Social research: why and how

Module items¶

Assignment¶

• Research design
  • See: [[How to submit an assignment]]

Sample assignment¶

• Sample Research design

Reading¶

Learning outcomes¶

1. Understand what constitutes social research
2. Learn different research methods
3. Learn sampling
4. Learn the steps of empirical social research
   1. Topic,
   2. Literature review,
   3. Research question,
   4. Hypotheses,
   5. Finding/collecting data, and
   6. Data analysis

What is [[social research]]?¶

• Social research is a research conducted by social scientists following a systematic plan, meaning scientific methods.
• A theory, model, or hypothesis is a set of statements about how concepts are related.
  • By themselves, those statements are just ideas.
  • They become scientific when we design studies that could support them or show them to be wrong.
    • A good scientific idea should be:
      • logically consistent,
      • compatible with known evidence,
      • testable,
      • clear enough that other researchers could repeat the work (reproducibility).
  • Objectivity in science does not mean researchers have no prior beliefs.
    • It means the procedure is public enough that others can inspect it, repeat it, and disagree about what it means.

[[Research methods]]¶

• There are two main research methods. The two approaches ask different kinds of questions about the same social world:
  • [[Quantitative research]]
    • Relies on numbers and statistical data.
    • Uses questions, measures, and counts that can be compared across many cases
    • Aims to describe patterns and test whether concepts are related.
    • We mostly use [[survey]]:
      • Patterns of attitudes and behavior among large groups of people
      • Standardized questions, usually with a questionnaire
  • [[Qualitative research]]
    • Relies on words, conversations, observations, and documents
    • Uses a smaller number of cases in greater depth
    • Aims to understand meanings, processes, and lived experience.
    • We use:
      • [[In-depth interview]]
        • Thought processes, meanings, and stories that lead to opinions or behaviors
        • Open-ended conversation rather than a fixed list of response options
      • [[Content analysis]]
        • Study of documents: texts, images, audio, or video
        • The “cases” are items of content, not always people
      • [[Ethnographic research]]
        • How people interact in everyday settings, rather than only how they say they interact
        • Observation over time, often with field notes

[[Sampling]]¶

• [[Population]]: the universe of units from which the sample is to be selected
  • Often people, but also newspapers, posts, classrooms, or organizations
• [[Sample]]: the segment of that population selected for investigation
• [[Sampling]]: the process of selecting units from a population so that findings from the sample can be used to say something about the population
• We almost always sample.
  • Time and cost make it impossible to study every relevant case.
• A representative sample is designed to act as a small version of the wider population.
• Other samples are chosen because they are information-rich for the question, not because they statistically represent everyone.

How large should a sample be?¶

• There is no single correct number.
• Sample size depends on:
  • how mixed people are on the thing you care about,
  • what kind of analysis you plan to do,
  • how precise you need the estimate to be,
  • time, access, and cost.
• A larger sample reduces some kinds of error, but a large unrepresentative sample is still a poor sample.
• Qualitative studies often use fewer cases on purpose.
  • The goal is depth, and stopping when new interviews stop adding new themes.

[[Social research process]]¶

1. [[Research topic]]
   1. The subject you want to work on
2. [[Literature review]]
   1. What is already known, how it was studied, and where the gaps are
3. [[Research question]]
   1. The specific question your study attempts to answer
4. [[Hypotheses]]
   1. Testable predictions about what you expect to find
5. [[Finding data]]
   1. Collecting new evidence, or locating evidence that already exists
6. [[Data analysis]]
   1. Reducing and interpreting the evidence so you can justify a conclusion

• These steps look linear.
  • In practice they loop. Reading can change the question. Fieldwork can change the coding. Analysis can send you back to the literature.

How is a [[research topic]] formulated?¶

• Literature ➜ Research topic
  • A gap, a debate, or a clash of findings gives you something to work on.
• Research topic ➜ Literature
  • Once you have a topic, you read in a more targeted way.
• Society ➜ Research topic
  • A change in everyday life can open a question: new technology, a campus policy, a migration pattern, a public worry.
• Personal experience can start a topic, but the topic still has to connect to social-science ideas.
• A topic is a territory. It is not yet a study.

Defining a [[literature review]]¶

• A literature review is how you become familiar with the body of research on a topic.
• That body of research is constantly changing.
• You read in order to find:
  • what is already known,
  • which concepts and theories have been used,
  • which methods have been used,
  • where the controversies are,
  • where the evidence clashes,
  • who the key contributors are.
• A literature review is not a summary.
• Its purpose is to locate your argument in existing work, and to show why a new study is still needed.

[[Research question]]¶

• An answerable inquiry into a specific concern or issue.
  • A research question is what the research attempts to answer.
  • It must actually be a question. It ends with a question mark.
• “I am interested in student belonging” is a topic.
  • “What roles do friendship networks, coursework load, and attendance play in a stronger sense of belonging among first-year CSUMB students?” is a research question.
• Poor or missing research questions produce unfocused data collection.

[[Research question]] and [[literature review]]¶

• The research question gives direction and structure to the literature review.
• Formulating a question and reading the literature depend on each other.
• Early reading suggests a question. Further reading revises it.
• A usable question should:
  • be specific enough to guide data collection,
  • be answerable with evidence you can actually get,
  • connect to concepts already used in the literature.
• Research question: To what extent do ethnic identification, contact level with natives, and discrimination influence return migration?

[[Research topic]] vs. [[research question]]¶

• A research topic is what the paper is about.
  • Health disparity, racial inequality, and gender are topics.
    • Research topic: Gender inequality in the workplace
• A research question is an answerable inquiry into a specific topic.
  • Research question: What are the current inequality issues that women face in the workplace?

[[Hypotheses]]¶

• Tentative, testable predictions about what we expect to find before the study is finished.
  • A hypothesis is derived from a broader idea, then stated in a form that evidence can support or weaken.
• Hypotheses are educated guesses that derived from [[literature review]].
  • Hypothesis 1: I expect to find that ethnic identification increases return migration
  • Hypothesis 2: I expect to find that contact level with natives decreases return migration
  • Hypothesis 3: I expect to find that discrimination increases return migration
    • Research question: To what extent do ethnic identification, contact level with natives, and discrimination influence return migration?

[[Finding data]]¶

• Data are the empirical material that let you answer the question.
• Two broad routes:
  • [[Primary data]]: you collect it yourself for this study.
  • [[Secondary data]]: someone else already collected it, and you analyze it.
• The data must actually contain the concepts in your research question.
  • If you want to study belonging and campus jobs, the dataset or interviews must include belonging and campus jobs.
  • People’s answers, field notes, documents, and existing surveys can all be data.

Finding data¶

• Quantitative data usually begin as answers that can be turned into numbers.
• Common sources:
  • existing social surveys and official statistics,
  • a questionnaire you write and administer,
  • coded counts from documents or media.
• Strengths:
  • many cases,
  • comparable measures,
  • a path from sample to population if the sample was designed for that.
• Limits:
  • you only see what the questions or codes already asked,
  • missing answers and a mismatched list of people can distort the picture.

Finding data - Collecting your own¶

• You design the questions or observations because no existing source measures what you need.
• Quantitative version: a survey of CSUMB students with items you wrote.
• Qualitative version: interviews, focus groups, or observation with a smaller set of people.
• Collecting your own data gives control, and it creates obligations:
  • you must recruit people,
  • you must keep the procedure consistent,
  • you must handle consent, privacy, and the mess of real fieldwork.
• The research question still comes first. Do not collect first and invent the question later.

[[Data analysis]]¶

• Analysis is the process of reducing a large pile of evidence until a justified conclusion is possible.
  • Raw interviews, questionnaires, or documents do not speak for themselves.
    • Quantitative analysis reduces numbers to tables, averages, and tests of relationships.
    • Qualitative analysis reduces text to coded categories, themes, and an interpretation.
      • In both cases, analysis is a claim about what the evidence shows, tied back to the research question.

Data analysis ([[Quantitative research]])¶

• Start by checking the data: missing answers, impossible values, and whether the measures match the concepts.
• Then describe the sample, and then test the relationships in the hypotheses.
• Example: a survey of immigrants asks about ethnic identification, contact with natives, discrimination, and intention to return.
  • Each concept becomes a variable.
  • Statistical models estimate whether those variables move together in the predicted direction, after other differences are taken into account.
• More advanced models can treat several relationships at once, but the logic stays the same: measured concepts, hypothesized links, and a test against the data.

Data analysis ([[Qualitative research]])¶

• Qualitative data arrive as transcripts, field notes, and documents. One interview can produce dozens of pages.
• Analysis usually begins while data are still being collected, so later questions can follow what is emerging.
• Typical steps:
  • read and reread,
  • index the text into codes and themes,
  • compare cases,
  • refine the themes until they explain the material.
• When several people code the same transcripts, they check how much their coding agrees. Disagreement is a signal to clarify the codebook, not a failure.
• Example: interviews with older immigrants about health, care, and family conflict.
  • The point is not to count return plans.
  • The point is to interpret what moving for care means in those lives, in the speakers’ own terms.