Keane’s Vaping Study Began with Users’ Accounts

• Helen Keane and colleagues monitored an online vapers’ forum
• They invited users to describe personal vaporizers in their own words
• The final analysis used 705 responses; 97% reported having been daily smokers
• Respondents described vaping spreading through friends, relatives, curiosity, and gifts

Why Sociologists Need Systematic Evidence

• Familiar claims can be repeated long after the evidence has been misunderstood
• A striking result may not reappear when other researchers repeat the study
• A large number of responses can still come from a badly selected sample
• A correlation can look causal when a third variable produces both outcomes
• Research on people can be methodologically strong and still be unethical

Questions to Ask About a “Studies Show” Headline

Our Research Project: Popularity Signals and Credibility

• Popularity signals and credibility are still abstract ideas
• We have not yet decided how to observe or measure either one
• We will make the question more precise as we build the study

What a Hypothesis Is—and When We Use One

• Hypothesis — a speculative statement about the relationship between two or more variables
• It is stated before the data are collected, so the evidence can support or refute it

A Hypothesis Is Not a Finding

Hypothesis: Posts that appear more popular will receive higher perceived-credibility ratings than posts that appear less popular.

The Scientific Method Is a Cycle

Define the Problem and Review the Literature

Define the problem

• State clearly what the study will investigate
• Replace a broad topic with a researchable question

Review the literature

• Examine relevant scholarly studies and information
• Refine the problem and possible data-collection techniques
• Reduce avoidable mistakes

Operational Definitions Make Concepts Measurable

• Operational definition — an explanation of an abstract concept specific enough for a researcher to assess it

Hypotheses, Variables, and Causal Logic

• Hypothesis — a speculative statement about the relationship between two or more variables
• Variable — a measurable trait or characteristic that changes under different conditions
• The independent variable (x) is hypothesized to influence the dependent variable (y)
• Causal logic links a condition or variable to a consequence, with one leading to the other

Ice Cream Sales and Drowning Deaths

Correlation Does Not Establish Causation

A Sample Represents a Larger Population

• Population — the entire group the researcher wants to understand
• Sample — the smaller group actually studied
• A representative sample reflects the population well enough to support conclusions about it

Probability Sampling Uses Chance

Convenience and Snowball Samples Are Nonrandom

• Convenience sample — recruits people who are readily accessible
• Snowball sample — participants help recruit others through their networks
• Self-selected sample — people decide for themselves whether to participate

Sample Size Cannot Repair Selection Bias

• A self-selected website poll may collect tens of thousands of responses
• Those responses still reflect only the people who encountered the poll and chose to answer
• A carefully selected sample of 1,500 can represent a population more accurately

Sampling the Credibility Project

• KU students, adults in Kuwait, or users of one platform are different populations
• Recruiting only the instructor’s followers would create a convenience sample
• Recruiting only heavy users could exaggerate platform familiarity
• A defensible conclusion names the population the sample can actually represent

Validity and Reliability Ask Different Questions

Control Variables Test Rival Explanations

• Control variable — a factor held constant to test the relative impact of an independent variable
• Keep the post wording, source name, image, topic, and layout identical
• Measure or balance prior belief, topic familiarity, media literacy, and platform familiarity

Select a Design That Can Answer the Question

Conclusions Lead to Reports and New Questions

• A conclusion states whether the evidence supports or refutes the hypothesis
• It must identify limits, exceptions, and rival explanations that remain
• Researchers then prepare a report, often beginning with an executive summary
• The conclusion should generate ideas for future research

The Textbook Hypothesis Links Education to Income

• Independent variable (x): level of education
• Dependent variable (y): income
• Operational definitions: years of schooling and income reported for the past year

Macro-Level Evidence Does Not Answer a Micro-Level Question

• The literature review found that U.S. states with more educated residents also had higher household incomes
• That is a macro-level relationship between state characteristics
• The hypothesis asks a micro-level question about individuals

Figure 2-4 Supports the Education–Earnings Hypothesis

Exceptions, Inequality, and Local Context Matter

• A high-earning school dropout and a modest-earning doctorate holder are exceptions within the pattern
• Family resources, occupation, gender, and other factors may alter the relationship
• Control variables help test education’s relative impact
• Kuwait’s credential-linked salary and occupational structures make “the degree itself” especially difficult to isolate as a recurring local project

Schaefer’s Four Major Research Designs

Surveys Use Interviews and Questionnaires

• Survey — an interview or questionnaire that provides information about how people think and act
• Interview — face-to-face, phone, or online questioning; higher response and opportunities to probe
• Questionnaire — a printed or written form; cheaper for large samples
• Surveys can reach many people, but sampling and wording determine what the answers mean

Cross-Sectional and Longitudinal Studies Use Time Differently

Cell-Phone and Web Surveys Change the Sample

• In 2018, 54% of U.S. households—and 72% of people ages 18–29—were reachable only by cell phone
• Completing a cell-phone survey took about nine calls, compared with five for a landline
• Excluding cell-only users could distort the population represented
• Web surveys are inexpensive, but opt-in participation can create self-selection
• A carefully drawn web sample can still produce valid results

Question Wording and Interviewers Shape Answers

• Effective questions are simple, clear, specific, and unbiased
• Wording must change as social life changes
• Leading, loaded, or double-barrelled questions damage validity
• Interviewer characteristics can influence what respondents disclose

Quantitative and Qualitative Research Answer Different Questions

• Quantitative research collects and reports data primarily in numerical form
• Qualitative research relies on field and naturalistic settings, often with smaller groups

Ethnography Studies a Social Setting in Depth

• Ethnography — study of an entire social setting through extended systematic fieldwork
• Observation — direct participation in closely watching a group or organization
• Ethnographers also collect historical information and conduct interviews
• Detailed notes make the method systematic rather than casual

Alice Goffman Used Long-Term Participant Observation

• Participant observation — the sociologist joins a setting for an extended period to understand how it operates
• Alice Goffman conducted six years of fieldwork in a heavily policed, low-income Philadelphia neighborhood
• The study argued that warrants and police contact shaped everyday routines and relationships
• The study later prompted debate about evidence, anonymization, researcher involvement, and ethical boundaries

Visual Sociology Treats Images as Data

• Visual sociology — the use of photographs, film, and video to study society
• Images are evidence, not decoration
• Charles Suchar photographed changing neighborhoods over time
• Repeated satellite or aerial images can track land use, building density, roads, and green space
• Images show what changed and where; interviews and records help explain why

Experiments Manipulate an Independent Variable

• Experiment — an artificially created situation that allows a researcher to manipulate variables
• The experimental group is exposed to the independent variable
• The control group is not
• A study can use more than one experimental group, each receiving a different version of the treatment
• Groups should be comparable except for the treatment

The Credibility Project Fits an Experiment

Comparable participants all see the same post. Only what appears beside it changes.

Then compare average perceived-credibility scores across the three groups.

The Hawthorne Effect Changes Observed Behavior

Existing Sources Allow Nonreactive Research

• Secondary analysis uses previously collected, publicly accessible information and data
• It is often nonreactive: the analysis does not influence people’s behavior
• Census, health, birth, death, marriage, and divorce records are common sources
• The limitation is fit: data collected for another purpose may omit what the new question needs

Content Analysis Codes Cultural Materials

• Content analysis — systematic coding and objective recording of data, guided by a rationale
• Sources can include newspapers, television, websites, scripts, diaries, songs, folklore, and legal documents
• Schaefer reports that women’s sports received 3.2% of airtime in a 2015 content analysis

Table 2-3: Advantages and Limitations

Identify the Research Design

Network Analysis and Digital Traces Extend the Toolkit

The ASA Code Sets Seven Ethical Duties

1. Maintain objectivity and integrity
2. Respect rights, dignity, and diversity
3. Protect subjects from personal harm
4. Preserve confidentiality
5. Seek informed consent in private contexts
6. Acknowledge collaboration and assistance
7. Disclose all financial support

Rik Scarce Chose Confidentiality

• In 1993, a federal grand jury questioned doctoral student Rik Scarce
• He was researching environmental activists and knew at least one suspect in a laboratory raid
• Scarce refused to disclose what he knew
• He spent 159 days in jail; the ASA supported his appeal

The Facebook Experiment Tested Informed Consent

• In 2014, researchers altered the News Feeds of 689,000 Facebook users
• The experiment tested whether emotional content spread through a network
• Users were not asked for meaningful informed consent

Exxon Funding Created a Conflict of Interest

Value Neutrality Applies to Interpretation

• Value neutrality — Weber’s principle that personal values may guide question choice but must not influence data interpretation
• Researchers must accept findings that contradict their views or theories
• Peter Rossi’s estimate of homelessness in 1980s Chicago fell below advocates’ estimates and drew strong criticism

Morris’s Result Did Not Show Mentoring Caused Harm

• Hypothesis: mentoring would reduce problems among children with an incarcerated parent
• Unexpected result: the mentored group had more reported problems
• The groups already differed in prior problems, age, and length of parental absence
• The program’s duration and emphasis on school success also shaped what it could change

Percentage, Mean, Mode, and Median Summarize Data

• Percentage — a portion of 100; useful for comparing groups of different sizes
• Mean — values added together and divided by the number of values
• Mode — the single most common value
• Median — the midpoint dividing ordered values into two equal-sized groups

Cross-Tabulations Show Relationships Between Variables

• Cross-tabulation — a table showing the relationship between two or more variables

Useful Data Requires a Defined Outcome and a Missing-Group Check

• Dave Eberbach found that local patterns disappeared inside state and county averages
• 1. Define the outcome: what would count as success or harm?
• 2. Choose the evidence: which measure and design could detect it?
• 3. Check who is missing: whose experience may be absent from the data?

A Research Report Completes the Cycle

• Narrow the topic enough to answer it well
• Search books, scholarly databases, government documents, and relevant organizations
• Evaluate online authorship, sponsorship, date, and possible bias
• Build an outline with a connected introduction and conclusion
• Draft, pause, revise, and read difficult passages aloud
• Cite every borrowed idea or exact phrase

The Credibility Project from Question to Conclusion

1. Define popularity signals and a three-item perceived-credibility score
2. State the hypothesis without calling it a finding
3. Select a population and defensible sample
4. Hold the post constant and manipulate only the signals
5. Check validity, reliability, and rival explanations
6. Protect consent, privacy, and confidentiality
7. Report support or refutation, limits, and future questions

Research Skills Travel Across Majors

• Medicine: randomized controlled trials assign participants to treatment or control groups by chance, then track benefits, adverse effects, and informed consent
• Architecture: compare similar hospital wards to ask how daylight, noise, layout, and wayfinding relate to sleep, stress, or recovery
• Business: market research depends on representative samples and unbiased questions
• Computer science: operationalize “interaction” and “time on platform,” then audit opaque algorithms through their inputs, outputs, errors, and overlooked groups

From a Plausible Claim to a Defensible Conclusion

• Define the problem and its concepts
• Treat hypotheses as questions for evidence, not preferred answers
• Match the design to the claim
• Ask who entered the sample and who did not
• Test rival explanations
• Protect participants and interpret results honestly
• Report limits and begin the next research cycle