Chapter 7: The Research Question

Learning Objectives

Distinguish between weak and strong research questions
Recognize and avoid common pitfalls in question formulation
Translate theoretical concepts into operational definitions
Write research questions that are specific, measurable, and answerable
Understand the relationship between research questions and project scope

Crossroads representing research decision points

The hardest part of research isn’t analyzing data or writing the final report. It’s asking the right question.

This seems counterintuitive. Students arrive at research methods courses assuming the challenge will be statistical tests or coding schemes, technical skills they haven’t yet learned. But most students can learn to run a chi-square test or code qualitative data with practice. What’s harder is the intellectual work that precedes analysis: formulating a question that’s specific enough to answer, broad enough to matter, and grounded enough in existing knowledge to avoid reinventing the wheel.

A weak research question dooms the entire project. You can have perfect data, sophisticated methods, and rigorous analysis, but if the question itself is vague, unanswerable, or trivial, the study contributes nothing.

Consider these two questions:

Question A: “How does music affect people?”

Question B: “Do songs with negative lyric sentiment achieve higher peak chart positions than songs with positive sentiment, controlling for tempo and artist popularity?”

Question A is a career-spanning research program disguised as a single question. You could write a thousand dissertations exploring different aspects of it and still not exhaust the topic. It’s so broad that it’s essentially unanswerable within any realistic scope.

Question B is researchable. It specifies what will be measured (lyric sentiment, chart position), what will be compared (negative vs. positive), and what will be controlled (tempo, artist popularity). You can design a study around it. You can determine what evidence would answer it. You can know when you’re done.

The same distinction operates in every domain. “How does news affect democracy?” is a Question A. “Does exposure to episodic framing of poverty reduce support for structural antipoverty programs compared to thematic framing among registered voters?” is a Question B. The craft of research question formulation is the craft of getting from A to B.

What Makes a Research Question “Good”?

A strong research question meets five criteria. It’s worth examining each carefully, because students often violate these principles without realizing it.

1. It’s Specific

Weak questions use vague language that could mean different things to different people:

Weak: “How does social media influence politics?”

What does “influence” mean? Change voting behavior? Shape discourse? Affect polarization? Mobilize activism? “Social media” is equally vague: Twitter? TikTok? All platforms? And “politics”: electoral politics? Policy debates? Political identity?

Stronger: “Does exposure to political content on TikTok increase political efficacy among Gen Z voters?”

This narrows platforms (TikTok), specifies the outcome (political efficacy), and bounds the population (Gen Z voters).

For music research:

Weak: “What role does music play in identity?”

Stronger: “How do fans of K-pop use parasocial language differently than fans of country music when discussing their favorite artists on Reddit?”

2. It’s Measurable

Some questions use concepts that resist operationalization, the translation into concrete, observable variables.

Weak: “Do authentic artists create better music?”

What does “authentic” mean in a measurable way? What counts as “better”? Without clear operational definitions, you can’t design a study.

Stronger: “Do artists who write their own lyrics receive higher critical ratings on Pitchfork than artists who use professional songwriters?”

Now “authentic” is operationalized as “self-written lyrics” (verifiable) and “better” is operationalized as “Pitchfork rating” (quantifiable). You can disagree with these operationalizations, but at least they’re concrete.

A non-music parallel: “Is this news article biased?” is unmeasurable as stated. “Do CNN and Fox News differ in the proportion of episodic vs. thematic frames used in coverage of immigration?” is measurable.

3. It’s Answerable Within Your Constraints

Some questions are good in principle but impossible given your resources, timeline, or skills.

Weak for a semester project: “How has hip-hop evolved globally from 1973 to 2024?”

This requires decades of data, expertise in multiple languages and cultures, and years of analysis.

Stronger: “How did framing of police violence in Billboard Hot 100 hip-hop songs change between two high-profile incidents: Ferguson (2014) and Minneapolis (2020)?”

This focuses on two specific moments, one genre, one measurable dimension (framing), using accessible data (charting songs).

4. It’s Not Already Definitively Answered

Your literature review (Chapter 4) should reveal whether your question is genuinely open or has already been resolved.

Problematic: “Does tempo affect perceived energy in music?”

If your literature review shows this has been tested dozens of times with consistent results, asking it again without a novel angle (new population, new genre, testing a moderator) is redundant.

Better: “Does the relationship between tempo and perceived energy differ between listeners with and without musical training?”

This tests a potential moderator that previous research may not have examined.

5. It Matters

The question should have intellectual or practical significance. It should contribute to theory, resolve a contradiction, or have applied value.

Weak: “Do songs released on Tuesdays chart higher than songs released on Thursdays?”

Even if true, so what? This is trivia, not insight. It doesn’t test theory, address a gap, or have clear implications.

Stronger: “Do songs released during culturally significant moments (post-tragedy, during social movements) achieve different chart longevity than songs released during neutral periods?”

This has theoretical implications (cultural resonance, collective emotion) and practical value (understanding audience engagement). The same logic applies beyond music: “Do corporate apology statements released within 24 hours of a crisis receive more favorable media coverage than those released after 72 hours?” is a question with both theoretical and practical significance.

Common Pitfalls in Question Formulation

Certain mistakes appear repeatedly in student research proposals. Learning to recognize and avoid them accelerates the refinement process.

Pitfall 1: The Impossible Comparison

Problematic: “Are TikTok users happier than non-users?”

This seems straightforward, but it’s confounded by self-selection. People who are already predisposed to certain traits choose to use or avoid platforms. Correlation doesn’t reveal causation.

Better: “Among current TikTok users, does time spent on the platform correlate with self-reported well-being, controlling for baseline mood and social connectedness?”

This is still correlational, but it’s more honest about what it can demonstrate. Or, if you wanted causal inference:

Even better: “Does random assignment to a two-week TikTok abstinence condition change self-reported well-being compared to a control group?”

Now you have an experimental design that can support causal claims.

Pitfall 2: The Unmeasurable Concept

Problematic: “What is the essence of musical beauty?”

“Essence” and “beauty” are philosophical concepts, not empirical variables. You can’t code for “essence.”

Better: “What musical features (tempo, key, harmonic complexity) predict listener ratings of aesthetic pleasure?”

You’ve operationalized “beauty” as “aesthetic pleasure” (measurable via ratings) and identified concrete features to test.

Pitfall 3: The Everything Question

Problematic: “How has music changed society over time?”

This is a book series, not a research question.

Better: “Did the rise of politically explicit hip-hop in the 1980s correlate with increased political engagement among Black youth, as measured by voter registration rates?”

Specific genre, specific time period, specific outcome.

Pitfall 4: The Circular Question

Problematic: “Do popular songs become popular because people like them?”

This is tautological. “Popular” and “people like them” are the same thing.

Better: “Do songs with specific audio features (high energy, major key, fast tempo) achieve higher initial chart positions in the first week of release compared to songs lacking these features?”

This tests whether specific features predict early success, which has theoretical implications about what drives initial popularity before word-of-mouth effects accumulate.

Pitfall 5: The False Binary

Problematic: “Is music or lyrics more important for chart success?”

This assumes you must choose one. Reality is more complex.

Better: “To what extent do lyric sentiment and audio features (tempo, energy) independently predict chart success, and do they interact?”

This allows both to matter and tests whether their effects combine in non-additive ways.

From Theoretical Concepts to Operational Definitions

Theory gives you abstract concepts. Research requires concrete variables. The bridge is operationalization, defining how you’ll measure theoretical constructs.

Example: Parasocial Relationships

Theoretical concept: Parasocial relationship (Horton & Wohl, 1956), a one-sided emotional bond between a fan and a media figure.

How do you measure it?

Option 1: Self-Report Scale

Ask fans to rate agreement with statements like:

“I feel like [artist] understands me.”
“I would be devastated if [artist] retired.”
“I think about [artist] even when not listening to their music.”

Sum the ratings to create a parasocial bond score.

Pros: Captures subjective experience directly. Cons: Subject to social desirability bias. People may over- or under-report.

Option 2: Behavioral Indicators

Measure observable behaviors:

Number of concerts attended
Frequency of social media interactions with artist content
Time spent in fan communities

Pros: Behavioral data may be more reliable than self-report. Cons: Behavior doesn’t necessarily reflect emotional bond (you might attend concerts for social reasons, not parasocial attachment).

Option 3: Language Analysis

Code fan-created content (forum posts, comments) for parasocial language:

First-person references to the artist (“I love how Taylor…”)
Expressions of intimacy (“She gets me”)
Emotional disclosure (“When I’m sad, I listen to…”)

Pros: Naturalistic data, reveals how fans actually talk. Cons: Labor-intensive coding. May miss fans who feel strongly but don’t post publicly.

Which is best? Depends on your research question, data access, and resources. The point is that “parasocial relationship” isn’t directly observable; you must choose a measurable proxy and defend why it’s valid. This logic applies identically to non-music concepts: “media trust,” “political polarization,” and “brand loyalty” all require the same translation from abstract construct to concrete measurement.

Example: Lyric Sentiment

Theoretical concept: Emotional valence of lyric content.

How do you measure it?

Option 1: Automated Sentiment Analysis (LIWC)

Use software like Linguistic Inquiry and Word Count (LIWC) to score lyrics based on positive and negative emotion words.

Pros: Fast, replicable, no coder training needed. Cons: Misses context (sarcasm, irony, metaphor). “I’m bad” in Michael Jackson’s song is positive, not negative. Napier and Shamir (2018) used this approach across decades of Billboard lyrics but acknowledged these limitations.

Option 2: Human Coding

Train coders to read lyrics and categorize overall sentiment as positive, negative, or neutral based on holistic judgment.

Pros: Captures context, nuance, metaphorical meaning. Cons: Labor-intensive. Requires inter-coder reliability testing (Lombard et al., 2002). Subjective.

Option 3: Dimensional Coding

Code for multiple dimensions:

Valence (positive/negative)
Arousal (high energy/low energy)
Dominance (powerful/submissive)

Pros: Captures emotional complexity better than a single valence dimension. Aligns with psychological models of emotion like those proposed by Zentner, Grandjean, and Scherer (2008). Cons: More complex, requires careful codebook development.

Again, no single “right” answer. You choose based on what your theory predicts and what your question requires.

Writing Strong Research Questions: A Formula

Here’s a template that consistently produces clear, researchable questions:

[Does/Is there/How does] + [Specific Variable or Concept] + [Relationship/Pattern/Difference] + [Among/Between/Within] + [Bounded Population or Context] + [Controlling for/Accounting for Confounds, if applicable]?

Examples Using the Formula

Example 1 (Descriptive, music): “What is the prevalence of mental health references in Billboard Hot 100 songs between 2015 and 2024?”

Specific variable: Mental health references
Pattern: Prevalence
Context: Billboard Hot 100, 2015-2024

Example 2 (Relational, music): “Is there a relationship between lyric sentiment (positive, negative, neutral) and chart peak position among pop songs in the Billboard Hot 100 (2015-2024), controlling for tempo and artist popularity?”

Variables: Lyric sentiment (IV), chart position (DV)
Relationship: Correlation or difference
Context: Pop songs, Billboard Hot 100, specific timeframe
Controls: Tempo, artist popularity

Example 3 (Comparative, music): “How does parasocial language use differ between K-pop fans and country music fans in Reddit fan communities, and does this difference persist when controlling for age and gender?”

Variables: Parasocial language
Comparison: K-pop vs. country fans
Context: Reddit fan communities
Controls: Age, gender

Example 4 (Descriptive, non-music): “What is the prevalence of episodic vs. thematic framing in New York Times coverage of homelessness between 2019 and 2024?”

Specific variable: Frame type (episodic vs. thematic)
Pattern: Prevalence
Context: NYT coverage, specific timeframe

Example 5 (Relational, non-music): “Is there a relationship between news frame type (episodic vs. thematic) and reader attribution of responsibility (individual vs. structural) among adults exposed to news about poverty, controlling for political ideology?”

Variables: Frame type (IV), attribution (DV)
Relationship: Difference
Context: Adults, poverty news
Controls: Political ideology

The formula works regardless of domain. Swap the variables and context; the structure holds.

The Relationship Between Questions and Hypotheses

Sometimes you pose research questions. Sometimes you pose hypotheses. Understanding when to use which is crucial.

Use Research Questions (RQs) When:

1. You’re exploring new territory

If no theory predicts the outcome and you’re genuinely uncertain what the data will show, use an RQ.

Example: “How do fans of electronic dance music use parasocial language differently than fans of classical music?”

You don’t have a theory predicting the direction of difference; you’re exploring.

2. You’re describing rather than explaining

Descriptive studies often use RQs because they’re mapping terrain, not testing causal mechanisms.

Example: “What percentage of top-charting songs contain political references?”

3. The literature is too contradictory to predict confidently

If previous studies show inconsistent patterns, an RQ is more honest than picking a side arbitrarily.

Example: “Is there a relationship between lyric sentiment and chart success, given that existing studies have tracked aggregate trends in lyric negativity (Brand et al., 2019; Napier & Shamir, 2018) but have not tested whether sentiment predicts individual song performance?”

Use Hypotheses (H) When:

1. Theory makes a specific prediction

If your theoretical framework predicts a direction or pattern, state it as a hypothesis.

Example: “H1: Songs in minor keys will be rated as sadder than songs in major keys (derived from appraisal theory; Scherer, 2004).”

2. Previous research suggests a likely outcome

If the literature consistently finds a pattern, you can hypothesize it will replicate in your context.

Example: “H1: Popular song lyrics will show increasing negative emotional expression over time, consistent with Brand et al. (2019) and Napier and Shamir (2018).”

3. You’re testing a causal claim

Causal research typically uses hypotheses because you’re predicting that X causes Y.

Example: “H1: Exposure to sad music will decrease self-reported mood compared to a control condition.”

Revising Questions: The Iteration Process

First drafts of research questions are rarely optimal. Expect to revise multiple times.

Iteration Example: From Vague to Precise

Draft 1: “How does music affect emotion?”

Problem: Too broad. What music? What emotions? In what context?

Draft 2: “How does sad music affect listeners’ emotions?”

Problem: “Sad music” is vague (minor key? sad lyrics? both?). “Affect emotions” is vague (make them sadder? provide catharsis?).

Draft 3: “Do songs with negative lyric sentiment make listeners feel sadder after listening?”

Problem: Good, but how are you measuring “sadder”? Self-report? And what’s the comparison? Sadder than before? Sadder than a control condition?

Draft 4: “Does listening to songs with negative lyric sentiment decrease self-reported mood compared to listening to songs with neutral sentiment?”

Much better: Specific stimulus (negative lyrics), specific outcome (self-reported mood), clear comparison (negative vs. neutral), measurable and testable.

Testing Your Question: The Checklist

Before finalizing your research question, run it through these diagnostics:

☐ Can I operationalize every concept? Every abstract term (authenticity, beauty, emotion) must translate into something measurable.

☐ Is the scope manageable? Can I answer this in one semester with available resources?

☐ Does my theory predict this pattern? If using a hypothesis, does it actually follow from the theoretical framework?

☐ Would a null finding still be interesting? If you find no relationship, does that tell us something valuable? Or is the question only interesting if the answer is “yes”?

☐ Can I specify what evidence would answer the question? If you can’t envision the data and analysis that would resolve the question, it’s not ready.

☐ Does the question advance beyond description to explanation? “What exists?” questions are fine for exploratory work, but “Why does it exist?” questions are typically more valuable.

Practice: Crafting Research Questions

Exercise 7.1: Diagnosing Weak Questions

For each question below, identify what’s wrong and propose a revision:

Question A: “How does technology change communication?”

Problem(s): _______________ Revision: _______________

Question B: “Are happy people more successful?”

Problem(s): _______________ Revision: _______________

Question C: “What is the true meaning of rap music?”

Problem(s): _______________ Revision: _______________

Question D: “Does media coverage matter?”

Problem(s): _______________ Revision: _______________

Exercise 7.2: Operationalizing Concepts

Choose one abstract concept and develop three different operational definitions:

Concept: Musical “authenticity”

Operationalization 1 (how you’d measure it): _______________ Operationalization 2: _______________ Operationalization 3: _______________

Which is best for your research question, and why?

Now do the same for a non-music concept: “media bias”

Operationalization 1: _______________ Operationalization 2: _______________ Operationalization 3: _______________

What do you notice about the choices involved in operationalizing an abstract concept?

Exercise 7.3: From Vague to Precise

Take this vague interest and narrow it through five iterations:

Starting point: “I’m interested in music and mental health.”

Iteration 1 (narrow the scope): _______________ Iteration 2 (specify the relationship): _______________ Iteration 3 (bound the population/context): _______________ Iteration 4 (add measurement specificity): _______________ Iteration 5 (final researchable question): _______________

Exercise 7.4: RQ or Hypothesis?

For each scenario, decide whether to use an RQ or hypothesis, and write it:

Scenario A: Social Identity Theory (Tajfel & Turner, 1979) predicts in-group favoritism. You hypothesize fans will rate their preferred genre higher than rival genres.

Use: RQ or H? Write it: _______________

Scenario B: No research has examined how EDM fans talk about their favorite DJs compared to how jazz fans talk about their favorite musicians. You’re exploring this.

Use: RQ or H? Write it: _______________

Scenario C: Multiple studies show that popular lyrics have become more negative over time (Brand et al., 2019; Napier & Shamir, 2018). You expect this to replicate in your dataset.

Use: RQ or H? Write it: _______________

Scenario D: You want to know whether nonprofit fundraising emails use more emotional language than for-profit marketing emails, but no prior research has examined this comparison.

Use: RQ or H? Write it: _______________

Exercise 7.5: The Revision Process

Take a weak research question you’ve written (or make one up). Revise it five times, addressing a different problem each time:

Draft 1 (original, weak): _______________ Draft 2 (more specific): _______________ Draft 3 (added measurement): _______________ Draft 4 (bounded context): _______________ Draft 5 (final, strong): _______________

What improved most between Draft 1 and Draft 5?

Reflection Questions

The Precision-Breadth Tradeoff: Strong research questions are narrow and specific, which means they can’t address all the interesting aspects of a topic. How do you decide which aspect to focus on and which to leave for future research?
Operationalization Choices: Every time you operationalize an abstract concept, you make choices that shape what you can discover. What gets lost when you reduce “musical beauty” to “listener ratings”? Is that loss acceptable, or does it fundamentally change the question?
Your Own Question: Write your current research question for this course project. Now critique it using the five criteria from this chapter. What needs revision?
Cross-Domain Transfer: Take your research question from this course and rewrite it for a completely different domain (news, health, politics, advertising). What changes? What stays the same? Does the translation reveal weaknesses in the original formulation?

Chapter Summary

This chapter focused on the craft of question formulation:

Five criteria for strong RQs: Specific, measurable, answerable, not already resolved, significant
Common pitfalls: Impossible comparisons, unmeasurable concepts, scope too broad, circular logic, false binaries
Operationalization translates theoretical concepts into measurable variables. Multiple valid operationalizations often exist; choose based on theory and resources.
Formula for strong questions: [Action verb] + [specific variable] + [relationship/pattern] + [bounded context] + [controls if applicable]
RQs vs. Hypotheses: Use RQs when exploring or when theory doesn’t predict direction. Use hypotheses when theory makes specific predictions.
Iteration is essential: First drafts are rarely optimal. Expect to revise questions multiple times.
The checklist: Can you operationalize it? Is scope manageable? Does theory predict it? Would null findings matter? Can you specify what evidence answers it?

Key Terms

Hypothesis (H): Testable prediction derived from theory (directional statement)
Null hypothesis (H₀): Prediction of no relationship or difference
Operationalization: Translating abstract concepts into concrete, measurable variables
Research question (RQ): Open-ended or bounded inquiry when theory doesn’t specify direction
Scope: The boundaries of what a study will examine (population, context, timeframe)

References

Brand, C. O., Acerbi, A., & Mesoudi, A. (2019). Cultural evolution of emotional expression in 50 years of song lyrics. Evolutionary Human Sciences, 1, e11. https://doi.org/10.1017/ehs.2019.11

Horton, D., & Wohl, R. R. (1956). Mass communication and para-social interaction: Observations on intimacy at a distance. Psychiatry, 19(3), 215-229. https://doi.org/10.1080/00332747.1956.11023049

Lombard, M., Snyder-Duch, J., & Bracken, C. C. (2002). Content analysis in mass communication: Assessment and reporting of intercoder reliability. Human Communication Research, 28(4), 587-604. https://doi.org/10.1111/j.1468-2958.2002.tb00826.x

Napier, K., & Shamir, L. (2018). Quantitative sentiment analysis of lyrics in popular music. Journal of Popular Music Studies, 30(4), 161-176. https://doi.org/10.1525/jpms.2018.300411

Scherer, K. R. (2004). Which emotions can be induced by music? What are the underlying mechanisms? And how can we measure them? Journal of New Music Research, 33(3), 239-251. https://doi.org/10.1080/0929821042000317822

Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin & S. Worchel (Eds.), The social psychology of intergroup relations (pp. 33-47). Brooks/Cole.

Zentner, M., Grandjean, D., & Scherer, K. R. (2008). Emotions evoked by the sound of music: Characterization, classification, and measurement. Emotion, 8(4), 494-521. https://doi.org/10.1037/1528-3542.8.4.494

Graduate Extension

Required Task: Locate three published empirical studies in communication journals that state explicit research questions or hypotheses. For each study:

Extract the stated RQ(s) or hypothesis/hypotheses exactly as written.
Evaluate each using the five criteria from this chapter (specific, measurable, answerable, not already resolved, significant). Assign a rating (strong, adequate, weak) for each criterion and justify your assessment in 1-2 sentences.
Identify the operational definitions used for the key variables. Are they explicitly stated? Could another researcher replicate the measurement from the information provided?
Assess theory-question alignment: Does the RQ or hypothesis follow logically from the theoretical framework cited? Or is there a gap between what the theory predicts and what the question actually tests?
Revise the weakest question of the three. Rewrite it using the formula from this chapter, making it more specific, better operationalized, or more tightly connected to theory.

Why this matters for graduate students: At the graduate level, you will not only formulate your own research questions but also evaluate others’ questions as a reviewer, committee member, or collaborator. The ability to diagnose weaknesses in question formulation, and to propose constructive revisions, is a core professional skill. This exercise trains that skill on real published work rather than textbook examples.

Looking Ahead

Chapter 8 (The Ethics of Inquiry) addresses the ethical responsibilities that accompany research. Before you collect data, code content, or survey participants, you need to understand the principles that govern ethical research: respect for persons, beneficence, and justice. You’ll examine landmark cases that shaped modern research ethics, navigate the institutional review process, and confront the specific ethical questions that arise in content analysis, survey research, and digital media research. Ethics isn’t a bureaucratic hurdle; it’s a foundational commitment that shapes every decision you make as a researcher.

--- title: "Chapter 7: The Research Question" --- ## Learning Objectives - Distinguish between weak and strong research questions - Recognize and avoid common pitfalls in question formulation - Translate theoretical concepts into operational definitions - Write research questions that are specific, measurable, and answerable - Understand the relationship between research questions and project scope --- ![Crossroads representing research decision points](../images/ch07-crossroads_3.jpg){fig-alt="Crossroads representing research decision points"} --------------------------------------------------------------------- The hardest part of research isn't analyzing data or writing the final report. It's asking the right question. This seems counterintuitive. Students arrive at research methods courses assuming the challenge will be statistical tests or coding schemes, technical skills they haven't yet learned. But most students can learn to run a chi-square test or code qualitative data with practice. What's harder is the intellectual work that precedes analysis: formulating a question that's specific enough to answer, broad enough to matter, and grounded enough in existing knowledge to avoid reinventing the wheel. A weak research question dooms the entire project. You can have perfect data, sophisticated methods, and rigorous analysis, but if the question itself is vague, unanswerable, or trivial, the study contributes nothing. Consider these two questions: **Question A**: "How does music affect people?" **Question B**: "Do songs with negative lyric sentiment achieve higher peak chart positions than songs with positive sentiment, controlling for tempo and artist popularity?" Question A is a career-spanning research program disguised as a single question. You could write a thousand dissertations exploring different aspects of it and still not exhaust the topic. It's so broad that it's essentially unanswerable within any realistic scope. Question B is researchable. It specifies what will be measured (lyric sentiment, chart position), what will be compared (negative vs. positive), and what will be controlled (tempo, artist popularity). You can design a study around it. You can determine what evidence would answer it. You can know when you're done. The same distinction operates in every domain. "How does news affect democracy?" is a Question A. "Does exposure to episodic framing of poverty reduce support for structural antipoverty programs compared to thematic framing among registered voters?" is a Question B. The craft of research question formulation is the craft of getting from A to B. ## What Makes a Research Question "Good"? A strong research question meets five criteria. It's worth examining each carefully, because students often violate these principles without realizing it. ### 1. It's Specific Weak questions use vague language that could mean different things to different people: **Weak**: "How does social media influence politics?" What does "influence" mean? Change voting behavior? Shape discourse? Affect polarization? Mobilize activism? "Social media" is equally vague: Twitter? TikTok? All platforms? And "politics": electoral politics? Policy debates? Political identity? **Stronger**: "Does exposure to political content on TikTok increase political efficacy among Gen Z voters?" This narrows platforms (TikTok), specifies the outcome (political efficacy), and bounds the population (Gen Z voters). **For music research**: **Weak**: "What role does music play in identity?" **Stronger**: "How do fans of K-pop use parasocial language differently than fans of country music when discussing their favorite artists on Reddit?" ### 2. It's Measurable Some questions use concepts that resist operationalization, the translation into concrete, observable variables. **Weak**: "Do authentic artists create better music?" What does "authentic" mean in a measurable way? What counts as "better"? Without clear operational definitions, you can't design a study. **Stronger**: "Do artists who write their own lyrics receive higher critical ratings on Pitchfork than artists who use professional songwriters?" Now "authentic" is operationalized as "self-written lyrics" (verifiable) and "better" is operationalized as "Pitchfork rating" (quantifiable). You can disagree with these operationalizations, but at least they're concrete. **A non-music parallel**: "Is this news article biased?" is unmeasurable as stated. "Do CNN and Fox News differ in the proportion of episodic vs. thematic frames used in coverage of immigration?" is measurable. ### 3. It's Answerable Within Your Constraints Some questions are good in principle but impossible given your resources, timeline, or skills. **Weak for a semester project**: "How has hip-hop evolved globally from 1973 to 2024?" This requires decades of data, expertise in multiple languages and cultures, and years of analysis. **Stronger**: "How did framing of police violence in Billboard Hot 100 hip-hop songs change between two high-profile incidents: Ferguson (2014) and Minneapolis (2020)?" This focuses on two specific moments, one genre, one measurable dimension (framing), using accessible data (charting songs). ### 4. It's Not Already Definitively Answered Your literature review (Chapter 4) should reveal whether your question is genuinely open or has already been resolved. **Problematic**: "Does tempo affect perceived energy in music?" If your literature review shows this has been tested dozens of times with consistent results, asking it again without a novel angle (new population, new genre, testing a moderator) is redundant. **Better**: "Does the relationship between tempo and perceived energy differ between listeners with and without musical training?" This tests a potential moderator that previous research may not have examined. ### 5. It Matters The question should have intellectual or practical significance. It should contribute to theory, resolve a contradiction, or have applied value. **Weak**: "Do songs released on Tuesdays chart higher than songs released on Thursdays?" Even if true, so what? This is trivia, not insight. It doesn't test theory, address a gap, or have clear implications. **Stronger**: "Do songs released during culturally significant moments (post-tragedy, during social movements) achieve different chart longevity than songs released during neutral periods?" This has theoretical implications (cultural resonance, collective emotion) and practical value (understanding audience engagement). The same logic applies beyond music: "Do corporate apology statements released within 24 hours of a crisis receive more favorable media coverage than those released after 72 hours?" is a question with both theoretical and practical significance. ## Common Pitfalls in Question Formulation Certain mistakes appear repeatedly in student research proposals. Learning to recognize and avoid them accelerates the refinement process. ### Pitfall 1: The Impossible Comparison **Problematic**: "Are TikTok users happier than non-users?" This seems straightforward, but it's confounded by self-selection. People who are already predisposed to certain traits choose to use or avoid platforms. Correlation doesn't reveal causation. **Better**: "Among current TikTok users, does time spent on the platform correlate with self-reported well-being, controlling for baseline mood and social connectedness?" This is still correlational, but it's more honest about what it can demonstrate. Or, if you wanted causal inference: **Even better**: "Does random assignment to a two-week TikTok abstinence condition change self-reported well-being compared to a control group?" Now you have an experimental design that can support causal claims. ### Pitfall 2: The Unmeasurable Concept **Problematic**: "What is the essence of musical beauty?" "Essence" and "beauty" are philosophical concepts, not empirical variables. You can't code for "essence." **Better**: "What musical features (tempo, key, harmonic complexity) predict listener ratings of aesthetic pleasure?" You've operationalized "beauty" as "aesthetic pleasure" (measurable via ratings) and identified concrete features to test. ### Pitfall 3: The Everything Question **Problematic**: "How has music changed society over time?" This is a book series, not a research question. **Better**: "Did the rise of politically explicit hip-hop in the 1980s correlate with increased political engagement among Black youth, as measured by voter registration rates?" Specific genre, specific time period, specific outcome. ### Pitfall 4: The Circular Question **Problematic**: "Do popular songs become popular because people like them?" This is tautological. "Popular" and "people like them" are the same thing. **Better**: "Do songs with specific audio features (high energy, major key, fast tempo) achieve higher initial chart positions in the first week of release compared to songs lacking these features?" This tests whether specific features predict early success, which has theoretical implications about what drives initial popularity before word-of-mouth effects accumulate. ### Pitfall 5: The False Binary **Problematic**: "Is music or lyrics more important for chart success?" This assumes you must choose one. Reality is more complex. **Better**: "To what extent do lyric sentiment and audio features (tempo, energy) independently predict chart success, and do they interact?" This allows both to matter and tests whether their effects combine in non-additive ways. ## From Theoretical Concepts to Operational Definitions Theory gives you abstract concepts. Research requires concrete variables. The bridge is **operationalization**, defining how you'll measure theoretical constructs. ### Example: Parasocial Relationships **Theoretical concept**: Parasocial relationship (Horton & Wohl, 1956), a one-sided emotional bond between a fan and a media figure. **How do you measure it?** **Option 1: Self-Report Scale** Ask fans to rate agreement with statements like: - "I feel like [artist] understands me." - "I would be devastated if [artist] retired." - "I think about [artist] even when not listening to their music." Sum the ratings to create a parasocial bond score. **Pros**: Captures subjective experience directly. **Cons**: Subject to social desirability bias. People may over- or under-report. **Option 2: Behavioral Indicators** Measure observable behaviors: - Number of concerts attended - Frequency of social media interactions with artist content - Time spent in fan communities **Pros**: Behavioral data may be more reliable than self-report. **Cons**: Behavior doesn't necessarily reflect emotional bond (you might attend concerts for social reasons, not parasocial attachment). **Option 3: Language Analysis** Code fan-created content (forum posts, comments) for parasocial language: - First-person references to the artist ("I love how Taylor...") - Expressions of intimacy ("She gets me") - Emotional disclosure ("When I'm sad, I listen to...") **Pros**: Naturalistic data, reveals how fans actually talk. **Cons**: Labor-intensive coding. May miss fans who feel strongly but don't post publicly. **Which is best?** Depends on your research question, data access, and resources. The point is that "parasocial relationship" isn't directly observable; you must choose a measurable proxy and defend why it's valid. This logic applies identically to non-music concepts: "media trust," "political polarization," and "brand loyalty" all require the same translation from abstract construct to concrete measurement. ### Example: Lyric Sentiment **Theoretical concept**: Emotional valence of lyric content. **How do you measure it?** **Option 1: Automated Sentiment Analysis (LIWC)** Use software like Linguistic Inquiry and Word Count (LIWC) to score lyrics based on positive and negative emotion words. **Pros**: Fast, replicable, no coder training needed. **Cons**: Misses context (sarcasm, irony, metaphor). "I'm bad" in Michael Jackson's song is positive, not negative. Napier and Shamir (2018) used this approach across decades of Billboard lyrics but acknowledged these limitations. **Option 2: Human Coding** Train coders to read lyrics and categorize overall sentiment as positive, negative, or neutral based on holistic judgment. **Pros**: Captures context, nuance, metaphorical meaning. **Cons**: Labor-intensive. Requires inter-coder reliability testing (Lombard et al., 2002). Subjective. **Option 3: Dimensional Coding** Code for multiple dimensions: - Valence (positive/negative) - Arousal (high energy/low energy) - Dominance (powerful/submissive) **Pros**: Captures emotional complexity better than a single valence dimension. Aligns with psychological models of emotion like those proposed by Zentner, Grandjean, and Scherer (2008). **Cons**: More complex, requires careful codebook development. Again, no single "right" answer. You choose based on what your theory predicts and what your question requires. ## Writing Strong Research Questions: A Formula Here's a template that consistently produces clear, researchable questions: **[Does/Is there/How does] + [Specific Variable or Concept] + [Relationship/Pattern/Difference] + [Among/Between/Within] + [Bounded Population or Context] + [Controlling for/Accounting for Confounds, if applicable]?** ### Examples Using the Formula **Example 1 (Descriptive, music)**: "What is the **prevalence of mental health references** in **Billboard Hot 100 songs** between **2015 and 2024**?" - Specific variable: Mental health references - Pattern: Prevalence - Context: Billboard Hot 100, 2015-2024 **Example 2 (Relational, music)**: "Is there a **relationship between lyric sentiment** (positive, negative, neutral) and **chart peak position** among **pop songs in the Billboard Hot 100** (2015-2024), **controlling for tempo and artist popularity**?" - Variables: Lyric sentiment (IV), chart position (DV) - Relationship: Correlation or difference - Context: Pop songs, Billboard Hot 100, specific timeframe - Controls: Tempo, artist popularity **Example 3 (Comparative, music)**: "How does **parasocial language use** differ **between K-pop fans and country music fans** in **Reddit fan communities**, and does this difference persist when **controlling for age and gender**?" - Variables: Parasocial language - Comparison: K-pop vs. country fans - Context: Reddit fan communities - Controls: Age, gender **Example 4 (Descriptive, non-music)**: "What is the **prevalence of episodic vs. thematic framing** in **New York Times coverage of homelessness** between **2019 and 2024**?" - Specific variable: Frame type (episodic vs. thematic) - Pattern: Prevalence - Context: NYT coverage, specific timeframe **Example 5 (Relational, non-music)**: "Is there a **relationship between news frame type** (episodic vs. thematic) and **reader attribution of responsibility** (individual vs. structural) among **adults exposed to news about poverty**, **controlling for political ideology**?" - Variables: Frame type (IV), attribution (DV) - Relationship: Difference - Context: Adults, poverty news - Controls: Political ideology The formula works regardless of domain. Swap the variables and context; the structure holds. ## The Relationship Between Questions and Hypotheses Sometimes you pose research questions. Sometimes you pose hypotheses. Understanding when to use which is crucial. ### Use Research Questions (RQs) When: **1. You're exploring new territory** If no theory predicts the outcome and you're genuinely uncertain what the data will show, use an RQ. *Example*: "How do fans of electronic dance music use parasocial language differently than fans of classical music?" You don't have a theory predicting the direction of difference; you're exploring. **2. You're describing rather than explaining** Descriptive studies often use RQs because they're mapping terrain, not testing causal mechanisms. *Example*: "What percentage of top-charting songs contain political references?" **3. The literature is too contradictory to predict confidently** If previous studies show inconsistent patterns, an RQ is more honest than picking a side arbitrarily. *Example*: "Is there a relationship between lyric sentiment and chart success, given that existing studies have tracked aggregate trends in lyric negativity (Brand et al., 2019; Napier & Shamir, 2018) but have not tested whether sentiment predicts individual song performance?" ### Use Hypotheses (H) When: **1. Theory makes a specific prediction** If your theoretical framework predicts a direction or pattern, state it as a hypothesis. *Example*: "H1: Songs in minor keys will be rated as sadder than songs in major keys (derived from appraisal theory; Scherer, 2004)." **2. Previous research suggests a likely outcome** If the literature consistently finds a pattern, you can hypothesize it will replicate in your context. *Example*: "H1: Popular song lyrics will show increasing negative emotional expression over time, consistent with Brand et al. (2019) and Napier and Shamir (2018)." **3. You're testing a causal claim** Causal research typically uses hypotheses because you're predicting that X causes Y. *Example*: "H1: Exposure to sad music will decrease self-reported mood compared to a control condition." ## Revising Questions: The Iteration Process First drafts of research questions are rarely optimal. Expect to revise multiple times. ### Iteration Example: From Vague to Precise **Draft 1**: "How does music affect emotion?" **Problem**: Too broad. What music? What emotions? In what context? **Draft 2**: "How does sad music affect listeners' emotions?" **Problem**: "Sad music" is vague (minor key? sad lyrics? both?). "Affect emotions" is vague (make them sadder? provide catharsis?). **Draft 3**: "Do songs with negative lyric sentiment make listeners feel sadder after listening?" **Problem**: Good, but how are you measuring "sadder"? Self-report? And what's the comparison? Sadder than before? Sadder than a control condition? **Draft 4**: "Does listening to songs with negative lyric sentiment decrease self-reported mood compared to listening to songs with neutral sentiment?" **Much better**: Specific stimulus (negative lyrics), specific outcome (self-reported mood), clear comparison (negative vs. neutral), measurable and testable. ## Testing Your Question: The Checklist Before finalizing your research question, run it through these diagnostics: **☐ Can I operationalize every concept?** Every abstract term (authenticity, beauty, emotion) must translate into something measurable. **☐ Is the scope manageable?** Can I answer this in one semester with available resources? **☐ Does my theory predict this pattern?** If using a hypothesis, does it actually follow from the theoretical framework? **☐ Would a null finding still be interesting?** If you find no relationship, does that tell us something valuable? Or is the question only interesting if the answer is "yes"? **☐ Can I specify what evidence would answer the question?** If you can't envision the data and analysis that would resolve the question, it's not ready. **☐ Does the question advance beyond description to explanation?** "What exists?" questions are fine for exploratory work, but "Why does it exist?" questions are typically more valuable. ------------------------------------------------------------------------ ## Practice: Crafting Research Questions ### Exercise 7.1: Diagnosing Weak Questions For each question below, identify what's wrong and propose a revision: **Question A**: "How does technology change communication?" **Problem(s)**: _______________ **Revision**: _______________ **Question B**: "Are happy people more successful?" **Problem(s)**: _______________ **Revision**: _______________ **Question C**: "What is the true meaning of rap music?" **Problem(s)**: _______________ **Revision**: _______________ **Question D**: "Does media coverage matter?" **Problem(s)**: _______________ **Revision**: _______________ ------------------------------------------------------------------------ ### Exercise 7.2: Operationalizing Concepts Choose one abstract concept and develop three different operational definitions: **Concept: Musical "authenticity"** **Operationalization 1** (how you'd measure it): _______________ **Operationalization 2**: _______________ **Operationalization 3**: _______________ **Which is best for your research question, and why?** Now do the same for a non-music concept: **"media bias"** **Operationalization 1**: _______________ **Operationalization 2**: _______________ **Operationalization 3**: _______________ **What do you notice about the choices involved in operationalizing an abstract concept?** ------------------------------------------------------------------------ ### Exercise 7.3: From Vague to Precise Take this vague interest and narrow it through five iterations: **Starting point**: "I'm interested in music and mental health." **Iteration 1** (narrow the scope): _______________ **Iteration 2** (specify the relationship): _______________ **Iteration 3** (bound the population/context): _______________ **Iteration 4** (add measurement specificity): _______________ **Iteration 5** (final researchable question): _______________ ------------------------------------------------------------------------ ### Exercise 7.4: RQ or Hypothesis? For each scenario, decide whether to use an RQ or hypothesis, and write it: **Scenario A**: Social Identity Theory (Tajfel & Turner, 1979) predicts in-group favoritism. You hypothesize fans will rate their preferred genre higher than rival genres. **Use**: RQ or H? **Write it**: _______________ **Scenario B**: No research has examined how EDM fans talk about their favorite DJs compared to how jazz fans talk about their favorite musicians. You're exploring this. **Use**: RQ or H? **Write it**: _______________ **Scenario C**: Multiple studies show that popular lyrics have become more negative over time (Brand et al., 2019; Napier & Shamir, 2018). You expect this to replicate in your dataset. **Use**: RQ or H? **Write it**: _______________ **Scenario D**: You want to know whether nonprofit fundraising emails use more emotional language than for-profit marketing emails, but no prior research has examined this comparison. **Use**: RQ or H? **Write it**: _______________ ------------------------------------------------------------------------ ### Exercise 7.5: The Revision Process Take a weak research question you've written (or make one up). Revise it five times, addressing a different problem each time: **Draft 1** (original, weak): _______________ **Draft 2** (more specific): _______________ **Draft 3** (added measurement): _______________ **Draft 4** (bounded context): _______________ **Draft 5** (final, strong): _______________ **What improved most between Draft 1 and Draft 5?** ------------------------------------------------------------------------ ## Reflection Questions 1. **The Precision-Breadth Tradeoff**: Strong research questions are narrow and specific, which means they can't address all the interesting aspects of a topic. How do you decide which aspect to focus on and which to leave for future research? 2. **Operationalization Choices**: Every time you operationalize an abstract concept, you make choices that shape what you can discover. What gets lost when you reduce "musical beauty" to "listener ratings"? Is that loss acceptable, or does it fundamentally change the question? 3. **Your Own Question**: Write your current research question for this course project. Now critique it using the five criteria from this chapter. What needs revision? 4. **Cross-Domain Transfer**: Take your research question from this course and rewrite it for a completely different domain (news, health, politics, advertising). What changes? What stays the same? Does the translation reveal weaknesses in the original formulation? ------------------------------------------------------------------------ ## Chapter Summary This chapter focused on the craft of question formulation: - **Five criteria for strong RQs**: Specific, measurable, answerable, not already resolved, significant - **Common pitfalls**: Impossible comparisons, unmeasurable concepts, scope too broad, circular logic, false binaries - **Operationalization** translates theoretical concepts into measurable variables. Multiple valid operationalizations often exist; choose based on theory and resources. - **Formula for strong questions**: [Action verb] + [specific variable] + [relationship/pattern] + [bounded context] + [controls if applicable] - **RQs vs. Hypotheses**: Use RQs when exploring or when theory doesn't predict direction. Use hypotheses when theory makes specific predictions. - **Iteration is essential**: First drafts are rarely optimal. Expect to revise questions multiple times. - **The checklist**: Can you operationalize it? Is scope manageable? Does theory predict it? Would null findings matter? Can you specify what evidence answers it? ------------------------------------------------------------------------ ## Key Terms - **Hypothesis (H)**: Testable prediction derived from theory (directional statement) - **Null hypothesis (H₀)**: Prediction of no relationship or difference - **Operationalization**: Translating abstract concepts into concrete, measurable variables - **Research question (RQ)**: Open-ended or bounded inquiry when theory doesn't specify direction - **Scope**: The boundaries of what a study will examine (population, context, timeframe) ------------------------------------------------------------------------ ## References Brand, C. O., Acerbi, A., & Mesoudi, A. (2019). Cultural evolution of emotional expression in 50 years of song lyrics. *Evolutionary Human Sciences*, *1*, e11. https://doi.org/10.1017/ehs.2019.11 Horton, D., & Wohl, R. R. (1956). Mass communication and para-social interaction: Observations on intimacy at a distance. *Psychiatry*, *19*(3), 215-229. https://doi.org/10.1080/00332747.1956.11023049 Lombard, M., Snyder-Duch, J., & Bracken, C. C. (2002). Content analysis in mass communication: Assessment and reporting of intercoder reliability. *Human Communication Research*, *28*(4), 587-604. https://doi.org/10.1111/j.1468-2958.2002.tb00826.x Napier, K., & Shamir, L. (2018). Quantitative sentiment analysis of lyrics in popular music. *Journal of Popular Music Studies*, *30*(4), 161-176. https://doi.org/10.1525/jpms.2018.300411 Scherer, K. R. (2004). Which emotions can be induced by music? What are the underlying mechanisms? And how can we measure them? *Journal of New Music Research*, *33*(3), 239-251. https://doi.org/10.1080/0929821042000317822 Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin & S. Worchel (Eds.), *The social psychology of intergroup relations* (pp. 33-47). Brooks/Cole. Zentner, M., Grandjean, D., & Scherer, K. R. (2008). Emotions evoked by the sound of music: Characterization, classification, and measurement. *Emotion*, *8*(4), 494-521. https://doi.org/10.1037/1528-3542.8.4.494 ------------------------------------------------------------------------ ::: {.callout-note title="Graduate Extension" collapse="true"} **Required Task**: Locate three published empirical studies in communication journals that state explicit research questions or hypotheses. For each study: 1. **Extract** the stated RQ(s) or hypothesis/hypotheses exactly as written. 2. **Evaluate** each using the five criteria from this chapter (specific, measurable, answerable, not already resolved, significant). Assign a rating (strong, adequate, weak) for each criterion and justify your assessment in 1-2 sentences. 3. **Identify the operational definitions** used for the key variables. Are they explicitly stated? Could another researcher replicate the measurement from the information provided? 4. **Assess theory-question alignment**: Does the RQ or hypothesis follow logically from the theoretical framework cited? Or is there a gap between what the theory predicts and what the question actually tests? 5. **Revise** the weakest question of the three. Rewrite it using the formula from this chapter, making it more specific, better operationalized, or more tightly connected to theory. **Why this matters for graduate students**: At the graduate level, you will not only formulate your own research questions but also evaluate others' questions as a reviewer, committee member, or collaborator. The ability to diagnose weaknesses in question formulation, and to propose constructive revisions, is a core professional skill. This exercise trains that skill on real published work rather than textbook examples. ::: ------------------------------------------------------------------------ ## Looking Ahead Chapter 8 (The Ethics of Inquiry) addresses the ethical responsibilities that accompany research. Before you collect data, code content, or survey participants, you need to understand the principles that govern ethical research: respect for persons, beneficence, and justice. You'll examine landmark cases that shaped modern research ethics, navigate the institutional review process, and confront the specific ethical questions that arise in content analysis, survey research, and digital media research. Ethics isn't a bureaucratic hurdle; it's a foundational commitment that shapes every decision you make as a researcher.