Week 10 — Sociology-in-Action Workshop · "What the Census Counts (and What It Doesn't)"

Course: Introduction to Sociology (SOC 1) · Silver Oak University (fictional sample) · Prof. Adeyemi
Objective: Objective 6 — read real Census race & ethnicity demographics: what the self-identified categories measure, what they show, and what they don't · SLO B (reason from evidence) & SLO A (apply theory)
Worth 50 points · Sociology Workshops group = 15% of the grade · Workshop 10
Mode this week: data interpretation (other weeks alternate with observation/reflection workshops). No special tools — just a browser and an approved chatbot.

This is the course's signature weekly component. Every instructional week has one Sociology Workshop. Some weeks you'll read real social data (a chart or table from the Census, Pew, BLS, the World Bank, or Our World in Data); other weeks you'll observe and reflect on your own social world. Either way you'll end by catching an AI's mistakes. All external resources are links — nothing to buy or download.

A note on the topic. This is a charged subject we handle factually and with respect. The point of this workshop is precision about what a demographic number is and isn't — not to "both-side" documented facts (race is socially constructed; measured gaps exist), and never to read a stereotype out of a statistic.

Part 1 — The Big Picture

This week you learned that race is a social construction — the categories are made by societies, have changed over time, and don't line up with biology (there's more genetic variation within so-called racial groups than between them). That raises a sharp question about the most-cited racial data in the country: what is the U.S. Census actually counting when it reports the population "by race"? The answer is the heart of this workshop — and it's stated by the Census itself: its racial categories "generally reflect a social definition of race … and not an attempt to define race biologically," and every response is based on self-identification. So the Census doesn't measure a biological fact about people; it measures how people identify themselves.

The guiding question: When a table reports the U.S. population "by race and ethnicity," what is being measured — and what does that number show, and not show?

Part 2 — The Data (identified, linked, and pre-stated — verified live)

We're reading the "Race and Hispanic Origin" rows from the U.S. Census Bureau. The figures below are from the Census Bureau's Vintage 2023 Population Estimates (population as of July 1, 2023). Open the pages yourself and find these numbers — that habit is the whole point.

Figure A — Two self-identified shares of the U.S. population (July 1, 2023):
- "White alone, not Hispanic or Latino": about 58% of the U.S. population (the Census reports the non-Hispanic White population at roughly 195 million).
- "Hispanic or Latino" (of any race): about 19.5% of the U.S. population (roughly 65 million people) — the second-largest group.
- (Other self-identified groups — "Black or African American alone," "Asian alone," "Two or More Races," "American Indian and Alaska Native alone," and "Native Hawaiian and Other Pacific Islander alone" — make up the rest. For example, the Census reported the non-Hispanic Asian population at about 7.4% in 2023.)

Sources (links only — verified live):
- U.S. Census Bureau QuickFacts: United States (the live "Race and Hispanic Origin" rows you'll read).
🔗 https://www.census.gov/quickfacts/fact/table/US
- U.S. Census Bureau, "About the Topic of Race" (the Bureau's own statement that the categories are a social definition based on self-identification).
🔗 https://www.census.gov/topics/population/race/about.html

Two reading rules before you use any of these numbers.
1. Self-identified, not biological. These are categories people chose for themselves on a survey. They reflect a social definition of race — the Census says so explicitly.
2. Race and Hispanic origin are asked SEPARATELY. On the Census, Hispanic/Latino is an ethnicity asked as a separate question from race, so "Hispanic, of any race" overlaps the race rows — which is exactly why you cannot read the rows as clean, mutually-exclusive "biological bins." (If you want to cite any figure, read it at the source first, with the year — that's the discipline of this course. The QuickFacts table is periodically updated to newer estimate vintages, so confirm the year shown when you open it.)

Part 3 — Read-the-Data Scaffold (fill this in)

Work the scaffold for Figure A (the self-identified U.S. population shares, July 1, 2023). This is the what-is-measured → over-what → what-it-shows-and-not → correlation-or-causation move.

Part 4 — Analysis Questions

Answer in a sentence or two each:
1. In your own words, what does it mean that the Census measures race by self-identification and treats its categories as a social (not biological) definition? Connect it to this week's idea that race is a social construction.
2. Name one thing a population share like "about 58% White alone, not Hispanic" does not tell you (think: within-group diversity? the causes of any disparity? how categories have changed over time?).
3. A friend says, "These Census race numbers prove race is biologically real — there are clear categories right there in the data." Using what the Census itself says about its categories, explain why that conclusion doesn't follow.
4. Pick one of the three perspectives from this week — functionalist, conflict, or interactionist — and explain in two sentences how it would interpret a documented racial gap (say, in homeownership). (Remember: the data describe the gap; the perspectives interpret it — and a gap is not, by itself, a cause.)

Part 5 — AI-Critique Moment (required — this is the BYOAI step)

Now bring in your approved chatbot (Gemini, Claude, or ChatGPT) and be the sociologist who checks its work.

1. Ask it: "Give me the U.S. population shares by race and ethnicity for the most recent year, and explain why one racial group has lower median household wealth than another."
2. Check everything it says against the sources and this week's ideas:
   - Did it invent or misdate a demographic statistic? Verify any share on the Census QuickFacts page (note the estimate year shown). If you can't find a number at the source — or it's tagged to the wrong year — treat it as fabricated or misdated, and say so.
   - Did it treat race as biological, or blur race and ethnicity? Watch for it describing the categories as biological facts, or merging the Hispanic-origin (ethnicity) question with the race question.
   - Did it pin a wealth gap on a stereotype? Watch for any explanation that blames a trait of a group ("because that group is/does X"). That's both unsupported and a stereotype — flag it.
   - Did it slide from a gap to a cause? A documented gap describes a pattern; it does not, by itself, prove a cause. Catch any leap from a number to a causal/moral conclusion.
3. Write 3–4 sentences reporting what the AI got right and at least one thing you had to correct, verify, or flag — an unverifiable or misdated figure, a treat-race-as-biological slip, a blurred race/ethnicity distinction, a stereotype, or a jump from a gap to a cause. (If it happened to get everything right, say how you verified each claim at the source — that's the skill.)

The habit all term: the tool drafts, you judge — and you verify every number at its source. On this topic a careless model will confidently fabricate a percentage or stereotype a group to "explain" a gap. Catching it is the point.

Part 6 — What to Submit

Submit a single document (or text entry) with: your completed Part 3 scaffold, your Part 4 answers, and your Part 5 AI-critique paragraph. Note the source and year for any figure you reference. Due Sunday, Nov 8, 11:59 p.m. (50 points).

Instructor answer key & model responses — REMOVE BEFORE PUBLISHING TO STUDENTS

All figures below were verified live at the source on the build date (2026-06-29). The self-identified U.S. population shares — "White alone, not Hispanic or Latino" ≈ 58% (non-Hispanic White ≈ 195 million) and "Hispanic or Latino, of any race" ≈ 19.5% (≈ 65 million), with non-Hispanic Asian ≈ 7.4% — are from the U.S. Census Bureau, Vintage 2023 Population Estimates (population as of July 1, 2023), and were confirmed against the Census Bureau's released estimates and its "About the Topic of Race" page on the build date. The QuickFacts U.S. table presents these same self-identified "Race and Hispanic Origin" rows and is periodically refreshed to newer estimate vintages — students should read the figure and the year shown at the source. No causal claim and no gap figure is asserted in this workshop; students read any specific gap statistic at its source (Census, Federal Reserve, Pew, BLS).

Model worked scaffold (Figure A — U.S. self-identified shares, July 1, 2023):
- What is measured? Self-identified social categories — people chose a box about themselves on a survey. The Census states its categories reflect "a social definition of race … and not an attempt to define race biologically." So the number measures identification, not a biological fact. It is a count/percentage of a population, not a measure of any trait or outcome.
- Over what population and period? The whole U.S. resident population, estimated as of July 1, 2023 (Vintage 2023 estimates). The categories and questions have changed over the decades (e.g., the option to mark more than one race), so cross-decade comparisons aren't apples-to-apples.
- What it shows / does NOT show: it shows the composition of the population (the relative size of self-identified groups). It does NOT show within-group diversity (e.g., "Asian alone" spans dozens of very different national-origin groups; "Hispanic or Latino" spans many origins and any race), and it does NOT explain any gap between groups or its causes.
- Correlation or causation? No. A composition table is descriptive. If you later saw an income or wealth gap between two groups, this table would not tell you the cause — causes (discrimination, wealth differences, schooling, geography, policy) are structural and must be shown with other evidence.
- Self-identification check: because Hispanic/Latino is an ethnicity asked as a separate question from race, "Hispanic, of any race" overlaps the race rows — so the rows aren't mutually-exclusive "biological bins"; they're self-identified social categories that can combine.

Expected answers:
- Q1: self-identification means people choose their own category; treating those categories as a social definition matches the week's point that race is socially constructed — the Census is recording a social identity, not measuring biology.
- Q2: any correct limit — it doesn't show within-group diversity, doesn't show the causes of any disparity, doesn't capture how categories changed over time, and isn't a biological breakdown.
- Q3: the Census itself says its categories are a social definition based on self-identification, "not an attempt to define race biologically." Clear, self-chosen categories in a table reflect a social classification system, not biological kinds — and the categories have changed over time, which a biological fact wouldn't.
- Q4: Functionalist: might frame disparities through assimilation/cohesion (and is critiqued for treating the dominant culture as the standard and underplaying structure). Conflict: the gap reflects a racial hierarchy and institutional racism reproducing advantage. Interactionist: racial meanings/labels in everyday interaction (and the contact hypothesis). Full credit names the perspective and keeps describe vs. interpret straight.
- Part 5 (AI-critique): full credit for a specific catch — most commonly the AI inventing or misdating a share (unverifiable at the Census), treating race as biological or blurring race and ethnicity, pinning a gap on a stereotype, or jumping from a gap to a cause/verdict. Full credit also if the student verified each figure at the Census (noting the year) and reported how.

Grading rubric — 50 points

Quality gate (self-checked): the published figures asserted in this workshop — the self-identified U.S. population shares for July 1, 2023 (≈58% White alone not Hispanic; ≈19.5% Hispanic of any race; ≈7.4% non-Hispanic Asian) — are from the U.S. Census Bureau, Vintage 2023 Population Estimates, and were verified live at the U.S. Census Bureau on the build date (2026-06-29); source and reference year are stated, and students are directed to confirm the figure and year shown on the live QuickFacts table (which refreshes to newer vintages). No gap figure and no causal claim is asserted — the workshop deliberately reframes any racial-gap number around the skill (read it at the source; the data describe but don't explain), per the discipline's load-bearing rule. The social-construction-of-race framing, the self-identification point, and the race-vs-ethnicity (separate-question) point are all accurate and match the Census Bureau's own statements. The AI-critique explicitly targets fabricated/misdated statistics, treating race as biological, blurring race and ethnicity, stereotyping, and the gap-to-cause leap — the discipline's load-bearing AI risks on this topic. No correlation is presented as causation: every gap is framed as descriptive, and the workshop asserts none. Documented facts (race is socially constructed; gaps exist) are reported plainly and not both-sidesed; competing interpretations are presented evenhandedly.