Week 2 · Chapter

Chapter 2 — Sociological Research Methods

Introduction to Sociology Free sample edition

The module's primary reading · ~35–45 minutes · 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), building on Objective 1 (apply the sociological imagination — connecting personal troubles to public issues — and the three major theoretical perspectives (structural functionalism, conflict theory, symbolic interactionism) to interpret a social phenomenon, naming the level of analysis each works at and placing the discipline's founders with the ideas and works they are known for) · Read it first this week; the lecture, the slide deck, the tutorial, the quiz, and the Sociology in Action Workshop all build on it.

In plain language

Sociologists make claims about whole groups of people, so they need fair ways to check those claims. This week you learn the main ways they gather evidence, from asking thousands of people the same questions to living inside one small group for months. You also learn when a pattern earns the word "cause." A correlation (two things that tend to change together) is a clue about a cause, but it is not proof of one. Finally, you learn the rules that protect the people who agree to be studied.

Why you already care about this

In 2025, Pew Research Center asked 5,022 U.S. adults whether they own a smartphone. Ninety-one percent said yes. Pew used those answers to describe every adult in the country, nearly all of whom were never asked. How can a few thousand answers speak for a whole country? And what, exactly, can that one number tell you?

You meet claims like this every day. A headline says people who use a certain app sleep worse. A coworker forwards a poll that "proves" most people agree with her. A group chat shares a chart about crime in two towns. Each claim sounds like evidence. This chapter gives you the questions to ask before you believe it: how was it gathered, whom does it speak for, and does it show a cause or only a pattern?

Sociologists ask those questions of their own work first, and by the end of this week you will too. You will define what you measure before you collect anything. You will pick a method that fits your question. You will check whether your sample can stand in for the larger group. And you will protect the people who make the research possible.

By the end of this chapter you will be able to:

turn a hunch into a hypothesis and name its independent and dependent variables;
match a research question to the method built for it: a survey, an experiment, field research, or secondary data analysis;
tell a reliable measure from a valid one;
explain why a random sample can beat a much bigger self-selected one;
say whether a pattern is a correlation or a cause, and name a third variable;
name the ethics rules every study of people must follow;
read a real survey figure four ways and report a change in percentage points and in percent.

Carry this sentence with you: a pattern is a clue, not a verdict — until a design earns "cause."

Where we are

Already covered: Week 1, The Sociological Imagination & Doing Sociology, taught you to connect personal troubles to public issues, to run one event through three perspectives, and to ask four questions of any statistic.
This week: how sociologists gather evidence, what each method can and cannot claim, and the ethics rules that protect the people they study.
How they connect: Week 1 said sociology tests its claims against evidence instead of trusting common sense; this week shows how that testing works.
Where we're going next: Week 3, Culture, studies the shared values, norms, and symbols of a group, and you will read every claim about culture with this week's methods in hand.

Week 1 is the chapter this one leans on. Recall the four questions you asked of the two illustrative towns: what is measured, over what population and period, what the figure shows and does not, and whether it is a correlation or a cause. This week you learn where those answers come from. Recall also the difference between a count and a rate. A count is a number of events; a rate divides that count by a population, so places of different sizes can be compared. Both ideas return in Section 8.

1. Turn a hunch into a study: the research cycle

Picture a vacant lot that neighbors turn into a community garden. A year later, one gardener tells you, "Since the garden opened, everybody here knows each other." That is a hunch. It might be right. But "knows each other" is fuzzy, and one gardener is one point of view. How would you check it?

You check a hunch by running it through the research cycle.

The research cycle is the ordered steps of a study: ask a question, state a hypothesis, define the variables so they can be measured, choose a method, collect data, and analyze them to see whether the hypothesis holds.
A hypothesis is a testable statement of the relationship a researcher expects to find between two or more variables, written before the data are collected.
A variable is a characteristic that can take different values from one person, group, or place to another, such as age, income, or the number of neighbors a person knows.

The most important step is easy to skip. Before you collect anything, you must operationalize your variables: define each one as an exact, observable measurement, such as counting the neighbors a resident can name on a questionnaire, so that different researchers can measure it the same way. "Knows each other" becomes "the number of neighbors on the block a resident can name." Now two researchers can measure it and get the same kind of answer.

Your hypothesis usually links two kinds of variable.

The independent variable (IV) is the presumed cause in a study: the variable that an experiment changes on purpose, or that a researcher expects to come first and shape another variable.
The dependent variable (DV) is the presumed effect in a study: the variable a researcher measures to see whether it changes along with the independent variable.

Worked example — the illustrative community-garden study. Walk the gardener's hunch through all six steps.

Question. Do community gardens build ties between neighbors?
Hypothesis. Residents who belong to the garden can name more of their neighbors than residents who do not.
Operationalize. Garden membership is yes or no, taken from the garden's sign-up list. Ties are measured as the number of neighbors a resident can name.
Method. A short questionnaire handed to every household on the block.
Data. Each resident's membership and their count of named neighbors.
Analysis. Compare the two groups' counts.

Here the IV is garden membership, because the hypothesis treats it as the presumed cause. The DV is the number of neighbors a resident can name, because that is the presumed effect you measure. In words: the study asks whether joining the garden comes with knowing more neighbors.

Misconception check — the swapped variables. Students often think the dependent variable is "the one that comes first," because "dependent" sounds like "the thing everything depends on." It is the reverse. The dependent variable depends on the independent one; it is the outcome you measure. The cure: ask "which one would change if the hypothesis is right?" That one is the DV.

2. Match the question to a method

A city bus agency, a sociologist studying a late-night diner crew, and a researcher reading old official records all want evidence. They do not use the same tool. Each research method is built for a different kind of question, and the first skill is matching the two.

The four major methods, and what each is for:

A survey is a research method that gathers data by asking many people the same questions, through a questionnaire or an interview, and is best suited to describing a large group. Its strength is breadth.
An experiment is a research method in which the researcher changes one variable for an experimental group, leaves it unchanged for a control group, and compares the two groups to test a cause. Its strength is cause.
Field research is a research method that collects firsthand data in the natural settings where people live, work, and play, rather than in a laboratory or through a questionnaire. Ethnography is a form of field research in which the researcher joins a group's daily life over a long period to understand its world from the inside. Their strength is depth.
Secondary data analysis is a research method that works with data someone else already collected, such as Census tables or official records, to answer a new question. Its strength is reach across time and place.

An experiment has two groups with names of their own.

The experimental group is the people in an experiment who receive the treatment, the change whose effect the researcher wants to test.
The control group is the people in an experiment who do not receive the treatment but are otherwise handled the same way, giving the comparison that isolates the treatment's effect.

Secondary data analysis is as old as the discipline. Émile Durkheim's Suicide (1897) worked from official statistics that already existed. He showed that suicide rates differed in regular ways between social groups. In the societies he compared, for example, Protestants had higher rates than Catholics, and unmarried people had higher rates than married people. He read those rates as social facts about how strongly groups bind their members. It is the discipline's founding use of existing data.

Worked example — four questions, four methods. Match each illustrative question to the method built for it.

"What share of the city's bus riders feel safe at night?" You need many answers from a large group, so you use a survey.
"Does a text-message reminder make library patrons return books on time?" You want a cause, so you run an experiment: the experimental group gets the reminder, the control group does not, and you compare on-time returns.
"How does a late-night diner crew keep each other going through a long shift?" You want depth and meaning, so you do field research, and if you work alongside the crew for months, ethnography.
"Has the share of one-person households changed over decades?" The Census already collected that, so you use secondary data analysis.

In words: breadth calls for a survey, cause calls for an experiment, depth calls for field research, and questions about the past or about huge populations often call for existing data.

Misconception check — the empty control group. Students often think the control group is "the group that nothing happens to," a group that barely matters. It is tempting because the control group gets no treatment. But the control group is the comparison. Its members are handled the same way in every other respect, so any difference between the two groups can be traced to the treatment. Without it, you would have nothing to compare the reminders against.

3. Check the measuring stick: reliability versus validity

Step on a bathroom scale three times in a row. If it reads the same number each time, you trust it. But suppose that scale has always read two pounds heavy. It is consistent, and it is wrong every time.

That scale separates two ideas that people often blur.

Reliability is the consistency of a measure: a reliable measure gives the same result each time it is used on the same thing under the same conditions.
Validity is whether a measure captures what it is meant to capture; a measure can be consistent and still miss its target, so reliability does not guarantee validity.

Worked example — the community-involvement question. Suppose an illustrative survey wants to measure how involved residents are in their neighborhood. It counts the number of clubs a resident has joined.

Is it reliable? Probably. Ask the same resident twice in a week and the count will likely match.
Is it valid? Weaker. A resident who has joined no club may still check on an elderly neighbor every day, coach a pickup game, or organize a block cleanup. Club counts miss that kind of involvement.
A better measure. Add questions about helping neighbors, attending local meetings, and volunteering. The measure now covers more of what "involvement" means.

In words: the club count is consistent but narrow, so it is reliable without being fully valid.

Here is a caveat worth keeping. Reliability is necessary for validity, but it is never enough. A measure that changes wildly from one day to the next cannot be valid. A measure that stays the same every day can still be aimed at the wrong thing.

Misconception check — the swapped pair. Students often swap the two words, calling the always-two-pounds-heavy scale "valid because it's consistent." The words sound alike, so the mix-up is natural. The cure is to tie each word to a question. Reliable asks, "Do I get the same reading twice?" Valid asks, "Am I measuring the right thing?" The scale says yes to the first and no to the second.

4. Decide whom you can speak for: population and sample

Picture a sports team's fan page that posts a poll: "Should the city build a new stadium?" By evening it has thousands of votes, and most say yes. Does that show what the city thinks?

To answer, you need two words, and you will use them all term.

A population is the whole group of people a researcher wants to draw conclusions about, such as all U.S. adults.
A sample is the smaller set of people drawn from a population who actually take part in a study.

The fan-page poll has a large sample, but of whom? Only people who follow the team's page, saw the post, and chose to vote. That is a self-selected sample. A self-selected sample is a sample made up of people who chose to respond, such as visitors who answer an online poll; it represents only the kind of person who opts in and cannot be generalized to a population.

The fix is a random sample: a sample chosen so that every member of the population has a known, equal chance of selection, which makes the sample representative and lets results be generalized to the population. A random sample of a few hundred city residents includes people who never visit the fan page, people who dislike sports, and people who never answer online polls.

The problem with the fan-page poll has a name. Sampling bias is a systematic difference between a sample and the population it is meant to represent, caused by how the sample was chosen; collecting more responses the same way does not remove it.

Worked example — two illustrative stadium polls. Compare two ways of asking.

Poll A. The fan page collects thousands of votes from anyone who clicks. Population the poll claims to describe: city residents. People it reached: fans who chose to vote.
Poll B. A research team draws a few hundred residents at random from a list of every household and asks the same question.
Which one speaks for the city? Poll B. Every household had a known, equal chance of being picked, so the sample looks like the city in miniature. Poll A's sampling bias is built in, and another thousand fan votes would only repeat it.

In words: a representative random sample beats a large self-selected sample, because size cannot fix the way a sample was chosen.

Misconception check — bigger is always better. Students often think a bigger sample is always better. It is tempting because more answers feel like more evidence. But a self-selected sample that grows only gets better at describing the people who opt in. The cure: before you ask "how many?", ask "how were they chosen?" Representative beats big.

5. Earn the word cause: correlation versus causation

Every summer in a beach town, two numbers climb together: ice-cream sales and drownings. Does ice cream make people drown? No. Something else is going on.

These two words carry the whole section, and you will meet them in every week that reads data.

A correlation is an association between two variables that change together, so that higher values of one tend to come with higher, or lower, values of the other.
Causation is the stronger claim that a change in one variable produces a change in another; it is supported only by an experiment with random assignment or a study designed to identify a cause.

The beach town shows the most common trap.

A third variable is an outside factor that affects both variables in a correlation, so the two move together even though neither one produces the other.
A spurious correlation is an association between two variables that is created by a third variable, such as ice-cream sales and drownings both rising in summer because of the heat.

Heat sends more people to the ice-cream stand. Heat also sends more people into the water, where some drown. Heat is the third variable, and the ice-cream link is spurious.

How does a design earn the word "cause"? The experiment from Section 2 does it with one more ingredient. Random assignment is placing each person in an experiment into the experimental group or the control group by chance, so the groups start out alike and a later difference can be traced to the treatment. If library patrons are assigned to get a reminder by a coin flip, then busy patrons, careful patrons, and forgetful patrons land in both groups. A difference in on-time returns can then be traced to the reminder.

Worked example — reading a claim. An illustrative headline reads: "Neighborhoods with more coffee shops have lower crime."

What kind of study? Someone compared neighborhoods as they already were. Nobody assigned coffee shops at random.
Correlation or cause? A correlation. The two variables change together across neighborhoods.
A third variable. Income could drive both. Wealthier neighborhoods may attract more coffee shops and also have lower crime rates for many other reasons.
The honest sentence. "More coffee shops tend to come with lower crime across these neighborhoods; this design cannot show that one causes the other."

In words: the headline describes a pattern, and only a different design could test a cause.

Pause and choose one of these two prompts.

(a) Think of a pattern you have heard in the news or on social media, and name a third variable that could explain it.
(b) Think of a claim you have heard about what "makes" people happier, and say what study design you would need to test it.

Misconception check — the arrow in the headline. Students often read "is associated with" as "causes," because a story with a cause is easier to remember. Headlines make it worse by using words like "drives" or "makes." The cure: look for random assignment. No random assignment, and no study designed to find a cause, means no "cause." Say "tends to come with" instead.

6. Name the three traditions: positivism, interpretivism, and two kinds of data

Imagine two researchers studying the same neighborhood food pantry. One counts how many households visit each week and charts the totals over a year. The other sits with visitors in the waiting line and asks what it feels like to come. Both are doing sociology. They are working in different traditions.

Positivism is the research tradition that studies society with the methods of the natural sciences, measuring observable facts and looking for general patterns. Auguste Comte, who gave the discipline its name, argued for a positive — that is, scientific — study of society.
Interpretivism is the research tradition that seeks to understand the meanings people give to their own actions, following Weber's idea of interpretive understanding of an action from the actor's point of view. Max Weber called this verstehen: interpretive understanding of the meaning an action has for the actor.

The two traditions tend to collect different kinds of data.

Quantitative research is research that collects and analyzes numbers, such as survey shares or rates, to describe patterns across many cases.
Qualitative research is research that collects and analyzes words, observations, and meanings, such as interview transcripts or field notes, to understand a setting in depth.

Worked example — sorting the two pantry studies. Place each study.

The counter. Weekly visit totals across a year are numbers, so this is quantitative research in the positivist tradition. It can show when visits rise and fall.
The listener. Conversations in the waiting line are words and meanings, so this is qualitative research in the interpretivist tradition. It can show what a visit means to the people making it.
Together. The counts show the size and timing of the need. The conversations show why some households wait until a crisis to come. Each answers a question the other cannot.

In words: positivism and interpretivism are two lenses on one setting, and a full picture often uses both.

Misconception check — "qualitative means unscientific." Students often think research with numbers is science and research with words is opinion. It is tempting because numbers look precise. But qualitative research follows a method too: careful notes, many conversations, and patterns checked against the evidence. The cure: judge a study by whether its method fits its question, not by whether it has numbers.

7. Protect the people you study: research ethics

Imagine a classmate planning a small study of how students juggle work and school. She plans to interview coworkers at her part-time job. Before she asks a single question, she has duties to every person she will study.

Three rules protect them, and they apply to you whenever you study people.

Informed consent is a person's voluntary agreement to take part in a study after being told its purpose, what it involves, its risks, and the right to stop at any time.
Confidentiality is the research rule that information about individual participants is kept private and never reported in a way that identifies them.
A review board is a committee that examines a proposed study involving human participants and must approve it as ethical before any data are collected.

Worked example — the coworker interviews. Walk the plan through the three rules.

Informed consent. She tells each coworker what the study is about, how long it takes, any risks, and that they may stop at any time. Only then does she begin.
Confidentiality. In her write-up, nobody is named, and she leaves out details such as a rare job title that would let a reader guess who said what.
Review board. If the study is part of a formal research project, a review board reads the plan and must approve it before the first interview.

In words: consent comes first, privacy lasts through the write-up, and an outside review checks the plan before it starts.

Why do these rules exist? The Tuskegee syphilis study is named as the reason they exist. The Readings & Resources page links the public-health agency's own account of what happened in that study.

Misconception check — the secret results. Students often think confidentiality means a study's results must stay secret. It is tempting because "confidential" sounds like "classified." But results are meant to be shared. Confidentiality protects the people: findings are reported in a way that never identifies who said or did what.

8. Read a real survey figure four ways

Back to the number from the start of this chapter: 91% of U.S. adults own a smartphone. You now have the tools to read it properly. You will ask the same four questions you asked in Week 1, and then report a change two ways.

Two units matter here, and you can easily mix them up.

A percentage point is the unit for the difference between two percentages; a share that moves from 35% to 91% has risen 56 percentage points, which is a different number from its percent change.
A percent change is a difference between two values divided by the starting value and multiplied by 100, so a share that moves from 10% to 12% has risen 20 percent.

Worked example — Pew Research Center's smartphone share, read four ways. The source is Pew Research Center's Mobile Fact Sheet.

What is measured? Self-reported ownership. 91% of U.S. adults said they own a smartphone, and 98% said they own a cellphone of some kind, in Pew Research Center's 2025 survey.
Over whom, and when? The population is U.S. adults. The sample is 5,022 U.S. adults surveyed in one window in 2025. The share describes that population at that time.
What does it show? Smartphone ownership was 35% when Pew first measured it in 2011 and 91% in 2025.
- Measured in percentage points, the change is 91 − 35 = 56 percentage points.
- Measured in percent, the change is 56 ÷ 35 × 100 = 160 percent.
- As a ratio: 91 ÷ 35 = 2.6, so the 2025 share is 2.6 times the 2011 share. A 160 percent rise and "2.6 times" are the same change.
- A second share from the same sheet: 98 − 91 = 7 percentage points of U.S. adults own a cellphone that is not a smartphone.
What does it not show? Why ownership rose, or what owning a smartphone does to anyone.
Correlation or cause? Neither. The figure is a description of one population at two points in time.

In words: in 2025, about nine in ten adults in the country said they own a smartphone, up 56 percentage points from 2011; the survey describes that change and does not explain it.

Worked example — counts versus rates, one more time. These two towns are illustrative, not real.

Town A. 12 events among 40,000 residents: 12 ÷ 40,000 × 100,000 = 30 per 100,000.
Town B. 15 events among 30,000 residents: 15 ÷ 30,000 × 100,000 = 50 per 100,000.
Compare rates. Town B's 50 per 100,000 versus Town A's 30 per 100,000. Town B has the larger count and the higher rate here, and a rate, never a count, is what you compare.

The percentage-point companion is illustrative too: a share that moves from 10% to 12% has risen 12 − 10 = 2 percentage points, which is 2 ÷ 10 × 100 = 20 percent.

One limitation is worth saying aloud. A survey share carries some uncertainty because it comes from a sample, not the whole population, so a share is best read as close to the true figure rather than exact to the last digit. That is one more reason to report the source, the population, and the year with every number.

Misconception check — points versus percent. Students often report the smartphone change as "a 56 percent rise." It is tempting because both numbers sit next to a percent sign. But 56 is the difference in percentage points. As a percent of the 2011 share, the rise is 160 percent. The cure: write the unit every time — "56 percentage points" or "160 percent" — and never a bare "56%" for a difference between two percentages.

For more on this week's methods, the module's Readings & Resources page links to short articles and videos.

Key terms

Research cycle — The ordered steps of a study: ask a question, state a hypothesis, define the variables so they can be measured, choose a method, collect data, and analyze them to see whether the hypothesis holds. (first used: 1. Turn a hunch into a study: the research cycle)
Hypothesis — A testable statement of the relationship a researcher expects to find between two or more variables, written before the data are collected. (first used: 1. Turn a hunch into a study: the research cycle)
Variable — A characteristic that can take different values from one person, group, or place to another, such as age, income, or the number of neighbors a person knows. (first used: 1. Turn a hunch into a study: the research cycle)
Operationalize — To define a variable as an exact, observable measurement, such as counting the neighbors a resident can name on a questionnaire, so that different researchers can measure it the same way. (first used: 1. Turn a hunch into a study: the research cycle)
Independent variable — The presumed cause in a study: the variable that an experiment changes on purpose, or that a researcher expects to come first and shape another variable. (first used: 1. Turn a hunch into a study: the research cycle)
Dependent variable — The presumed effect in a study: the variable a researcher measures to see whether it changes along with the independent variable. (first used: 1. Turn a hunch into a study: the research cycle)
Survey — A research method that gathers data by asking many people the same questions, through a questionnaire or an interview, and is best suited to describing a large group. (first used: 2. Match the question to a method)
Experiment — A research method in which the researcher changes one variable for an experimental group, leaves it unchanged for a control group, and compares the two groups to test a cause. (first used: 2. Match the question to a method)
Field research — A research method that collects firsthand data in the natural settings where people live, work, and play, rather than in a laboratory or through a questionnaire. (first used: 2. Match the question to a method)
Ethnography — A form of field research in which the researcher joins a group's daily life over a long period to understand its world from the inside. (first used: 2. Match the question to a method)
Secondary data analysis — A research method that works with data someone else already collected, such as Census tables or official records, to answer a new question. (first used: 2. Match the question to a method)
Experimental group — The people in an experiment who receive the treatment, the change whose effect the researcher wants to test. (first used: 2. Match the question to a method)
Control group — The people in an experiment who do not receive the treatment but are otherwise handled the same way, giving the comparison that isolates the treatment's effect. (first used: 2. Match the question to a method)
Reliability — The consistency of a measure: a reliable measure gives the same result each time it is used on the same thing under the same conditions. (first used: 3. Check the measuring stick: reliability versus validity)
Validity — Whether a measure captures what it is meant to capture; a measure can be consistent and still miss its target, so reliability does not guarantee validity. (first used: 3. Check the measuring stick: reliability versus validity)
Population — The whole group of people a researcher wants to draw conclusions about, such as all U.S. adults. (first used: 4. Decide whom you can speak for: population and sample)
Sample — The smaller set of people drawn from a population who actually take part in a study. (first used: 4. Decide whom you can speak for: population and sample)
Self-selected sample — A sample made up of people who chose to respond, such as visitors who answer an online poll; it represents only the kind of person who opts in and cannot be generalized to a population. (first used: 4. Decide whom you can speak for: population and sample)
Random sample — A sample chosen so that every member of the population has a known, equal chance of selection, which makes the sample representative and lets results be generalized to the population. (first used: 4. Decide whom you can speak for: population and sample)
Sampling bias — A systematic difference between a sample and the population it is meant to represent, caused by how the sample was chosen; collecting more responses the same way does not remove it. (first used: 4. Decide whom you can speak for: population and sample)
Correlation — An association between two variables that change together, so that higher values of one tend to come with higher, or lower, values of the other. (first used: 5. Earn the word cause: correlation versus causation)
Causation — The stronger claim that a change in one variable produces a change in another; it is supported only by an experiment with random assignment or a study designed to identify a cause. (first used: 5. Earn the word cause: correlation versus causation)
Third variable — An outside factor that affects both variables in a correlation, so the two move together even though neither one produces the other. (first used: 5. Earn the word cause: correlation versus causation)
Spurious correlation — An association between two variables that is created by a third variable, such as ice-cream sales and drownings both rising in summer because of the heat. (first used: 5. Earn the word cause: correlation versus causation)
Random assignment — Placing each person in an experiment into the experimental group or the control group by chance, so the groups start out alike and a later difference can be traced to the treatment. (first used: 5. Earn the word cause: correlation versus causation)
Positivism — The research tradition that studies society with the methods of the natural sciences, measuring observable facts and looking for general patterns. (first used: 6. Name the three traditions: positivism, interpretivism, and two kinds of data)
Interpretivism — The research tradition that seeks to understand the meanings people give to their own actions, following Weber's idea of interpretive understanding of an action from the actor's point of view. (first used: 6. Name the three traditions: positivism, interpretivism, and two kinds of data)
Quantitative research — Research that collects and analyzes numbers, such as survey shares or rates, to describe patterns across many cases. (first used: 6. Name the three traditions: positivism, interpretivism, and two kinds of data)
Qualitative research — Research that collects and analyzes words, observations, and meanings, such as interview transcripts or field notes, to understand a setting in depth. (first used: 6. Name the three traditions: positivism, interpretivism, and two kinds of data)
Informed consent — A person's voluntary agreement to take part in a study after being told its purpose, what it involves, its risks, and the right to stop at any time. (first used: 7. Protect the people you study: research ethics)
Confidentiality — The research rule that information about individual participants is kept private and never reported in a way that identifies them. (first used: 7. Protect the people you study: research ethics)
Review board — A committee that examines a proposed study involving human participants and must approve it as ethical before any data are collected. (first used: 7. Protect the people you study: research ethics)
Percentage point — The unit for the difference between two percentages; a share that moves from 35% to 91% has risen 56 percentage points, which is a different number from its percent change. (first used: 8. Read a real survey figure four ways)
Percent change — A difference between two values divided by the starting value and multiplied by 100, so a share that moves from 10% to 12% has risen 20 percent. (first used: 8. Read a real survey figure four ways)

Summary

The research cycle runs question, hypothesis, operationalize, method, data, analysis.
The independent variable is the presumed cause; the dependent variable is the measured effect.
Surveys buy breadth, experiments buy cause, field research buys depth, and secondary data reuses what exists.
A reliable measure is consistent; a valid measure hits its target; you need both.
A representative random sample beats a big self-selected one.
A correlation is a pattern; only random assignment or a design built to find a cause earns "cause."
Positivism counts patterns; interpretivism seeks meaning; good research often uses both.
Informed consent, confidentiality, and review-board approval protect the people studied.
Read every figure four ways, and write "percentage points" or "percent" every time.

Check your understanding

Work these before looking at the answers — they are ungraded, and the answers (with reasoning) are at the bottom of the page.

An illustrative county records 14 bike thefts in a town of 20,000 people and 27 bike thefts in a city of 60,000. Compute each place's rate per 100,000 and say which has more thefts relative to its size.
In an illustrative city, the share of commuters who bike to work rises from 8% to 12%. Report the change in percentage points and in percent.
An after-school program wants to know whether a weekly check-in call raises homework completion. Name the method it should use, the independent variable, the dependent variable, and the control group.
A late-night radio show asks listeners to call in about a new city curfew, and 3,000 people call. Can the show say what the city thinks? Explain using this week's terms.

Answers (worked reasoning — read after attempting)

The town: 14 ÷ 20,000 × 100,000 = 70 per 100,000. The city: 27 ÷ 60,000 × 100,000 = 45 per 100,000. The city has the larger count (27 versus 14), but the town has the higher rate (70 versus 45 per 100,000). Relative to its size, the town has more thefts. This is the count-versus-rate trap: compare rates, never counts.
Measured in percentage points, the change is 12 − 8 = 4 percentage points. Measured in percent, it is 4 ÷ 8 × 100 = 50 percent. Both describe one change. Writing "a 4% rise" would be wrong, because 4 is a difference in points, not a percent of the starting share.
The program wants a cause, so it should run an experiment, ideally with random assignment. The independent variable is whether a student gets the weekly call, because that is the change the program makes on purpose. The dependent variable is homework completion, because that is what it measures. The control group is the students who do not get the call but are otherwise treated the same. Random assignment makes the two groups alike at the start, so a difference in completion can be traced to the calls.
No. The 3,000 callers are a self-selected sample: they are late-night listeners who chose to call, and people with strong feelings are more likely to bother. That is sampling bias, and 3,000 more callers would not fix it. A random sample of city residents, even a few hundred, could speak for the city; the call-in count cannot.

Next up in this module: the Readings & Resources page, the lecture tutorial, the practice set, the quiz, the discussion, and the assignment — and next week the course turns to culture, where the norms model ranks folkways, mores, and taboos.

Built with The Course Maker · samples.thecoursemaker.com