Week 2 Sociology in Action Workshop: Two Polls and Three Neighborhoods

Part 1 — The Big Picture

This week you learned that a number is only as good as the way it was gathered. A sample speaks for a population only if it was chosen fairly, a pattern is not a cause, and places of different sizes are compared by rate, never by count. In this workshop you do that reading yourself on a small, made-up city. First you compare two polls about a new park: one that anyone could answer on the city's website and one drawn at random from every household. Then you turn three neighborhoods' noise-complaint counts into rates per 100,000 residents and decide what those rates can and cannot say. Finally you ask a chatbot the same questions and catch where it goes wrong, the way you will check any figure you meet in the rest of this course.

Time budget: Plan for about 90 minutes.

Part 2 — The Task

Guiding question: Which of two park polls can speak for this city's residents, and what can three neighborhoods' complaint rates tell you, and not tell you, about the neighborhoods and the people who live in them?

Workshop data: two park polls and three neighborhoods' noise complaints (illustrative)

Source: An illustrative dataset constructed for this workshop, not from a real city: two hypothetical polls about a new park and one year of hypothetical noise complaints in three neighborhoods.

• Rows: 5 rows of data: 2 polls and 3 neighborhoods (the scaffold shows them in 2 tables of 2 and 3 rows)
• Poll columns: how the poll was gathered; responses; yes answers
• Neighborhood columns: neighborhood; complaints in one year; residents

Part 3 — Materials & Procedure

Tools and materials

• Free public data tables (Census Bureau, Bureau of Labor Statistics, Pew Research Center, World Bank, Our World in Data, General Social Survey) plus a spreadsheet (Google Sheets or any free spreadsheet)
• A document or notebook for your written answers

Ways of working

• apply-a-perspective-to-a-case
• read-and-interpret-social-data
• structured-observation-and-reflection

Procedure

1. Open a blank sheet. In row 1 type the headers Poll, Responses, and Yes answers in columns A, B, and C. In rows 2 and 3 copy the two polls from the first table in Part 4. Tool: Free public data tables (Census Bureau, Bureau of Labor Statistics, Pew Research Center, World Bank, Our World in Data, General Social Survey) plus a spreadsheet (Google Sheets or any free spreadsheet). Check: Row 2 holds the website poll and row 3 holds the random-sample poll.
2. In D1 type Yes share (%). In D2 type =C2/B2*100 and copy it to D3. Copy both shares into the poll table, then in the results table subtract the smaller share from the larger and label the difference in percentage points. Tool: Free public data tables (Census Bureau, Bureau of Labor Statistics, Pew Research Center, World Bank, Our World in Data, General Social Survey) plus a spreadsheet (Google Sheets or any free spreadsheet). Check: Each share falls between 0 and 100, and the difference is labelled in percentage points, not percent.
3. In row 6 type the headers Neighborhood, Complaints, and Residents in columns A, B, and C. In rows 7 to 9 copy the three neighborhoods from the second table in Part 4. Tool: Free public data tables (Census Bureau, Bureau of Labor Statistics, Pew Research Center, World Bank, Our World in Data, General Social Survey) plus a spreadsheet (Google Sheets or any free spreadsheet). Check: Rows 7 to 9 hold North, South, and East.
4. In D6 type Rate per 100,000. In D7 type =B7/C7*100000 and copy it to D8 and D9. Copy each rate into the neighborhood table and rank the neighborhoods by rate and, separately, by count. Tool: Free public data tables (Census Bureau, Bureau of Labor Statistics, Pew Research Center, World Bank, Our World in Data, General Social Survey) plus a spreadsheet (Google Sheets or any free spreadsheet). Check: Each rate is a whole number, and the neighborhood with the most complaints is not the one with the highest rate.
5. Answer the analysis questions in Part 5 in your document, using only the numbers in your tables. Check: Every answer points to a value in your tables or to a definition from this week.
6. Do the AI-critique step in Part 6, or the equivalent route printed there, and write the 3–5 sentence report. Check: Your report quotes at least two errors and corrects each with evidence.

Fallback route

When to use it: Use this route if no spreadsheet is available.

Work the same steps by hand or with a calculator. Divide each poll's yes answers by its responses and multiply by 100 to get each share, then subtract the smaller share from the larger. Divide each neighborhood's complaints by its residents and multiply by 100,000 to get each rate. Write every division in your document so it can be checked, complete every scaffold table, answer every analysis question, do the critique step, and submit the same items for equal credit.

This route earns the same credit as the primary route.

Part 4 — The Scaffold

These are the workshop's two illustrative park polls; compute each yes share so you can compare how they were gathered, not only how big they are.

Student scaffold table 1

Poll	How it was gathered	Responses	Yes answers	Yes share (%)
Website poll	Anyone visiting the city's website could answer	1,200	900
Random-sample poll	400 households drawn at random from a list of every address	400	220

These are the workshop's illustrative noise-complaint counts for one year; turn each count into a rate so the neighborhoods can be compared fairly.

Student scaffold table 2

Neighborhood	Complaints in one year	Residents	Rate per 100,000 residents	Rank by count	Rank by rate
North	18	12,000
South	30	40,000
East	24	20,000

Record your results and your reasoning here so every interpretation rests on numbers you produced yourself.

Student scaffold table 3

Quantity	Your value or words	How you found it
Gap between the two yes shares, in percentage points
Poll that can speak for the city's residents, and why
Neighborhood with the most complaints
Neighborhood with the highest rate
One reason complaints might not measure noise

Part 5 — Analysis Questions

1. Report each poll's yes share and the gap between them in percentage points, and show the division you used for each share.
2. Which poll can speak for the city's residents? Name the population, each poll's sample, and the sampling bias in the other poll, and say whether more website responses would fix it.
3. Which neighborhood has the most complaints, and which has the highest rate per 100,000 residents? Explain in one or two sentences why the two rankings differ.
4. A council member says East's residents are noisier people. Explain what East's rate does and does not say about the neighborhood and about any one resident, whether it is a correlation, a cause, or neither, and name one research method you would use to learn why the rates differ.
5. Name two limits of these figures, such as that complaints may measure people's willingness to report rather than noise itself, that the data cover one year, or that the city is illustrative, and say how each limit affects what you can conclude.
6. Choose one and answer it in 3–4 sentences: (a) a friend shares the website poll's result as proof that the whole city wants the park; what would you tell them, and what evidence would change your answer? (b) think of a poll or rating you have seen online; who could answer it, and whom can it speak for?

Part 6 — AI-Critique Moment

Choose one equal-credit critique route. Route A: ask an approved AI tool to critique your completed work, then check its response against your own evidence. Route B is the no-AI equivalent printed below.

Approved tools: Claude, Gemini, ChatGPT, Grok, Microsoft Copilot.

Prompt to paste:

I've pasted my completed workshop tables above. An illustrative city asked residents about a new park two ways. A poll on the city's website drew 1,200 responses, and 900 said yes. A research team also asked 400 households drawn at random from a list of every address, and 220 said yes. The same city's three neighborhoods recorded noise complaints in one year: North had 18 complaints among 12,000 residents, South had 30 among 40,000, and East had 24 among 20,000. Please do six things. First, say which poll better describes the city's residents and why. Second, give the gap between the two yes shares in percentage points. Third, compute each neighborhood's complaint rate per 100,000 residents and say which is highest. Fourth, explain why East's rate differs from South's. Fifth, tell me which sociologist introduced the idea of verstehen and what it means. Sixth, tell me what East's rate says about any one East resident.

Test the response for these likely failures:

• Chatbots often add an outside figure or study that is not in the data, such as a claim that online polls are typically off by a stated number of points, with no source.
• Chatbots often credit verstehen to Durkheim or Comte.
• Chatbots often describe who answers online polls, or who lives in East, from a stereotype, such as calling them young renters or louder people.
• Chatbots often say East's higher complaint rate shows that its residents cause more noise, or that the park would reduce noise.
• Chatbots often turn East's neighborhood rate into a statement about any one East resident, such as saying each East resident is more likely to complain.

In 3–5 sentences, say what the chatbot got right, quote at least two sentences it got wrong, and correct each one with evidence from your tables or this week's definitions.

Equivalent route (no AI required)

No AI account is required for this workshop. This route assesses the same objective, Objective 2 (describe and evaluate the major sociological research methods (surveys, experiments, field research and ethnography, secondary-data analysis), distinguish reliability from validity and a population from a sample, and read a social statistic critically — what it measures, over what population and period, what it shows and does not, and whether it is a correlation or a cause), takes comparable effort of about the same 90 minutes, uses the same rubric, and earns the same credit. Instead of the chatbot, critique this flawed analysis, printed here in full: "The city's website poll proves that 75% of residents want the park, and with 1,200 answers it is more accurate than the random sample of 400. A well-known national study shows that online polls are accurate to within 2 percentage points. The idea of verstehen, which Durkheim introduced, means we should trust the numbers over what residents say. People who answer online polls are mostly young renters who want more parks. East has the highest complaint rate because its residents are louder, so the park would cut noise there. Any East resident is more likely to complain than a North resident." Mark every error you can find, quote at least two, and correct each one in 3–5 sentences with evidence from your tables or this week's definitions. Submit the marked-up analysis and your correction in place of the chatbot report and share link.

Part 7 — What to Submit

• All three completed scaffold tables
• Written answers to all six analysis questions
• The chatbot critique report with its share link, or the equivalent route's marked-up flawed analysis

Deliverable:

A short Sociology in Action Workshop report on two park polls and three neighborhoods: three completed tables (the two polls' yes shares and their gap in percentage points, the three neighborhoods' rates per 100,000 residents with both rankings, and your results with reasons), written answers to six analysis questions, and a critique of a chatbot's answers to the same questions, or, on the equivalent route, a marked-up correction of the printed flawed analysis. Submit in any accessible file type or as a link; a recorded response may be submitted as audio or video with a transcript, or as text. Choose how you show it: written answers, a recorded explanation with a transcript, or a labelled diagram of the shares and rates with short notes; every form is graded by the same scoring criteria.

Rubric

The rubric below shows how each submitted part is evaluated; bands are full, partial, or none.

Workshop evaluation criteria

Criterion	Full	Partial	None
Computing the shares, the gap, and the rates	Both yes shares, their gap in percentage points, and all three rates per 100,000 are computed from the workshop's data with the divisions shown, and the gap is labelled in percentage points.	Most values are computed, but one share or rate is missing or wrong, a division is not shown, or the gap is labelled as a percent.	No shares or rates are reported, or the values are invented rather than computed.
Population, sample, and sampling bias	The answer names the city's residents as the population, describes each poll's sample and how it was chosen, names the website poll's sampling bias, and explains that more responses gathered the same way would not fix it.	The answer picks the random-sample poll but leaves out the population, how a sample was chosen, or why size does not fix the bias.	No answer is given, or the bigger poll is chosen because it is bigger.
Reading the rates at the right level	The answer compares rates rather than counts, explains why the two rankings differ, says a neighborhood rate cannot describe one resident, calls the rate a description rather than a cause, and names a fitting research method.	The answer compares rates but either applies the rate to a resident, reads it as a cause, or leaves out the research method.	No answer is given, or neighborhoods are compared by count alone.
Critique and correction	The chatbot critique report or the marked-up flawed analysis quotes at least two errors and corrects each with evidence from the tables or this week's definitions; both routes earn this band equally.	The chatbot critique report or the marked-up flawed analysis names an error but corrects it without evidence, or corrects only one; both routes earn this band equally.	No chatbot critique report or marked-up flawed analysis is submitted, or the output is pasted without any check; both routes are scored the same way.
Limits and reflection	Two real limits are named with their effect on the conclusion, and the chosen reflection option is answered with a reason.	One limit is named, or a limit is named without its effect, or the reflection gives no reason.	No limit or reflection is given.

Quality Gate

• Every share and rate in your tables comes from your own formulas or your own hand calculation from the workshop's data, never from memory or a chatbot.
• Every difference between two shares is written in percentage points.
• Your poll answer names the population and each poll's sample and says how each sample was chosen.
• Your neighborhood answer compares rates, not counts, and never applies a neighborhood rate to one resident.
• No statistic, study, or source is invented anywhere in the submission.
• The critique quotes the chatbot or the flawed analysis and corrects at least two errors with evidence.