Week 2 · Lecture outline

Week 2 — Lecture Outline · Sociological Research Methods

Introduction to Sociology · SOC 1 Fall 2026 · Prof. Adeyemi Fictional sample

Course: Introduction to Sociology (SOC 1) · Silver Oak University (fictional sample) · Prof. Adeyemi
Objectives covered: Objective 2 — Describe the major sociological research methods and read social data critically, distinguishing correlation from causation.
SLOs touched: A (apply theory — positivist vs. interpretivist traditions; methods serve all three perspectives) · B (read and evaluate evidence; correlation vs. causation) — this is the home week for SLO B
Meeting pattern: 2 sessions × 75 min = 150 min. Segment minutes below total ~150; scale to your own pattern.

Week at a Glance


The week's big question	"How do sociologists turn a curiosity about society into trustworthy evidence — and what is the single most expensive mistake in all of social science?"
By the end of the week, students can…	(1) walk the research cycle and define operationalization, independent/dependent variables; (2) match the four methods (survey, experiment, field/ethnography, secondary data) to their uses and say why the experiment is built for causation; (3) tell reliability from validity and a population from a sample, and explain why representativeness beats size and why a self-selected sample can't be generalized; (4) explain correlation vs. causation, the third variable, and the spurious correlation; (5) read a real social statistic critically and state the core research ethics (informed consent, confidentiality, IRB).
Key vocabulary	research cycle, hypothesis, operational definition/operationalization, variable, independent vs. dependent variable, survey, questionnaire/interview, experiment, control group, random assignment, field research/ethnography, participant observation, secondary (existing-data) analysis, reliability, validity, population, sample, representative sample, probability/random sampling, generalizability, margin of error, sampling bias, self-selection, correlation, causation, spurious correlation, third/confounding variable, positivism, interpretivism, verstehen, qualitative vs. quantitative, research ethics, informed consent, confidentiality, IRB
Materials	slides (Deck 2), the week's readings + video links, one approved chatbot (Gemini / Claude / ChatGPT) for the AI-critique moment and the tutorial
Timing note	8 segments, ~150 min total. Session 1 = Segments 1–4 (~75). Session 2 = Segments 5–8 (~75).

Segment 1 — Hook & the Promise (8 min) · Session 1 opens

Hook. Put two headlines on a slide: "Study: kids who eat breakfast get better grades" and "Poll: 80% of our readers oppose the new law." Have the room vote on which one they'd trust. Then the turn: "A sociologist doesn't ask 'is that true?' first. She asks 'how do you know?' — Who was studied? How were they chosen? What was actually measured? And does this show a cause, or just two things moving together?"

Then the payoff line: "Last week we said sociology runs on evidence, not anecdote. This week is how we get the evidence — and the one habit that separates a careful reader from a sucker: telling a correlation from a cause."

The promise (write it on the board): "By Friday you'll be able to walk the research cycle, match a method to a question, tell reliability from validity and a population from a sample, know why a good sample beats a big one, and explain why correlation is not causation."

Why it matters line (memory hook): "Anyone can say 'a study says.' A sociologist asks how the study knows."

Segment 2 — The Research Cycle & Operationalization (18 min)

Plain language first. Sociology is a science, so it follows a research cycle:
1. Question — a clear question about social life ("Does part-time work affect students' grades?").
2. Hypothesis — a testable prediction about a relationship between variables ("More work hours are associated with lower grades").
3. Operationalization — the key step: turn an abstract concept into something concrete and measurable (an operational definition). E.g., "work" = paid hours per week; "grades" = term GPA.
4. Method — choose how to collect data (Segments 3–4).
5. Data — collect it.
6. Analysis — analyze and report, so others can check and replicate.

Two terms to lock in now (a classic quiz pair):
- Independent variable = the presumed cause (what you change or compare — e.g., work hours).
- Dependent variable = the effect that depends on it (e.g., GPA). "The dependent variable depends on the independent one."

Why operationalization matters. A concept like "social isolation" or "religiosity" means nothing to a study until you pin down how you'll measure it. A good operational definition is one two different researchers would apply the same way.

Memory hook (put it on a slide):

"Operationalize = define it so two strangers would measure it the same."

Segment 3 — Two Methods: Surveys & Experiments (18 min)

Plain language first — the first two of four methods (one-line picture each):
- Survey — ask a sample of people standardized questions (questionnaires or interviews). Strength: measures attitudes/behaviors across many people; can generalize if the sample is good. Limit: usually shows correlation, not causation; self-report can be biased.
- Experiment — manipulate an independent variable under controlled conditions and compare groups. This is the one method built to establish CAUSATION, and it does so with two features: a control group (for comparison) and random assignment of subjects to groups (which evens out other differences). Can be lab or field. Limit: artificial settings; many social questions can't ethically be experimented on.

Land the key idea (preview of Segment 6): "If you want to know whether X causes Y, the experiment is your best tool — control + random assignment are what let you rule out other explanations."

Segment 4 — Two More Methods + Reading a Statistic (the data move) (16 min) · Session 1 closes (~75)

Plain language first — methods three and four:
- Field research / ethnography — observe people in their natural setting, often as a participant-observer, to understand social life and meaning from the inside. Strength: depth, context, the participants' own meanings (the interpretivist tradition). Limit: hard to generalize; the observer can affect what's observed.
- Secondary (existing-data) analysis — analyze records or data someone else already collected (government statistics, historical documents, prior surveys). Strength: efficient; great for trends over time. The classic example, named factually: Émile Durkheim's Suicide (1897) used existing official records to show that suicide rates track social integration — the imagination proven with secondary data.

A short read-the-data walkthrough (the move you'll do every Workshop) — with a figure verified at the source:

Put one real statistic on a slide: "Per the Pew Research Center Mobile Fact Sheet (data from a 2025 survey of U.S. adults), 91% of U.S. adults own a smartphone — up from 35% in 2011." Walk the class through the four questions you ask of any social statistic:
1. What is measured? Self-reported smartphone ownership, from a survey (not a head count of every adult).
2. Over what population and period? U.S. adults, 2025; with a margin of error (about ±1.9 points).
3. What does it show — and what does it not? Ownership is now near-universal but uneven (≈97% of adults under 50 vs. ≈78% of those 65+). It does not tell us why, or what people do with the phones.
4. Correlation or causation? Younger and higher-income adults own more — a descriptive pattern/correlation, not proof that income causes ownership. (Figure pre-verified live at pewresearch.org; cite Pew + 2025.)

Segment 5 — Sampling: Why a Good Sample Beats a Big One (20 min) · Session 2 opens

Hook back in: "Last session: how we gather data. Today, the question that sinks more 'studies' than any other — who did you actually ask, and how did you choose them?"

Plain language first.
- Population = the whole group you want to know about (e.g., all U.S. adults). Sample = the smaller subset you actually study.
- The goal is a representative sample — one whose makeup mirrors the population — so you can generalize from sample to population within a margin of error.
- The gold standard is a probability (random) sample: every member of the population has a known, non-zero chance of being selected.

Kill the big misconception (this is a named distractor):
- ❌ "A bigger sample is always better."
✅ Cure: representativeness beats size. The classic lesson (named factually): a 1936 U.S. presidential-election magazine poll drew millions of responses but used a biased sample (drawn largely from telephone and automobile owners) and predicted the wrong winner — while a much smaller scientific sample got it right. "Size can't fix bias." (We name the principle, not a precise figure.)

Memory hook: "A representative sample beats a big one. Size can't fix bias."

Segment 6 — Sampling Bias, the Three Traditions & One Question Three Ways (18 min)

The everyday trap — sampling bias / self-selection:
- A self-selected sample (an online click-in poll, a call-in survey, "our readers") lets people opt in. Those who opt in differ systematically from those who don't — so the result can't be generalized, no matter how many respond. "50,000 of our readers said…" is not evidence about the country.

Run the three-perspective move (the spine of the course) — methods serve all three, but the key contrast this week is two research traditions:
- Positivist tradition (associated with Durkheim): seek objective, often quantitative evidence — rates, distributions, law-like patterns about social facts. Surveys and secondary-data analysis fit here.
- Interpretivist tradition (associated with Weber's verstehen — interpretive understanding): seek the meaning social action has for the people involved, usually through qualitative methods like ethnography.
- (And the three Week-1 lenses still apply: a functionalist might survey how an institution maintains order; a conflict theorist might ask who funded a study and whose interests its questions serve; an interactionist might do ethnography to capture meaning.)

One question, three ways (do this out loud) — take: why do some students drop out?

Positivist / quantitative (survey or existing data): look for rates and predictors — which factors correlate with dropping out across thousands of students.

Interpretivist / qualitative (ethnography/interviews): embed in students' lives to understand how dropping out feels and what it means to them.

Experiment (where ethical/possible): test whether a specific support program causes higher persistence, using a comparison group.

Land it: "No method is 'more scientific' than the others — they answer different questions, and they're strongest combined."

Misconception + cure:
- ❌ "Numbers are real science; talking to people isn't."
✅ Cure: quantitative and qualitative methods answer different questions (how many / what pattern vs. what meaning). The best sociology often uses both.

Segment 7 — Correlation vs. Causation & Research Ethics (20 min)

The load-bearing rule (the most important idea of the week):
- A correlation = two variables move together. Causation = one variable actually produces a change in the other — a much stronger claim.
- A correlation is consistent with three stories: X→Y, Y→X (reverse direction), or a third (confounding) variable causing both (a spurious correlation).
- To claim causation you generally need: (1) association, (2) correct time order (cause precedes effect), and (3) elimination of plausible third variables — which is exactly what experiments are designed to do.

Name the misconception + cure (the correlation-vs-causation beat):
- ❌ "Two things move together, so one causes the other."
✅ Cure (signature example): ice-cream sales and drowning deaths rise and fall together — not because ice cream causes drowning, but because a third variable, hot summer weather, drives both (more ice cream and more swimming). A spurious correlation. "Correlation is a clue, not a verdict."
- ❌ "More police → more reported crime, so police cause crime."
✅ Cure: a third variable (city/population size) drives both, and more officers may simply record more crime — the arrow may even run in reverse. Direction and confounders are unaddressed.

Research ethics (named factually, non-graphic): studying humans carries obligations — informed consent (told the purpose/risks, freely agree), confidentiality/privacy, avoiding harm, and review by an Institutional Review Board (IRB) before the study begins. Three touchstones, named as ethics lessons, not for shock:
- The Tuskegee Syphilis Study — U.S. researchers withheld treatment from Black men for decades without informed consent; a central reason today's consent rules exist.
- The Milgram obedience experiments and Zimbardo's Stanford prison study — raised lasting questions about deception and psychological harm.

Memory hook: "Correlation is a clue, not a verdict — hunt the third variable. And: consent, confidentiality, no harm, IRB first."

Segment 8 — Technology Workflow + AI-Critique, Callback & Hand-off (12 min) · Session 2 closes (~75)

Technology workflow — verify at the source:
1. Real figures live at data portals: the U.S. Census / QuickFacts, Pew Research Center, the Bureau of Labor Statistics (BLS), the World Bank, Our World in Data.
2. The habit: never repeat a number you haven't seen on the source's own page — and note the year and what's measured.
3. Before believing any claim, run the four questions and the correlation-vs-causation check.

AI-critique moment (students verify, not consume) — this is the heart of this week's Workshop:

Paste this to an approved chatbot: "Give me a statistic about social-media use and a study showing it causes loneliness, with the citation."
Then check its work against a real source:
- Did it invent a statistic or a plausible-sounding "study" or citation? Chatbots fabricate these constantly — this week's load-bearing risk. Search for the number/study at the source; if you can't find it, treat it as fabricated.
- Did it slide from correlation to causation — reporting that social-media use correlates with loneliness as if it proved a cause (ignoring reverse direction and third variables)?
- Did it overgeneralize from a biased or self-selected sample?
Your job all semester: the tool drafts, you verify. This is exactly how the weekly Sociology Workshop's AI-critique step works.

Callback + tease:
- Callback: "This week is the backbone of SLO B. Sociology earns trust by being explicit about how it knows — the research cycle, matching method to question, measuring reliably and validly, sampling representatively, and never mistaking a correlation for a cause."
- Tease next week: "Now that we can gather and read evidence, we turn it on culture — the values, norms, and symbols that hold a society together and that we mostly never even notice."

Hand-off (the week's graded work):
- Lecture Tutorial 2 (AI tutor, share-link submission) — the research cycle, the four methods, reliability/validity, sampling, correlation vs. causation.
- Quiz 2 (end of week) and Discussion 2 ("What Does This Statistic Actually Show?").
- Assignment 2 — "Critique the Study": classify methods, sort reliability/validity, diagnose a causal-claim flaw, and build a short evidence-based argument.
- Workshop 2 — "Read the Data": read a real Pew statistic with the four-question scaffold, run a correlation-vs-causation drill, then catch an AI's reasoning slips.

Instructor FAQ — Common Stumbles

Student says / does	Quick cure
Confuses reliability and validity.	Reliable = consistent (same answer twice); valid = accurate (measures the real thing). The scale that's always 5 lb high is reliable but not valid.
"A bigger sample is always more accurate."	Representativeness beats size. A huge biased sample (1936 poll) is worse than a small random one. Size can't fix bias.
Trusts an online click-in poll ("50,000 readers!").	That's a self-selected sample — people opt in and differ systematically from those who don't, so it can't be generalized.
Slides from correlation to causation ("X and Y move together, so X causes Y").	A correlation has three possible stories (X→Y, Y→X, or a third variable). To claim a cause you need time order and the third variables ruled out — what experiments do.
Mixes up independent and dependent variables.	The dependent variable depends on the independent one. IV = presumed cause; DV = effect.
Thinks numbers = real science and ethnography = "just talking."	They answer different questions (how many / what pattern vs. what meaning) — positivist vs. interpretivist (verstehen). Best work uses both.
Calls every study an experiment.	Only an experiment manipulates a variable with control + random assignment. A survey describes/associates; it usually shows correlation, not cause.
Treats a chatbot's statistic or "study" as a source.	Chatbots fabricate figures and citations. Never repeat a number you haven't seen at the source (Census, Pew, BLS, World Bank).

Scope flag

This outline stays within Objective 2 (the research cycle; the four major methods; reliability/validity; population/sample and sampling bias; correlation vs. causation; research ethics; reading a social statistic). It is not a statistics course — there's no formula work, no significance testing, no regression; we build conceptual data literacy. The studies named (Durkheim's Suicide 1897; the 1936 election-poll lesson; the Tuskegee study; Milgram; Zimbardo) are referenced factually, briefly, and non-graphically as part of the discipline's real history and its ethics record. The one published figure used (the Pew 91% smartphone-ownership statistic, 2025) was verified live at the source before shipping. The instructor and institution remain fictional.

~ Prof. Adeyemi's edition · Fall 2026 · built with thecoursemaker.com