions in designed studies.
- Evaluate observational studies, distinguish association from causal evidence, and use simple causal diagrams to reason about confounding, adjustment, and backdoor paths.
- Explain the importance of sampling frames, coverage, nonresponse, and population definition, and compare simple random, stratified, cluster, and multistage sampling.
- Critique a study claim with attention to design, data quality, analysis, and communication, and write a clear design memo stating what a study supports, what it does not, and what evidence would strengthen it.
Weekly rhythm
This is a lecture/activity course meeting three days a week, and each day has a settled role:
- Monday — design concept + checkpoint. We introduce a design or evidence concept, work through examples, and complete a short checkpoint near the end of class.
- Wednesday — case, diagram, or analysis day. We apply the week’s idea to a study, dataset, causal diagram, survey scenario, or experimental-design problem.
- Friday — quiz / critique day. A short quiz on recent material, then a study-claim critique, a design revision, or work on a design memo.
This rhythm is the plan; the authoritative weekly schedule, including any shifts, lives in Blackboard. Attendance is not graded directly, but checkpoints, quizzes, labs, and in-class activities happen during class, so consistent attendance matters.
Assessment shape (indicative — not a contract)
The table below conveys the relative emphasis of each graded category in qualitative terms only. It is not a grading contract: it carries no percentages and no point values, and the actual weights live in Blackboard.
| Category | Rough emphasis |
|---|---|
| Design checkpoints | small |
| Weekly quizzes | small |
| Design memos and homework | the largest single category |
| Applied design labs | moderate |
| Midterm | moderate |
| Final design project | moderate |
| Final exam | small |
Read this as a picture of where the weight sits: regular design memos and homework are the backbone of the grade, the labs and the two big design instruments (midterm, project) carry moderate weight, and the smaller pieces — checkpoints, quizzes, and the final — add up around them. For the exact weighting, consult Blackboard.
Software and reproducibility
We use R (through RStudio or Posit Cloud) and Quarto, along with spreadsheets and browser-based tools, for randomization activities, blocking and sampling demonstrations, and adjustment. Computation supports the reasoning rather than replacing it: you read, edit, and run code; interpret output; and write short written conclusions. Software output does not rescue a weak design — you are expected to explain what the data- production process allows you to claim. Setup instructions are on the Causal-diagram guide and the resource pages.
AI use (summary)
Generative AI tools may be used as a study and workflow aid — to explain a concept a second way, generate practice questions, suggest possible threats to validity, check your understanding of a causal diagram, or help debug code used in a lab. They may not produce work you submit as your own, complete study critiques, write design memos, invent study limitations, fabricate sources, or replace your responsibility to understand and verify the reasoning, and they are prohibited on quizzes and exams unless explicitly allowed.
Because design and causal evidence are highly sensitive to assumptions, AI output must be checked carefully: many incorrect design explanations come from confusing random sampling with random assignment, overstating causal claims from observational data, ignoring measurement error, adjusting for the wrong variable, missing a selection problem, or treating statistical significance as proof that a design is strong. Whenever you use AI on a graded written assignment, lab, code submission, or project component, include a brief AI Use Note with three labeled lines:
| Field | What to record |
|---|---|
| Tool | which assistant you used (with approximate date or version) |
| Purpose | what you used it for |
| Verification | how you checked, tested, revised, or validated the output |
Verification is the load-bearing line: redraw the causal diagram yourself, confirm whether a variable is pre- or post-treatment, compare the claim to the actual sampling and assignment mechanism, check a calculation, verify a source, or rewrite the explanation in your own words after checking it. The full policy lives in Blackboard.
Materials
You will need:
- Instructor notes, examples, and design guides — the primary course materials, posted on this site and in Blackboard.
- Introduction to Modern Statistics, 2nd ed. (Çetinkaya-Rundel & Hardin) — a free supplement, CC BY-SA 3.0, at openintro-ims.netlify.app, used for study design, sampling, and confounding.
- Statistical Inference via Data Science: A ModernDive into R and the Tidyverse, 2nd ed. (Ismay, Kim & Valdivia) — a free supplement, CC BY-NC-SA 4.0, at moderndive.com/v2, used for randomization and sampling simulations.
- R, RStudio or Posit Cloud, and Quarto (plus spreadsheets or browser tools) — for the design labs.
- Blackboard (the LMS) — for all graded work, dates, and announcements.
- A non-graphing scientific calculator for in-class quizzes; most computation is done in R.
Not used in this course: Cengage, WebAssign, MyLab, or any paid homework platform.
Where things live
Keep the two homes of the course straight:
- Blackboard (the LMS) is the operational home: graded design checkpoints, quizzes, design memos and homework, applied design labs, the midterm, the final design project, the final exam, all due dates, all submissions, and all grades. It is authoritative.
- This public site is the public notes home: weekly notes, labs as study material, resources, and orientation pages. It is ungraded.
This companion exists to orient you. Whenever a graded specific is at stake — a weight, a deadline, a policy — go to Blackboard, which is authoritative.