Notes on "The Science in Social Science" – Pages 34-43
Page 34
Principle: State theories precisely & concretely
Vague hypotheses merely obfuscate; precise ones can be disproved and thereby improved.
Principle of parsimony
Word used incoherently in casual/scholarly talk; clarified by Jeffreys (1961):
“Simple theories have higher prior probabilities.” → AKA Jeffreys-Wrinch Simplicity Postulate (akin to Occam’s razor).
Parsimony = assumption about the world’s simplicity, not a universal research design rule.
Appropriate only when we already know the domain is simple (e.g., some physics fields).
In social sciences, useful only occasionally; excessive insistence unwarranted.
Avoiding over-complexity follows directly from another maxim: make theory only as complex as evidence warrants.
‘Indeterminate research designs’ emerge when evidence is insufficient relative to theory complexity (see §4.1) — a research-design, not ontological, problem.
Pre-versus post-data collection
Above advice presumes no data collected yet.
If data already gathered, researchers can still revise theory & collect more data → expensive, time-consuming, wastes earlier data.
Problem: theory needs improvement but cannot collect new data.
Situation common → requires “great caution and self-restraint.”
Human minds excel at pattern recognition—even in randomness—so post-hoc theories risk being ‘wildly wrong.’
Ad hoc adjustments must be rare & disciplined.
Page 35
Two rules for post-hoc theory modification when new data are unaffordable
Less-restrictive modification allowed
Extend hypothesis to broader range so that theory faces more falsification opportunities.
Example: from “democracies with advanced welfare don’t fight” → “all modern democracies don’t fight.”
More-restrictive modification generally inappropriate
Adding qualifications after inspecting data smells like saving a bad theory.
E.g., restricting “modern democracies don’t fight” to only those with advanced welfare after exceptions found.
Likewise, bolting on multiple vague caveats (economic structure, climate, repressive leadership, etc.) = disguised admission: “my theory is correct, except in country x.”
Without new data these patches have no admissible evidence.
When stuck (no new data possible): accept being wrong; solid negative findings > flimsy positive ones.
After admitting defeat
Add sections on future research & speculative modifications; freedom to propose new conditions, theories, data designs—but label as speculative until tested.
Creativity vs. rules
Social science not rule-bound; data can inspire theoretical insights even if restrictive; acceptable if authors credibly show that such modification could have been made pre-data.
Until new data, status of revised theory = very uncertain.
Pilot projects
Valuable especially when data collection costly (interviews, etc.).
Preliminary data may reshape questions/theory; subsequent data then test new theory, avoiding double-dipping.
Page 36
Improving Data Quality
Definition: “Data” = systematically collected qualitative/quantitative info.
Data often gathered before precise research question; still need quality rules.
Guideline 1: Record & report the data-generation process
Necessary for detecting bias and enabling valid inferences.
Quantitative: sampling frame, wording of questions.
Qualitative: explicit case-selection rules.
Fear of sharing data is unwarranted; publication & sharing secure credit and spur citations.
Guideline 2: Collect data on as many observable implications as possible
More diverse contexts → stronger evaluation of theory.
Two expansion paths:
a. More observations on same DV (finer time or geography).
b. Additional dependent variables predicted by theory.Example: Rational deterrence theory
Direct test: attack decisions given threats (Huth 1988).
But must also include cases where threats deterred → else selection bias.
Consider lab experiments, oligopolistic firms, organized-crime analogies—cross-domain support builds plausibility.
Practical constraint → infinite data impossible; scholars risk ruin by over-collection; balance needed.
Guideline 3: Maximize measurement validity
Measure what you think you measure; stick close to observed indicators; beware of alternative meanings (e.g., “ignorance” reply in repressive regimes).
Guideline 4: Ensure reliability
Procedures must yield same measurement upon repetition (same researcher, different times, or different coders).
Achieved via explicit coding rules, double-coding, test-retest.
Guideline 5: Strive for replicability of both data & reasoning
Full detail so others can duplicate data & logic.
Quantitative: start with same dataset → replicate analysis (often harder than expected; see Dewald et al. 1986).
Qualitative: footnotes, bibliographic essays, accessible archives; share field notes/audio when possible.
Extensive example: Middletown sociological studies (1929, 1937, 1980s replications).
Scholars with privileged access should secure similar future access for others.
Page 37
Replicability rarely executed but always prepare as if; aids reader evaluation.
Improving Use of Existing Data
Better to collect new data, but often impossible → must make best of flawed data.
First principle: aim for unbiased inferences (correct on average).
Beware selection bias—cases chosen distort population (deliberate or subtle).
Beware omitted-variable bias—missing controls confound causality.
Second principle: seek efficiency—use all information to maximize inference precision.
Use disaggregated data when available; finer units add information albeit with higher uncertainty.
Page 38
Themes of the Volume (intro to §1.3 begins) — only headings summarized on current pages (full discussion continues pp. 38–43):
Using Observable Implications to connect theory & data.
Maximizing Leverage: explain as much as possible with as little.
Page 39
Theme 1 elaborated: tight linkage between theory & empirical inquiry.
Good theories yield observable implications; good data collection guided by those implications.
Questions to ask: What are the theory’s observable implications? Are observations relevant to them?
Page 40
Theme 2 elaborated: search for additional implications increases leverage.
Leverage ≠ parsimony; it is about ratio of explained variance to information used.
Areas with traditionally low leverage often attract qualitative methods; goal should be to raise leverage.
Techniques:
Improve theory for more implications.
Improve data to observe more implications.
Improve data-use to extract more information.
Page 41
Cross-level observations can raise leverage; data need not match theory’s aggregation.
Ecological fallacy warning (Robinson 1950) acknowledged but shouldn’t prohibit using multi-level data if informative.
Example: revolution theory—collect individual interviews, small-community reactions, leader statements.
One aggregate observation (revolution occurs) = only one implication; deeper digging distinguishes theories.
Page 42
Theme 3: Reporting Uncertainty
All inference is uncertain; qualitative as error-prone as quantitative.
Good social scientists estimate and report uncertainty; qualitative research often fails here.
Neustadt & May (1986) heuristic: “How much of your own money would you wager—and at what odds?”
Page 43
Theme 4: Thinking like a Social Scientist — skepticism & rival hypotheses.
On any causal claim, ask about data accuracy, alternative explanations, reversed causality.
Process of causal inference = successive approximations; each conclusion sparks further inquiry.
Example: lower heart-attack rates in Japan → plausible “less red meat” explanation checked against confounders (genetics, lifestyle, reverse causation).
Principle: State theories precisely & concretely- Vague hypotheses merely obfuscate; precise ones can be disproved and thereby improved.
Principle of parsimony- Word used incoherently in casual/scholarly talk; clarified by Jeffreys (1961):
“Simple theories have higher prior probabilities.” → AKA Jeffreys-Wrinch Simplicity Postulate (akin to Occam’s razor).
Parsimony = assumption about the world’s simplicity, not a universal research design rule.
Appropriate only when we already know the domain is simple (e.g., some physics fields).
In social sciences, useful only occasionally; excessive insistence unwarranted.
Avoiding over-complexity follows directly from another maxim: make theory only as complex as evidence warrants.
‘Indeterminate research designs’ emerge when evidence is insufficient relative to theory complexity (see §4.1) — a research-design, not ontological, problem.
Pre-versus post-data collection- Above advice presumes no data collected yet.
If data already gathered, researchers can still revise theory & collect more data → expensive, time-consuming, wastes earlier data.
Problem: theory needs improvement but cannot collect new data.- Situation common → requires “great caution and self-restraint.”
Human minds excel at pattern recognition—even in randomness—so post-hoc theories risk being ‘wildly wrong.’
Ad hoc adjustments must be rare & disciplined.
Two rules for post-hoc theory modification when new data are unaffordable
Less-restrictive modification allowed
Extend hypothesis to broader range so that theory faces more falsification opportunities. - Example: from “democracies with advanced welfare don’t fight” → “all modern democracies don’t fight.”
More-restrictive modification generally inappropriate
Adding qualifications after inspecting data smells like saving a bad theory. - E.g., restricting “modern democracies don’t fight” to only those with advanced welfare after exceptions found.
Likewise, bolting on multiple vague caveats (economic structure, climate, repressive leadership, etc.) = disguised admission: “my theory is correct, except in country x.”
Without new data these patches have no admissible evidence.
When stuck (no new data possible): accept being wrong; solid negative findings > flimsy positive ones.
After admitting defeat- Add sections on future research & speculative modifications; freedom to propose new conditions, theories, data designs—but label as speculative until tested.
Creativity vs. rules- Social science not rule-bound; data can inspire theoretical insights even if restrictive; acceptable if authors credibly show that such modification could have been made pre-data.
Until new data, status of revised theory = very uncertain.
Pilot projects- Valuable especially when data collection costly (interviews, etc.).
Preliminary data may reshape questions/theory; subsequent data then test new theory, avoiding double-dipping.
Improving Data Quality- Definition: “Data” = systematically collected qualitative/quantitative info.
Data often gathered before precise research question; still need quality rules.
Guideline 1: Record & report the data-generation process
Necessary for detecting bias and enabling valid inferences.
Quantitative: sampling frame, wording of questions.
Qualitative: explicit case-selection rules.
Fear of sharing data is unwarranted; publication & sharing secure credit and spur citations.
Guideline 2: Collect data on as many observable implications as possible- More diverse contexts → stronger evaluation of theory.
Two expansion paths:
a. More observations on same DV (finer time or geography).
b. Additional dependent variables predicted by theory.
Example: Rational deterrence theory
Direct test: attack decisions given threats (Huth 1988).
But must also include cases where threats deterred → else selection bias.
Consider lab experiments, oligopolistic firms, organized-crime analogies—cross-domain support builds plausibility.
Practical constraint → infinite data impossible; scholars risk ruin by over-collection; balance needed.
Guideline 3: Maximize measurement validity- Measure what you think you measure; stick close to observed indicators; beware of alternative meanings (e.g., “ignorance” reply in repressive regimes).
Guideline 4: Ensure reliability- Procedures must yield same measurement upon repetition (same researcher, different times, or different coders).
Achieved via explicit coding rules, double-coding, test-retest.
Guideline 5: Strive for replicability of both data & reasoning- Full detail so others can duplicate data & logic.
Quantitative: start with same dataset → replicate analysis (often harder than expected; see Dewald et al. 1986).
Qualitative: footnotes, bibliographic essays, accessible archives; share field notes/audio when possible.
Extensive example: Middletown sociological studies (1929, 1937, 1980s replications).
Scholars with privileged access should secure similar future access for others.
Replicability rarely executed but always prepare as if; aids reader evaluation.
Improving Use of Existing Data - Better to collect new data, but often impossible → must make best of flawed data.
First principle: aim for unbiased inferences (correct on average).
Beware selection bias—cases chosen distort population (deliberate or subtle).
Beware omitted-variable bias—missing controls confound causality.
Second principle: seek efficiency—use all information to maximize inference precision.
Use disaggregated data when available; finer units add information albeit with higher uncertainty.
Themes of the Volume (intro to §1.3 begins) — only headings summarized on current pages (full discussion continues pp. 38–43):
Using Observable Implications to connect theory & data.
Maximizing Leverage: explain as much as possible with as little.
Theme 1 elaborated: tight linkage between theory & empirical inquiry. - Good theories yield observable implications; good data collection guided by those implications.
Questions to ask: What are the theory’s observable implications? Are observations relevant to them?
Theme 2 elaborated: search for additional implications increases leverage.- Leverage ≠ parsimony; it is about ratio of explained variance to information used.
Areas with traditionally low leverage often attract qualitative methods; goal should be to raise leverage.
Techniques:
Improve theory for more implications.
Improve data to observe more implications.
Improve data-use to extract more information.
Cross-level observations can raise leverage; data need not match theory’s aggregation.- Ecological fallacy warning (Robinson 1950) acknowledged but shouldn’t prohibit using multi-level data if informative.
Example: revolution theory—collect individual interviews, small-community reactions, leader statements.
One aggregate observation (revolution occurs) = only one implication; deeper digging distinguishes theories.
Theme 3: Reporting Uncertainty - All inference is uncertain; qualitative as error-prone as quantitative.
Good social scientists estimate and report uncertainty; qualitative research often fails here. This includes acknowledging limitations in data collection, measurement, and theoretical scope.
The degree of uncertainty should be quantified where possible (e.g., -values, confidence intervals, standard errors) or clearly articulated through qualitative caveats and discussions of alternative interpretations.
Neustadt & May (1986) heuristic: “How much of your own money would you wager—and at what odds?” This heuristic encourages researchers to honestly assess the subjective probability of their findings being correct and the potential for error, thereby fostering greater transparency in reporting.
Theme 4: Thinking like a Social Scientist — skepticism & rival hypotheses.- On any causal claim, ask about data accuracy, alternative explanations, and reversed causality. This critical approach ensures that conclusions are robust and not merely coincidental correlations.
The process of causal inference is a cycle of successive approximations; each conclusion sparks further inquiry, leading to refinement and deeper understanding.
Actively seeking out plausible rival hypotheses is crucial. For instance, if a theory suggests A causes B, a social scientist would consider if C causes B, if B causes A, or if A and B are both caused by an unobserved D.
Example: lower heart-attack rates in Japan → initial plausible “less red meat” explanation. A skeptical social scientist would then rigorously check this against confounders (genetics, lifestyle, reverse causation like healthier people choosing less red meat), seeking to rule out these alternatives before concluding causation. This iterative process of proposing and eliminating rival explanations strengthens the validity of the final claim.
How to Develop a Sound Research Question
Developing a sound research question in political science is a significant and often time-consuming process, crucial for effective scholarly analysis. It is distinct from typical undergraduate writing assignments.
What Makes for a Good Research Question?
A good research question possesses several key traits:
Non-normative and Answerable: Questions should seek to understand "what is," not "what ought to be." Avoid questions beginning with "should," as they often lead to position papers rather than scholarly analysis and rely on subjective beliefs or values. Empirical political scientists aim to test alternative explanations for phenomena, contributing to an understanding of how the world works.
Generates Implications for Real-World Problems: Although non-normative, a question should connect to significant real-world issues (e.g., representation, justice). It should address the "so what?" question, demonstrating broader relevance beyond a specific, narrow context.
Addresses a Debate or Puzzle in the Literature: A sound question emerges from and contributes to existing scholarly literature. It should explain its potential to shed new light on an unresolved puzzle, problem, or debate.
Not Overly Broad: Initial research ideas may be grand but must be refined into manageable questions. For example, instead of "Why do countries go to war?," ask "Are authoritarian governments more likely to start wars than democracies?"
Not Too Narrow: A question should not be so specific that its answer is only relevant to a very small audience. For instance, instead of "Which campaign strategies were most effective in helping Candidate Smith win more votes?," reframe as "Which campaign strategies are most effective in helping state legislative candidates win more votes?" (even if the study focuses on a few campaigns).
Beginning the Research Process: What Do You Want to Know?
Adopt a Suitable Frame of Mind: Aim to formulate a question you genuinely want to know the answer to, rather than one you want to "prove." Ideally, you do not know the answer with certainty.
Record Keeping: Maintain a dedicated research notebook or file to collect ideas, leads, sources (with citations and dates), iterations of your question, and notes. This stimulates thinking, helps maintain momentum, and ensures the integrity of your final work (e.g., by quoting/citing carefully).
Consider Feasibility: Take into account available time, familiarity with the topic, faculty mentorship, and accessibility of data and methods. While important, do not let potential data/methodological challenges discourage an interesting question, as many questions can be reframed for various data sources.
Generate Preliminary Questions: Reflect on personal interests, current events, expand on previous papers, and consult with faculty or graduate students. Reading review essays (e.g., in Annual Review of Political Science) or handbooks (e.g., Oxford/Cambridge Handbook series) can provide shortcuts into subfields and identify research directions.
What Do Scholars Already Know? The Core of a Research Question: What Is the Controversy, Debate, or Puzzle?
Question Formulation is a Process: It requires continuous refinement, especially for new researchers, as you engage with existing scholarship.
Address a Controversy, Debate, or Puzzle: A good question identifies something overlooked, under-studied, or debated within the academic literature, or a puzzle observed outside academia that can be illuminated by scholarly examination. Consulting faculty is highly recommended.
Question Development: Intermediate Stages
Study How Scholars Articulate Questions: Examine journal article abstracts and introductions to see how authors move from broad observations to concise, focused questions. Look for common phrases like "To what extent does…", "Under what conditions do…", or "Given X, what accounts for Y?"
Engage with Journal Articles: Move from skimming to fully reading relevant articles. Scan bibliographies for leads, note continually cited authors to identify ongoing scholarly conversations. Focus on recent articles to grasp current debates and findings, and identify inconsistencies or opposing arguments.
Question Development: Advanced Stages
Reached when:
You have enough information for a clear, specific question.
The question has a variety of possible answers.
It addresses a debate or puzzle in the literature.
It is manageable given your skills, knowledge, and time.
It is guided by prior scholarship and can contribute to it.
At this stage, identify the specific community of scholars asking similar questions by tracking repeated citations. This helps understand debates, methods, central theories, and how questions are articulated within that circle. This groundwork provides a head start on theory, hypotheses, case selection, data sources, and methods.
Keeping the Big Picture in Mind: How Will You Execute the Study?
Consider Methods and Data Early: Think about how your question might be answered and what data you might use. This helps refine the question's scope and allows you to identify suitable data sources (e.g., datasets, interviews, surveys, experiments, case studies).
Creative Approaches: Don't feel limited by current skill sets; basic statistical or qualitative analysis techniques can be learned during the project.
Alternative Theories
As you develop your question, avoid becoming overly attached to one explanation. Social scientists explicitly and fairly test rival theories to advance knowledge, rather than trying to "prove" a preferred one.
Definitions
Use common terminology from the literature rather than reinventing definitions. Note inconsistencies in definitions, as they might signal scholarly disagreements that can inspire a research question.
Summing Up: The Research Question
The research question is central to political science research and requires time and effort to refine. It should relate to and contribute to scholarly conversations. The iterative process of question development means it will likely be adjusted as you learn more about your subject and available data. This careful foundation will prepare you for developing theories and hypotheses and structuring your entire study.
Key Terms
Literature Review: A comprehensive summary and critical analysis of existing scholarly works related to a research topic, usually published in academic journals and books.
Normative Implications: The prescriptive or value-laden aspects of a research question or finding, concerning what should be, as opposed to what is.
Peer-Reviewed: Research that has undergone a vetting process by multiple anonymous experts in the field who critique the work and recommend publication or rejection.
Review Essays: Scholarly articles that summarize and critically evaluate the state of research in a specific area, often identifying directions for future study.
Scholarly Literature: Academic works, typically peer-reviewed journal articles and books by doctoral candidates or those holding doctorates, published by university or reputable commercial presses.
“So What” Question: A core inquiry in academic research that demands justification for a study’s broader relevance and importance, explaining why potential findings matter to scholars and real-world phenomena.
How to Develop a Sound Research Question
Developing a sound research question in political science is a significant and often time-consuming process, crucial for effective scholarly analysis. It is distinct from typical undergraduate writing assignments, which may focus on argumentation rather than empirical inquiry. The time and effort required for an initial research idea to evolve into a bona fide research question can be surprising, especially for those new to social science academic standards. While unusual political or public policymaking events can spark intellectual curiosity, the scholarly research question is structured and arrived at through a different, more rigorous process.
Students often underestimate the time needed to craft a research question, prioritizing gathering research, data analysis, and writing. However, experienced researchers emphasize that considerable time and care must be invested in investigating the scholarly literature and other relevant material to ensure the question is both important and feasible. A poorly formulated question, regardless of the effort to answer it, is unlikely to yield useful results for other scholars and can derail a project. The justification for a study, the literature review, and the research design are all intricately linked to the research question. While the research process is non-linear and revisions are common, a strong initial question facilitates smoother execution and produces more valuable results.
What Makes for a Good Research Question?
Most research questions are ignited by personal passion or interest, such as an interest in a political system or an observation of under-representation. While personal enthusiasm is an excellent starting point, moving from a general topic to a sound research question requires understanding its key traits:
Non-normative and Answerable: Research questions should inquire “what is,” not “what ought to be.” Questions beginning with “should” (e.g., “Should the United States invade Iraq?” or “Should there be more women in Congress?”) tend to lead to position papers rather than scholarly analyses. They often rely on subjective beliefs, values, and political contexts, assuming ideal, universally applicable solutions where politics don't matter. Purely normative questions ask for judgment calls based on opinion rather than testable evidence. Empirically oriented political scientists, unlike philosophers or political theorists, aim to contribute to understanding how the world works by testing alternative explanations. Instead of asking, “Should there be more women in Congress?”, an empirical political scientist might ask, “To what extent does the presence of women legislators influence agenda setting and policy outcomes in the U.S. Congress?”
Generates Implications for Real-World Problems: Although not normatively constructed, a research question must connect to significant, broader issues like representation and justice. It must address the “so what?” question, demonstrating its relevance beyond a narrow instance (e.g., why a specific political candidate lost). For example, while fairness is a normative concern, researchers can empirically ask if legislatures with more women produce different policies, thereby allowing for discussion of normative implications (e.g., why male legislator over-representation matters).
Addresses a Debate or Puzzle in the Literature: A sound question emerges from and contributes to existing scholarly literature. It should explain its potential to shed new light on an unresolved puzzle, problem, or debate within academic research.
Not Overly Broad: Research questions must be manageable. Initial grand questions (e.g., “Why do countries go to war?”) must be refined into more discrete, feasible inquiries (e.g., “Are authoritarian governments more likely to start wars than democracies?” or “How effective are treaties in preventing wars?”).
Not Too Narrow: A question should not be so specific that its answer is only relevant to a very small audience. For instance, instead of focusing on a single candidate's campaign strategies (“Which campaign strategies were most effective in helping Candidate Smith win more votes?”), broaden it to be more generally applicable (“Which campaign strategies are most effective in helping state legislative candidates win more votes?”), even if the study still focuses on a few campaigns.
Beginning the Research Process: What Do You Want to Know?
To develop a research question, consider these practical steps:
Adopt a Suitable Frame of Mind: Your goal is to formulate a question whose answer you genuinely want to know, rather than one you aim to “prove.” Ideally, you should not know the answer with certainty, even if you have a strong hunch.
Record Keeping: Establish a dedicated research notebook or word-processing file from the project's outset. Use it to collect ideas, leads, sources (with full citations, page numbers, and dates of access), iterations of your question, and meeting notes. This practice helps stimulate thinking, suggests fruitful avenues, and maintains momentum, especially during breaks in work. Following advice from Ernest Hemingway and Cory Doctorow, always stop work with a “rough edge”—knowing your next concrete steps or even mid-sentence—to avoid getting “stuck.” Diligent record-keeping also ensures the integrity of your final work product by preventing accidental plagiarism from cut-and-paste content or mixing paraphrased with verbatim text without proper attribution.
Consider Feasibility: Assess your available time, familiarity with the topic, faculty mentorship opportunities, and the accessibility of data and methods. While a reasonable timeline is crucial, do not let potential data/methodological challenges immediately discourage an interesting question, as many questions can be reframed to utilize various data sources and tools.
Generate Preliminary Questions: Reflect on personal interests, current events, or previous research papers that could be expanded. Consult with professors or graduate students in your department who work in relevant areas. Reading review essays (e.g., in Annual Review of Political Science) or handbooks (e.g., Oxford/Cambridge Handbook series) offers a shortcut to understand current scholarly debates and identify potential new research directions within specific subfields (e.g., comparative politics, American politics, international relations). Skimming introductions and conclusions of journal articles and academic books can also reveal motivating questions and suggestions for future studies.
What Do Scholars Already Know? The Core of a Research Question: What Is the Controversy, Debate, or Puzzle?
Once research interests are narrowed, the next step is carving out a sound, researchable question. This is an iterative process, especially for new researchers, as the question will continuously be refined while engaging with existing scholarship. A good question aims to address a controversy, debate, or puzzle that has been overlooked, under-studied, or is currently debated within academic literature. It may also stem from an external puzzle that scholarly examination can illuminate. Consulting a faculty member with research interests in your chosen area is highly recommended.
Preliminary questions (e.g., “Why aren’t there more women in political office in the United States?” or “How has social media transformed contemporary social movements?”) are good starting points but are often too broad. They need further refinement based on existing research. For instance, the question about women in political office points to a central puzzle in scholarship but needs to be honed in light of specific dimensions already studied.
Question Development: Intermediate Stages
To grasp how political scientists articulate research questions, study journal article abstracts and introductions. Observe how authors transition from a broad observation to a concise, answerable, and focused question. Common formulations include: “To what extent does…”, “Under what conditions do…”, or “Given X, what accounts for Y?” Practice formulating your questions using these phrases.
Move from skimming to fully reading relevant journal articles. Scan their bibliographies for leads, noting repeatedly cited authors, as this indicates ongoing scholarly conversations. Prioritize recent articles to quickly grasp current debates and findings. Identify inconsistencies, puzzles, and opposing arguments among scholars, as these often signal topics needing further study. For example, Mona Lena Krook’s book Quotas for Women in Politics concludes with specific directions for future research, which can be invaluable. When studying women’s political representation, scholars often narrow their focus to specific dimensions, such as the political campaign process, institutional or state-level characteristics, or the impact of women’s increased educational attainment. At this stage, intermediate questions might be formulated as: “To what extent does the role of money in politics affect women’s ability to attain political office as compared with men?” or “Under what conditions are women in Western industrialized nations likely to constitute a higher proportion of nationally elected officials?” However, these still need refinement; for example, the latter might require focusing on nations with comparable political systems due to research design limitations. The iterative nature of the research process means questions are frequently revised to match what is reasonably answerable given available resources.
Why Do We Care? Or the “So What” Question
Social scientists must justify their research questions to demonstrate their broader importance to scholars and connection to real-world political phenomena (e.g., democratic principles, political behavior, policy outcomes). This is often referred to as the “so what?” question, ensuring the findings have practical implications. Good research papers typically begin by stating the question and immediately justifying its significance (“This question is important because…”). The literature review further defends the question by showing its relation to ongoing academic conversations.
Question Development: Advanced Stages
Further refinement to an advanced stage occurs when:
You have gathered enough information to craft a clear question with a high level of specificity.
Your question has a variety of possible answers.
It addresses a debate or puzzle in the literature.
It is manageable given your skills, knowledge base, and time.
It is guided by prior scholarship and has the potential to contribute to it.
At this stage, you will focus on a narrower community of scholars asking similar questions, identified by frequently cited authors. This helps understand specific debates, methods, central theories, and question articulation within that scholarly circle. This groundwork provides a head start on developing theory, hypotheses, selecting cases, and choosing data sources and methods. For example, a refined question might be: “How does the sex of political candidates affect voting perceptions and behavior in Turkey?” (Matland & Tezcur, 2011), justified by Turkey’s unique democratic context as a Muslim-majority country.
Keeping the Big Picture in Mind: How Will You Execute the Study?
It is essential to consider how you might answer your research question and what data you might use early in the process. Habitually moving between question framing and execution considerations offers several benefits. It helps define a doable project scope that still contributes value to scholarship. As you read, note the data sources and methods employed by authors (e.g., small- techniques like interviews, surveys, experiments, or large- datasets with statistical techniques). Many authors share data. Don't feel limited by current skill sets; basic statistical or qualitative analysis techniques can be learned during the project.
Alternative Theories
While developing your question, take note of competing theories presented by various scholars. Your question will address an unresolved area in the literature. As you read, you will naturally form hunches about the answer, which are useful. However, avoid becoming overly attached to one explanation or set of hypotheses. Social scientists are obligated to explicitly and fairly test rival theories to advance knowledge, rather than trying to “prove” a preferred one.
Definitions
Use common terminology from the literature for phenomena or concepts. Make note of these definitions and their sources, using them in your work rather than creating new ones or prematurely revising existing ones. Reinventing definitions, especially for minor tweaks, wastes time, signals a lack of engagement with ongoing scholarly conversations, and makes it harder for others to build on common understandings. However, note inconsistencies in central concept definitions; these might indicate scholarly disagreements that could inspire a viable research question.
Summing Up: The Research Question
The research question is central to political science research. It requires time and effort to develop, as it should relate to and contribute to one or more scholarly conversations. The iterative nature of research means the question will likely be adjusted multiple times as you learn more about your subject and available data. Seek advice from faculty mentors. By following these strategies for beginning, intermediate, and advanced stages of question development, you will be well-prepared to situate your question within scholarly conversations (which implicitly involve competing theories) and to develop your own theories and hypotheses for your study.
Key Terms
Literature Review: A comprehensive summary and critical analysis of existing scholarly works related to a research topic, usually published in academic journals and books.
Normative Implications: The prescriptive or value-laden aspects of a research question or finding, concerning what should be, as opposed to what is.
Peer-Reviewed: Research that has undergone a vetting process by multiple anonymous experts in the field who critique the work and recommend publication or rejection. This process includes additional editorial scrutiny.
Review Essays: Scholarly articles that summarize and critically evaluate the state of research in a specific area, often identifying directions for future study.
Scholarly Literature: Academic works, typically peer-reviewed journal articles and books by doctoral candidates or those holding doctorates, published by university or reputable commercial presses.
“So What” Question: A core inquiry in academic research that demands justification for a study’s broader relevance and importance, explaining why potential findings matter to scholars and real-world phenomena.
Note on Scholarly Sources: Throughout this text, “scholarly” or “academic” literature refers to peer-reviewed social science journal articles and books written by doctoral candidates or those holding doctorates, published by university presses or reputable commercial presses
Page 34
Principle: State theories precisely & concretely
Vague hypotheses merely obfuscate; precise ones can be disproved and thereby improved.
Principle of parsimony
Word used incoherently in casual/scholarly talk; clarified by Jeffreys (1961):
“Simple theories have higher prior probabilities.” → AKA Jeffreys-Wrinch Simplicity Postulate (akin to Occam’s razor).
Parsimony = assumption about the world’s simplicity, not a universal research design rule.
Appropriate only when we already know the domain is simple (e.g., some physics fields).
In social sciences, useful only occasionally; excessive insistence unwarranted.
Avoiding over-complexity follows directly from another maxim: make theory only as complex as evidence warrants.
‘Indeterminate research designs’ emerge when evidence is insufficient relative to theory complexity (see §4.1) — a research-design, not ontological, problem.
Pre-versus post-data collection
Above advice presumes no data collected yet.
If data already gathered, researchers can still revise theory & collect more data → expensive, time-consuming, wastes earlier data.
Problem: theory needs improvement but cannot collect new data.
Situation common → requires “great caution and self-restraint.”
Human minds excel at pattern recognition—even in randomness—so post-hoc theories risk being ‘wildly wrong.’
Ad hoc adjustments must be rare & disciplined.
Page 35
Two rules for post-hoc theory modification when new data are unaffordable
Less-restrictive modification allowed
Extend hypothesis to broader range so that theory faces more falsification opportunities.
Example: from “democracies with advanced welfare don’t fight” → “all modern democracies don’t fight.”
More-restrictive modification generally inappropriate
Adding qualifications after inspecting data smells like saving a bad theory.
E.g., restricting “modern democracies don’t fight” to only those with advanced welfare after exceptions found.
Likewise, bolting on multiple vague caveats (economic structure, climate, repressive leadership, etc.) = disguised admission: “my theory is correct, except in country x.”
Without new data these patches have no admissible evidence.
When stuck (no new data possible): accept being wrong; solid negative findings > flimsy positive ones.
After admitting defeat
Add sections on future research & speculative modifications; freedom to propose new conditions, theories, data designs—but label as speculative until tested.
Creativity vs. rules
Social science not rule-bound; data can inspire theoretical insights even if restrictive; acceptable if authors credibly show that such modification could have been made pre-data.
Until new data, status of revised theory = very uncertain.
Pilot projects
Valuable especially when data collection costly (interviews, etc.).
Preliminary data may reshape questions/theory; subsequent data then test new theory, avoiding double-dipping.
Page 36
Improving Data Quality
Definition: “Data” = systematically collected qualitative/quantitative info.
Data often gathered before precise research question; still need quality rules.
Guideline 1: Record & report the data-generation process
Necessary for detecting bias and enabling valid inferences.
Quantitative: sampling frame, wording of questions.
Qualitative: explicit case-selection rules.
Fear of sharing data is unwarranted; publication & sharing secure credit and spur citations.
Guideline 2: Collect data on as many observable implications as possible
More diverse contexts → stronger evaluation of theory.
Two expansion paths:
a. More observations on same DV (finer time or geography).
b. Additional dependent variables predicted by theory.
Example: Rational deterrence theory
Direct test: attack decisions given threats (Huth 1988).
But must also include cases where threats deterred → else selection bias.
Consider lab experiments, oligopolistic firms, organized-crime analogies—cross-domain support builds plausibility.
Practical constraint → infinite data impossible; scholars risk ruin by over-collection; balance needed.
Guideline 3: Maximize measurement validity
Measure what you think you measure; stick close to observed indicators; beware of alternative meanings (e.g., “ignorance” reply in repressive regimes).
Guideline 4: Ensure reliability
Procedures must yield same measurement upon repetition (same researcher, different times, or different coders).
Achieved via explicit coding rules, double-coding, test-retest.
Guideline 5: Strive for replicability of both data & reasoning
Full detail so others can duplicate data & logic.
Quantitative: start with same dataset → replicate analysis (often harder than expected; see Dewald et al. 1986).
Qualitative: footnotes, bibliographic essays, accessible archives; share field notes/audio when possible.
Extensive example: Middletown sociological studies (1929, 1937, 1980s replications).
Scholars with privileged access should secure similar future access for others.
Page 37
Replicability rarely executed but always prepare as if; aids reader evaluation.
Improving Use of Existing Data
Better to collect new data, but often impossible → must make best of flawed data.
First principle: aim for unbiased inferences (correct on average).
Beware selection bias—cases chosen distort population (deliberate or subtle).
Beware omitted-variable bias—missing controls confound causality.
Second principle: seek efficiency—use all information to maximize inference precision.
Use disaggregated data when available; finer units add information albeit with higher uncertainty.
Page 38
Themes of the Volume (intro to §1.3 begins) — only headings summarized on current pages (full discussion continues pp. 38–43):
Using Observable Implications to connect theory & data.
Maximizing Leverage: explain as much as possible with as little.
Page 39
Theme 1 elaborated: tight linkage between theory & empirical inquiry.
Good theories yield observable implications; good data collection guided by those implications.
Questions to ask: What are the theory’s observable implications? Are observations relevant to them?
Page 40
Theme 2 elaborated: search for additional implications increases leverage.
Leverage ≠ parsimony; it is about ratio of explained variance to information used.
Areas with traditionally low leverage often attract qualitative methods; goal should be to raise leverage.
Techniques:
Improve theory for more implications.
Improve data to observe more implications.
Improve data-use to extract more information.
Page 41
Cross-level observations can raise leverage; data need not match theory’s aggregation.
Ecological fallacy warning (Robinson 1950) acknowledged but shouldn’t prohibit using multi-level data if informative.
Example: revolution theory—collect individual interviews, small-community reactions, leader statements.
One aggregate observation (revolution occurs) = only one implication; deeper digging distinguishes theories.
Page 42
Theme 3: Reporting Uncertainty
All inference is uncertain; qualitative as error-prone as quantitative.
Good social scientists estimate and report uncertainty; qualitative research often fails here. This includes acknowledging limitations in data collection, measurement, and theoretical scope.
The degree of uncertainty should be quantified where possible (e.g., -values, confidence intervals, standard errors) or clearly articulated through qualitative caveats and discussions of alternative interpretations.
Neustadt & May (1986) heuristic: “How much of your own money would you wager—and at what odds?” This heuristic encourages researchers to honestly assess the subjective probability of their findings being correct and the potential for error, thereby fostering greater transparency in reporting.
Page 43
Theme 4: Thinking like a Social Scientist — skepticism & rival hypotheses.
On any causal claim, ask about data accuracy, alternative explanations, and reversed causality. This critical approach ensures that conclusions are robust and not merely coincidental correlations.
The process of causal inference is a cycle of successive approximations; each conclusion sparks further inquiry, leading to refinement and deeper understanding.
Actively seeking out plausible rival hypotheses is crucial. For instance, if a theory suggests A causes B, a social scientist would consider if C causes B, if B causes A, or if A and B are both caused by an unobserved D.
Example: lower heart-attack rates in Japan → initial plausible “less red meat” explanation. A skeptical social scientist would then rigorously check this against confounders (genetics, lifestyle, reverse causation like healthier people choosing less red meat), seeking to rule out these alternatives before concluding causation.
How to Develop a Sound Research Question
Developing a sound research question in political science is a significant and often time-consuming process, crucial for effective scholarly analysis. It is distinct from typical undergraduate writing assignments, which may focus on argumentation rather than empirical inquiry. The time and effort required for an initial research idea to evolve into a bona fide research question can be surprising, especially for those new to social science academic standards. While unusual political or public policymaking events can spark intellectual curiosity, the scholarly research question is structured and arrived at through a different, more rigorous process.
Students often underestimate the time needed to craft a research question, prioritizing gathering research, data analysis, and writing. However, experienced researchers emphasize that considerable time and care must be invested in investigating the scholarly literature and other relevant material to ensure the question is both important and feasible. A poorly formulated question, regardless of the effort to answer it, is unlikely to yield useful results for other scholars and can derail a project. The justification for a study, the literature review, and the research design are all intricately linked to the research question. While the research process is non-linear and revisions are common, a strong initial question facilitates smoother execution and produces more valuable results.
What Makes for a Good Research Question?
Most research questions are ignited by personal passion or interest, such as an interest in a political system or an observation of under-representation. While personal enthusiasm is an excellent starting point, moving from a general topic to a sound research question requires understanding its key traits:
Non-normative and Answerable: Research questions should inquire “what is,” not “what ought to be.” Questions beginning with “should” (e.g., “Should the United States invade Iraq?” or “Should there be more women in Congress?”) tend to lead to position papers rather than scholarly analyses. They often rely on subjective beliefs, values, and political contexts, assuming ideal, universally applicable solutions where politics don't matter. Purely normative questions ask for judgment calls based on opinion rather than testable evidence. Empirically oriented political scientists, unlike philosophers or political theorists, aim to contribute to understanding how the world works by testing alternative explanations. Instead of asking, “Should there be more women in Congress?”, an empirical political scientist might ask, “To what extent does the presence of women legislators influence agenda setting and policy outcomes in the U.S. Congress?”
Generates Implications for Real-World Problems: Although not normatively constructed, a research question must connect to significant, broader issues like representation and justice. It must address the “so what?” question, demonstrating its relevance beyond a narrow instance (e.g., why a specific political candidate lost). For example, while fairness is a normative concern, researchers can empirically ask if legislatures with more women produce different policies, thereby allowing for discussion of normative implications (e.g., why male legislator over-representation matters).
Addresses a Debate or Puzzle in the Literature: A sound question emerges from and contributes to existing scholarly literature. It should explain its potential to shed new light on an unresolved puzzle, problem, or debate within academic research.
Not Overly Broad: Research questions must be manageable. Initial grand questions (e.g., “Why do countries go to war?”) must be refined into more discrete, feasible inquiries (e.g., “Are authoritarian governments more likely to start wars than democracies?” or “How effective are treaties in preventing wars?”).
Not Too Narrow: A question should not be so specific that its answer is only relevant to a very small audience. For instance, instead of focusing on a single candidate's campaign strategies (“Which campaign strategies were most effective in helping Candidate Smith win more votes?”), broaden it to be more generally applicable (“Which campaign strategies are most effective in helping state legislative candidates win more votes?”), even if the study still focuses on a few campaigns.
Beginning the Research Process: What Do You Want to Know?
To develop a research question, consider these practical steps:
Adopt a Suitable Frame of Mind: Your goal is to formulate a question whose answer you genuinely want to know, rather than one you aim to “prove.” Ideally, you should not know the answer with certainty, even if you have a strong hunch.
Record Keeping: Establish a dedicated research notebook or word-processing file from the project's outset. Use it to collect ideas, leads, sources (with full citations, page numbers, and dates of access), iterations of your question, and meeting notes. This practice helps stimulate thinking, suggests fruitful avenues, and maintains momentum, especially during breaks in work. Following advice from Ernest Hemingway and Cory Doctorow, always stop work with a “rough edge”—knowing your next concrete steps or even mid-sentence—to avoid getting “stuck.” Diligent record-keeping also ensures the integrity of your final work product by preventing accidental plagiarism from cut-and-paste content or mixing paraphrased with verbatim text without proper attribution.
Consider Feasibility: Assess your available time, familiarity with the topic, faculty mentorship opportunities, and the accessibility of data and methods. While a reasonable timeline is crucial, do not let potential data/methodological challenges immediately discourage an interesting question, as many questions can be reframed to utilize various data sources and tools.
Generate Preliminary Questions: Reflect on personal interests, current events, or previous research papers that could be expanded. Consult with professors or graduate students in your department who work in relevant areas. Reading review essays (e.g., in Annual Review of Political Science) or handbooks (e.g., Oxford/Cambridge Handbook series) offers a shortcut to understand current scholarly debates and identify potential new research directions within specific subfields (e.g., comparative politics, American politics, international relations). Skimming introductions and conclusions of journal articles and academic books can also reveal motivating questions and suggestions for future studies.
What Do Scholars Already Know? The Core of a Research Question: What Is the Controversy, Debate, or Puzzle?
Once research interests are narrowed, the next step is carving out a sound, researchable question. This is an iterative process, especially for new researchers, as the question will continuously be refined while engaging with existing scholarship. A good question aims to address a controversy, debate, or puzzle that has been overlooked, under-studied, or is currently debated within academic literature. It may also stem from an external puzzle that scholarly examination can illuminate. Consulting a faculty member with research interests in your chosen area is highly recommended.
Preliminary questions (e.g., “Why aren’t there more women in political office in the United States?” or “How has social media transformed contemporary social movements?”) are good starting points but are often too broad. They need further refinement based on existing research. For instance, the question about women in political office points to a central puzzle in scholarship but needs to be honed in light of specific dimensions already studied.
Question Development: Intermediate Stages
To grasp how political scientists articulate research questions, study journal article abstracts and introductions. Observe how authors transition from a broad observation to a concise, answerable, and focused question. Common formulations include: “To what extent does…”, “Under what conditions do…”, or “Given X, what accounts for Y?” Practice formulating your questions using these phrases.
Move from skimming to fully reading relevant journal articles. Scan their bibliographies for leads, noting repeatedly cited authors, as this indicates ongoing scholarly conversations. Prioritize recent articles to quickly grasp current debates and findings. Identify inconsistencies, puzzles, and opposing arguments among scholars, as these often signal topics needing further study. For example, Mona Lena Krook’s book Quotas for Women in Politics concludes with specific directions for future research, which can be invaluable. When studying women’s political representation, scholars often narrow their focus to specific dimensions, such as the political campaign process, institutional or state-level characteristics, or the impact of women’s increased educational attainment. At this stage, intermediate questions might be formulated as: “To what extent does the role of money in politics affect women’s ability to attain political office as compared with men?” or “Under what conditions are women in Western industrialized nations likely to constitute a higher proportion of nationally elected officials?” However, these still need refinement; for example, the latter might require focusing on nations with comparable political systems due to research design limitations. The iterative nature of the research process means questions are frequently revised to match what is reasonably answerable given available resources.
Why Do We Care? Or the “So What” Question
Social scientists must justify their research questions to demonstrate their broader importance to scholars and connection to real-world political phenomena (e.g., democratic principles, political behavior, policy outcomes). This is often referred to as the “so what?” question, ensuring the findings have practical implications. Good research papers typically begin by stating the question and immediately justifying its significance (“This question is important because…”). The literature review further defends the question by showing its relation to ongoing academic conversations.
Question Development: Advanced Stages
Further refinement to an advanced stage occurs when:
You have gathered enough information to craft a clear question with a high level of specificity.
Your question has a variety of possible answers.
It addresses a debate or puzzle in the literature.
It is manageable given your skills, knowledge base, and time.
It is guided by prior scholarship and has the potential to contribute to it.
At this stage, you will focus on a narrower community of scholars asking similar questions, identified by frequently cited authors. This helps understand specific debates, methods, central theories, and question articulation within that scholarly circle. This groundwork provides a head start on developing theory, hypotheses, selecting cases, and choosing data sources and methods. For example, a refined question might be: “How does the sex of political candidates affect voting perceptions and behavior in Turkey?” (Matland & Tezcur, 2011), justified by Turkey’s unique democratic context as a Muslim-majority country.
Keeping the Big Picture in Mind: How Will You Execute the Study?
It is essential to consider how you might answer your research question and what data you might use early in the process. Habitually moving between question framing and execution considerations offers several benefits. It helps define a doable project scope that still contributes value to scholarship. As you read, note the data sources and methods employed by authors (e.g., small- techniques like interviews, surveys, experiments, or large- datasets with statistical techniques). Many authors share data. Don't feel limited by current skill sets; basic statistical or qualitative analysis techniques can be learned during the project.
Alternative Theories
While developing your question, take note of competing theories presented by various scholars. Your question will address an unresolved area in the literature. As you read, you will naturally form hunches about the answer, which are useful. However, avoid becoming overly attached to one explanation or set of hypotheses. Social scientists are obligated to explicitly and fairly test rival theories to advance knowledge, rather than trying to “prove” a preferred one.
Definitions
Use common terminology from the literature for phenomena or concepts. Make note of these definitions and their sources, using them in your work rather than creating new ones or prematurely revising existing ones. Reinventing definitions, especially for minor tweaks, wastes time, signals a lack of engagement with ongoing scholarly conversations, and makes it harder for others to build on common understandings. However, note inconsistencies in central concept definitions; these might indicate scholarly disagreements that could inspire a viable research question.
Summing Up: The Research Question
The research question is central to political science research and requires time and effort to refine. It should relate to and contribute to scholarly conversations. The iterative process of question development means it will likely be adjusted as you learn more about your subject and available data. This careful foundation will prepare you for developing theories and hypotheses and structuring your entire study.
Key Terms
Literature Review: A comprehensive summary and critical analysis of existing scholarly works related to a research topic, usually published in academic journals and books.
Normative Implications: The prescriptive or value-laden aspects of a research question or finding, concerning what should be, as opposed to what is.
Peer-Reviewed: Research that has undergone a vetting process by multiple anonymous experts in the field who critique the work and recommend publication or rejection.
Review Essays: Scholarly articles that summarize and critically evaluate the state of research in a specific area, often identifying directions for future study.
Scholarly Literature: Academic works, typically peer-reviewed journal articles and books by doctoral candidates or those holding doctorates, published by university or reputable commercial presses.
“So What” Question: A core inquiry in academic research that demands justification for a study’s broader relevance and importance, explaining why potential findings matter to scholars and real-world phenomena.
Note on Scholarly Sources: Throughout this text, “scholarly” or “academic” literature refers to peer-reviewed social science journal articles and books written by doctoral candidates or those holding doctorates, published by university presses or reputable commercial presses
Linking Theory and Inference
Theory is the most crucial part of the research enterprise due to its centrality in making inferences. It informs every part of the research process, from formulating a research question to designing a study and interpreting results. As political scientists, our goal is to test, modify, or construct new theories to better explain phenomena. This is not done in isolation; we build upon prior scholarship, proposing new ideas or arguing against existing ones when they are incomplete or incorrect. While developing a grand theory is rare, making modest but significant theoretical contributions—such as showing how existing theories are incomplete, demonstrating the necessity of previously omitted variables, or resolving inconsistent findings—constitutes valid “middle range” theoretical contributions.
What Is Theory? Why Are Theories So Important and So Valuable?
A theory is fundamentally a generalization, defined as a set of principles that explains why people behave as they do across various contexts, or broadly, a statement about how one believes the world works. The purpose of scientific studies is not to explain single events but to develop theories that can be applied to other related phenomena. Theories provide a foundation of general knowledge, reducing complex observations into regular patterns and relationships, which can then be applied to past, present, or future problems. Well-developed theories serve as critical shortcuts or heuristics for scholars, policymakers, and the public in decision-making.
Individuals constantly rely on, construct, and apply theories in everyday life, often subconsciously. These can be inductively generated from personal observation (e.g., studying with distractions leads to lower grades) or deductively informed by existing knowledge (e.g., from trusted professionals or media). Without theories, every situation would require individual, repeated investigation. A toolkit of theories allows the application of knowledge from one context to another, saving immense time and resources (e.g., theories on Congressional operations preclude needing a new study for every new bill).
Beyond Generalizability, theory plays a special role in scholarly research by promoting sound research design. It is critical for:
Guiding researchers in determining which alternative theories to consider.
Setting the stage for developing interesting hypotheses to test.
Helping discern which factors (independent variables) to include and control for, and which way the causal arrow goes. Theory is the primary guide for identifying independent and dependent variables.
Ensuring the robustness of observed relationships, distinguishing them from spurious correlations.
Interpreting findings; data alone cannot explain patterns. Theory informs inferences to help interpret and explain results.
What Characterizes a Good Theory?
A good social science theory is a reasoned and precise speculation about the answer to a research question, stating why the proposed answer is correct. It usually implies several specific descriptive or causal hypotheses and must be consistent with prior evidence. Good theories are also defined as an interrelated set of constructs (variables) formed into propositions or hypotheses that specify relationships among variables, often in terms of magnitude or direction.
Good Theory Builds on Existing Theory
Conceptual and theoretical understandings in political science undergo a continuous process of refinement, as new theories emerge to question, refine, or replace older ones. Knowledge advances most efficiently when studies build upon prior scholarship. Theories that are “well grounded” in prior literature are valuable because they speak to the common interests and mutual understandings of others who are interested in the subject that the theory addresses. Well-grounded theories are therefore accessible to others. By drawing on prior theories and addressing ongoing conversations, a researcher is more likely to influence others’ thinking on the topic they are studying.
For example, in their article “Protest and democracy in Latin America’s market era,” Paul T. Bellinger, Jr. and Moises Arce aimed to understand “whether and how political democracy has influenced societal responses to economic liberalization.” They explicitly noted the contradictory implications of two existing theoretical streams: one emphasizing the “depoliticizing” effects of economic reforms in democracies (suppressing protest), and another, the “repoliticization” literature (promoting protest). To mediate these, they introduced a third literature on contentious politics, which posits that grievances increase mobilization willingness while democracy creates a favorable environment. This allowed Bellinger and Arce to formulate their own testable theory: democratic politics, even if imperfect, should encourage collective political activity, not render it obsolete. This example demonstrates how well-grounded theories are “leveraged” or applied across topics and situations by different scholars and policymakers.
Good Theory Concretely Specifies the Concepts and/or Variables It Invokes
Concepts are ideas represented by words and must be clearly defined in any research project. For instance, a scholar studying corruption must specify whether it refers to “petty administrative corruption” or “grand corruption by high-level officials,” as their causes differ. Precise definitions are crucial. It is generally advised to avoid substantially revising existing widely used definitions to ensure work is “leverageable” or applicable to other scholars, contributing to a common vocabulary and advancing knowledge, rather than just producing information. However, when scholars are divided over definitions, these differences can be leveraged to refine one's approach (e.g., the concept of representation as debated by Pitkin, Mansbridge, and Rehfeld).
Many common political science concepts (e.g., power, democracy, representation, equality, political efficacy) have multiple definitions across disciplines. Researchers should investigate existing definitions, consider their implications, and clearly state their chosen definition. Even seemingly self-evident terms like “voter turnout” require careful specification; defining turnout as a percentage of voting-age public vs. eligible voters significantly alters conclusions about its decline, demonstrating the important consequences of conceptual definition for findings.
Good Theory Clarifies the Relationship between Concepts and What Is to Be Explained or Described
Sound political science requires precise postulation of relationships. Whether descriptive or causal, theories must clearly describe the conclusion about the relationship between phenomena and the factors that explain, shape, influence, or cause them. Theories must state how the world works and why it works that way. For causal research questions, the theory should explain why independent variables are expected to cause changes in the dependent variable, elucidating the causal mechanisms (e.g., why education relates to voting involves explaining that more educated individuals tend to know more about politics, making vote decisions easier). These explanations are rooted in prior scholarship.
For descriptive inferences, the theory should describe how an examination of prior theories, combined with new observations, influences a debate or problem in the literature. For instance, Barakso’s study intervened in the debate on declining civic engagement by theorizing that organizational operation and internal democracy within groups, rather than just changes in group numbers or membership, influence civic participation. This study posited a clear theoretical relationship and laid out implications for future researchers.
Good Theory Is Falsifiable
Falsifiability, the ability for a theory to be proven wrong, is critical. Theories cannot be “proven” correct, but their soundness can be estimated through testing. A non-falsifiable theory (e.g., “Wars in Iraq and Afghanistan prevented another terrorist attack” because the counterfactual cannot be observed) prevents meaningful testing and evaluation. A falsifiable theory, conversely, allows for multiple tests of its validity (e.g., “military invasions of terrorist states reduce future terrorist attacks,” which can generate hypotheses like “worldwide terrorist incidents will diminish in the wake of an external military intervention”).
Some robust scholarly conversations revolve around non-falsifiable theories (e.g., deliberative democracy theory), often due to their engagement with salient political concepts. Critics argue that deliberative democracy, often conceived normatively, lacks empirical testing on its effectiveness or conditions for suboptimal outcomes. Vague and variable definitions also “insulate” it from refutation. Mutz suggests that instead of grand, overarching theories, scholars should develop and test “middle-range theories,” which are intermediate, precise, and falsifiable components of broader theories. This approach, by replacing vague entities with concrete concepts and requiring empirically grounded hypotheses, helps understand which elements are crucial to specific outcomes.
Good Theory Leads to Testable Hypotheses
Good theory specifies expected observations if the theory accurately describes how the world works. Whether descriptive or causal, theories must lead to specific, testable hypotheses—or implications. Testable hypotheses allow researchers to establish a theory's soundness. Furthermore, a theory that can generate multiple testable hypotheses, especially those extending beyond the immediate study, benefits the broader academic community by providing more avenues for exploration and knowledge improvement. For example, comparing a theory that “incumbents win re-election more often because they tend to have more money” (limited observable implications) with one stating “incumbents win re-election more often because they tend to have more resources” (many observable implications, including financial donations, name recognition, campaign workers, casework goodwill, and favorable media coverage) shows that a theory with more observable implications tends to be broader and more useful.
Incorporating Theory into Your Study: the Literature Review
The development of a research question is intricately tied to prior literature. The literature review is the practical means of incorporating others’ theories into a study. By this stage, a researcher has already gained tools to grasp and outline their literature review, as the question likely grew from competing theories or perceived shortcomings in existing literature, informing the researcher’s own views.
The researcher must decide whether to test an existing theory or propose a new explanation/theory, a decision often made while reading and refining the research question within the literature.
Thinking about the Literature Review
The literature review explains the logic of a study, grounded in prior research (theory). It reveals the main theories that justify the research question’s salience and particular formulation. It also specifies the key theories that led to the selection of a particular theory to explore or test, or if proposing a new theory, the shortcomings of prior theories that prompted the new contribution. It contains the theoretical justification for hypotheses (what variables are important and why) and concept definitions. Ultimately, theory informs the interpretation of findings.
The term “literature review” can be misleading, as its purpose is not to list every source read. Thinking of it as a “theory section” helps maintain focus on its purpose.
Three Goals of the Literature Review
Expanded Discussion of the Research Question
: Systematically and selectively discuss key problems, theories, and data that justify the salience, importance, and specific formulation of the research question.
Delineate Key Discussions and Debates
: Delineate the key discussions, debates, and data in the literature directly related to the question and the theory being examined or proposed. It should logically progress from general “big picture” concepts to specific debates leading to the proposed theory. Explicitly state (or restate) plausible alternative or rival theories that will be examined.
Present Working Answer/Theory and Hypotheses
: Present the researcher’s own working answer or theory, typically tested through hypotheses. Hypotheses are explicit statements of expected findings if the theory is correct, often worded as “if–then” or “when–then” statements (e.g., “Women’s estimation of the costs of running a political campaign are significantly less accurate than men’s.”). This component links to the data and methods section, explaining concept definitions and selected factors (variables) for analysis (e.g., voter pool, definition of “major scandal,” election laws, and other relevant factors from literature that might be omitted with justification). It also briefly reviews control variables (common factors like demographics or partisan affiliation).
Writing the Literature Review
The literature review is a complex but essential task, requiring thoughtful integration of all central project elements. It serves as a puzzle where each piece must interlock. There is no single formula for writing one, but common elements include discussing (and supporting with literature) the broader problem, its implications, the research question and its importance, theories, concepts, empirical evidence, hypotheses, one’s own theory, rival theories, and variables.
Political science journal articles show variations in literature review organization: some have a dedicated “literature review” section, others use descriptive subheadings, and some integrate citations throughout. Regardless of structure, the review requires clear leadership and communication of the study’s outline. It is not a summary of all read material or a dumping ground for citations.
A common misconception is that the literature drives the review, leading students to begin paragraphs with author names. Instead, the focus should be on the empirical finding or theoretical insight itself, with citations supporting the author’s theory building. The difference may seem subtle but helps maintain focus on the central goal: theory building. The literature review acts as a vital roadmap, a tightly focused discussion and justification of the research design. It can be conceived as a funnel, starting broad with the problem/puzzle and its implications, then narrowing to the specific research question, directly relevant theories, concept definitions, variables, and hypotheses. This structure ensures a logical progression from broad context to specific study design.
Two Examples of Theory Building
Racial Prejudice and Voting for Obama:
Existing Research
: Debates surrounding the 2008 presidential election prompted scholars to investigate racial prejudice and vote choice. Previous studies found that some whites do not support minority candidates, though these findings alone do not constitute a theory.
Causal Mechanisms
: Schaffner (2011) summarized causal mechanisms: overt racism or stereotypes (e.g., black candidates perceived as less competent/more liberal).
Theoretical Contribution (Schaffner)
: Schaffner’s work built on existing theory by incorporating “priming,” the ability of campaigns/events to make certain considerations (like race) more or less important to voters. He posited that not all prejudiced whites would necessarily vote against Obama; only those primed to think about race would be less likely to support him. This nuanced the existing theory that treated all prejudiced voters the same.
Hypothesis and Findings
: Schaffner hypothesized that whites would be least likely to support minority candidates when both racial prejudice is high AND they are placing more weight on the candidate’s race. His study found support for this hypothesis, illustrating how combining previously unrelated concepts (priming) adds nuance to existing theory.
Are Women's Organizations More Democratic?
Research Question
: Barakso (2007) asked “Is there a ‘woman’s way’ of governing?” specifically concerning how women’s interest groups govern themselves, an area with virtually no existing research.
Theory Building from Other Disciplines
: Despite the lack of direct literature, Barakso built her theory by drawing from extensive research in psychology, business administration, sociology, and political science. This literature consistently showed that women are more likely to encourage cooperative behavior, seek consensus, and delegate authority (e.g., female corporate managers).
Hypothesis and Findings
: This interdisciplinary research led Barakso to hypothesize that women’s organizations should be more democratic. However, her study did not find support for this expectation; women’s organizations were no more likely to be democratically structured than other groups. This unexpected finding creates a new puzzle for future scholars.
Political science research is interconnected with other social science disciplines (e.g., economics, sociology, psychology). Scholars frequently cite works from other fields, demonstrating how the best research draws insights across disciplines.
Taking Alternative Theories Seriously: What Do You Do When Your Theories and Hypotheses Don’t Match Your Findings?
Unexpected results (especially those in the “wrong” direction or null findings) can be unsettling. Researchers must first re-examine assumptions, the model, and data for errors. If no errors are found, the study should not be abandoned. Unexpected results suggest several possibilities:
Insufficient Theoretical Breadth
: The researcher might have failed to draw widely enough on extant theory, omitting relevant information or key variables. Actively considering and testing alternative theories reduces this risk. This aligns with the imperative to reduce vast information parsimoniously while mitigating the risk of omitting relevant data that could taint findings.
Policy Implications Example
: If a study finds no relationship between the number of women in a state legislature and policy outcomes (contrary to expectations, as women’s presence in legislatures often influences policy outcomes), the researcher should discuss this possibility. Even if the findings don't support the theory that women's equal representation is problematic due to different policy preferences, it doesn't necessarily undermine the overall notion that women's under-representation affects policymaking.
Interpreting Contradictory Results
: The author should discuss why her findings contradict existing literature and expectations, how her methodology might have skewed results, and various reasons why women might not behave differently in legislative settings despite theory (e.g., similar policy preferences, unconscious pressure to conform, or constraints of electoral pressure). Unforeseen results can be highly illuminating and contribute valuable insights when carefully analyzed.
Summing Up: Theory and Inference
Serious attention to theory building is essential for making strong causal inferences. Existing theories guide expectations and hypotheses, maximizing the potential to contribute to knowledge. Theory is crucial for informing study design choices, helping identify relevant variables (both those of primary interest and control variables necessary for strong inferences), and elucidating causal mechanisms. Importantly, theory demonstrates how research can contribute to general knowledge about the political world, thereby serving as a fundamental building block for political science research.
Key Terms
concepts
: Words that represent some idea and must be clearly defined in any research project.
falsifiability
: The ability for a theory to be proven wrong through empirical testing.
generalizations
: Broad statements or principles about how one thinks the world works; theories are fundamentally generalizations.
hypotheses
: Specific, testable statements based on a theory, expressing what one expects to find if the theory is correct.
literature review
: A systematic discussion within a study that explains the logic of the research, grounded in prior theories and research, and justifies all aspects of the research design.
observable implications
: Specific observations or outcomes that would be expected if a theory is accurate, allowing for empirical testing of the theory and its hypotheses.
theory
: A general explanation or a set of principles that attempts to explain why phenomena
Page 34
Principle: State theories precisely & concretely
Vague hypotheses merely obfuscate; precise ones can be disproved and thereby improved.
Principle of parsimony
Word used incoherently in casual/scholarly talk; clarified by Jeffreys (1961):
“Simple theories have higher prior probabilities.” → AKA Jeffreys-Wrinch Simplicity Postulate (akin to Occam’s razor).
Parsimony = assumption about the world’s simplicity, not a universal research design rule.
Appropriate only when we already know the domain is simple (e.g., some physics fields).
In social sciences, useful only occasionally; excessive insistence unwarranted.
Avoiding over-complexity follows directly from another maxim: make theory only as complex as evidence warrants.
‘Indeterminate research designs’ emerge when evidence is insufficient relative to theory complexity (see §4.1) — a research-design, not ontological, problem.
Pre-versus post-data collection
Above advice presumes no data collected yet.
If data already gathered, researchers can still revise theory & collect more data → expensive, time-consuming, wastes earlier data.
Problem: theory needs improvement but cannot collect new data.
Situation common → requires “great caution and self-restraint.”
Human minds excel at pattern recognition—even in randomness—so post-hoc theories risk being ‘wildly wrong.’
Ad hoc adjustments must be rare & disciplined.
Page 35
Two rules for post-hoc theory modification when new data are unaffordable
Less-restrictive modification allowed
Extend hypothesis to broader range so that theory faces more falsification opportunities.
Example: from “democracies with advanced welfare don’t fight” → “all modern democracies don’t fight.”
More-restrictive modification generally inappropriate
Adding qualifications after inspecting data smells like saving a bad theory.
E.g., restricting “modern democracies don’t fight” to only those with advanced welfare after exceptions found.
Likewise, bolting on multiple vague caveats (economic structure, climate, repressive leadership, etc.) = disguised admission: “my theory is correct, except in country x.”
Without new data these patches have no admissible evidence.
When stuck (no new data possible): accept being wrong; solid negative findings > flimsy positive ones.
After admitting defeat
Add sections on future research & speculative modifications; freedom to propose new conditions, theories, data designs—but label as speculative until tested.
Creativity vs. rules
Social science not rule-bound; data can inspire theoretical insights even if restrictive; acceptable if authors credibly show that such modification could have been made pre-data.
Until new data, status of revised theory = very uncertain.
Pilot projects
Valuable especially when data collection costly (interviews, etc.).
Preliminary data may reshape questions/theory; subsequent data then test new theory, avoiding double-dipping.
Page 36
Improving Data Quality
Definition: “Data” = systematically collected qualitative/quantitative info.
Data often gathered before precise research question; still need quality rules.
Guideline 1: Record & report the data-generation process
Necessary for detecting bias and enabling valid inferences.
Quantitative: sampling frame, wording of questions.
Qualitative: explicit case-selection rules.
Fear of sharing data is unwarranted; publication & sharing secure credit and spur citations.
Guideline 2: Collect data on as many observable implications as possible
More diverse contexts → stronger evaluation of theory.
Two expansion paths:
a. More observations on same DV (finer time or geography).
b. Additional dependent variables predicted by theory.
Example: Rational deterrence theory
Direct test: attack decisions given threats (Huth 1988).
But must also include cases where threats deterred → else selection bias.
Consider lab experiments, oligopolistic firms, organized-crime analogies—cross-domain support builds plausibility.
Practical constraint → infinite data impossible; scholars risk ruin by over-collection; balance needed.
Guideline 3: Maximize measurement validity
Measure what you think you measure; stick close to observed indicators; beware of alternative meanings (e.g., “ignorance” reply in repressive regimes).
Guideline 4: Ensure reliability
Procedures must yield same measurement upon repetition (same researcher, different times, or different coders).
Achieved via explicit coding rules, double-coding, test-retest.
Guideline 5: Strive for replicability of both data & reasoning
Full detail so others can duplicate data & logic.
Quantitative: start with same dataset → replicate analysis (often harder than expected; see Dewald et al. 1986).
Qualitative: footnotes, bibliographic essays, accessible archives; share field notes/audio when possible.
Extensive example: Middletown sociological studies (1929, 1937, 1980s replications).
Scholars with privileged access should secure similar future access for others.
Page 37
Replicability rarely executed but always prepare as if; aids reader evaluation.
Improving Use of Existing Data
Better to collect new data, but often impossible → must make best of flawed data.
First principle: aim for unbiased inferences (correct on average).
Beware selection bias—cases chosen distort population (deliberate or subtle).
Beware omitted-variable bias—missing controls confound causality.
Second principle: seek efficiency—use all information to maximize inference precision.
Use disaggregated data when available; finer units add information albeit with higher uncertainty.
Page 38
Themes of the Volume (intro to §1.3 begins) — only headings summarized on current pages (full discussion continues pp. 38–43):
Using Observable Implications to connect theory & data.
Maximizing Leverage: explain as much as possible with as little.
Page 39
Theme 1 elaborated: tight linkage between theory & empirical inquiry.
Good theories yield observable implications; good data collection guided by those implications.
Questions to ask: What are the theory’s observable implications? Are observations relevant to them?
Page 40
Theme 2 elaborated: search for additional implications increases leverage.
Leverage ≠ parsimony; it is about ratio of explained variance to information used.
Areas with traditionally low leverage often attract qualitative methods; goal should be to raise leverage.
Techniques:
Improve theory for more implications.
Improve data to observe more implications.
Improve data-use to extract more information.
Page 41
Cross-level observations can raise leverage; data need not match theory’s aggregation.
Ecological fallacy warning (Robinson 1950) acknowledged but shouldn’t prohibit using multi-level data if informative.
Example: revolution theory—collect individual interviews, small-community reactions, leader statements.
One aggregate observation (revolution occurs) = only one implication; deeper digging distinguishes theories.
Page 42
Theme 3: Reporting Uncertainty
All inference is uncertain; qualitative as error-prone as quantitative.
Good social scientists estimate and report uncertainty; qualitative research often fails here. This includes acknowledging limitations in data collection, measurement, and theoretical scope.
The degree of uncertainty should be quantified where possible (e.g., -values, confidence intervals, standard errors) or clearly articulated through qualitative caveats and discussions of alternative interpretations.
Neustadt & May (1986) heuristic: “How much of your own money would you wager—and at what odds?” This heuristic encourages researchers to honestly assess the subjective probability of their findings being correct and the potential for error, thereby fostering greater transparency in reporting.
Page 43
Theme 4: Thinking like a Social Scientist — skepticism & rival hypotheses.
On any causal claim, ask about data accuracy, alternative explanations, and reversed causality. This critical approach ensures that conclusions are robust and not merely coincidental correlations.
The process of causal inference is a cycle of successive approximations; each conclusion sparks further inquiry, leading to refinement and deeper understanding.
Actively seeking out plausible rival hypotheses is crucial. For instance, if a theory suggests A causes B, a social scientist would consider if C causes B, if B causes A, or if A and B are both caused by an unobserved D.
Example: lower heart-attack rates in Japan → initial plausible “less red meat” explanation. A skeptical social scientist would then rigorously check this against confounders (genetics, lifestyle, reverse causation like healthier people choosing less red meat), seeking to rule out these alternatives before concluding causation.
How to Develop a Sound Research Question
Developing a sound research question in political science is a significant and often time-consuming process, crucial for effective scholarly analysis. It is distinct from typical undergraduate writing assignments, which may focus on argumentation rather than empirical inquiry. The time and effort required for an initial research idea to evolve into a bona fide research question can be surprising, especially for those new to social science academic standards. While unusual political or public policymaking events can spark intellectual curiosity, the scholarly research question is structured and arrived at through a different, more rigorous process.
Students often underestimate the time needed to craft a research question, prioritizing gathering research, data analysis, and writing. However, experienced researchers emphasize that considerable time and care must be invested in investigating the scholarly literature and other relevant material to ensure the question is both important and feasible. A poorly formulated question, regardless of the effort to answer it, is unlikely to yield useful results for other scholars and can derail a project. The justification for a study, the literature review, and the research design are all intricately linked to the research question. While the research process is non-linear and revisions are common, a strong initial question facilitates smoother execution and produces more valuable results.
What Makes for a Good Research Question?
Most research questions are ignited by personal passion or interest, such as an personal interest in a political system or an observation of under-representation. While personal enthusiasm is an excellent starting point, moving from a general topic to a sound research question requires understanding its key traits:
Non-normative and Answerable: Research questions should inquire “what is,” not “what ought to be.” Questions beginning with “should” (e.g., “Should the United States invade Iraq?” or “Should there be more women in Congress?”) tend to lead to position papers rather than scholarly analyses. They often rely on subjective beliefs, values, and political contexts, assuming ideal, universally applicable solutions where politics don't matter. Purely normative questions ask for judgment calls based on opinion rather than testable evidence. Empirically oriented political scientists, unlike philosophers or political theorists, aim to contribute to understanding how the world works by testing alternative explanations. Instead of asking, “Should there be more women in Congress?”, an empirical political scientist might ask, “To what extent does the presence of women legislators influence agenda setting and policy outcomes in the U.S. Congress?”
Generates Implications for Real-World Problems: Although not normatively constructed, a research question must connect to significant, broader issues like representation and justice. It must address the “so what?” question, demonstrating its relevance beyond a narrow instance (e.g., why a specific political candidate lost). For example, while fairness is a normative concern, researchers can empirically ask if legislatures with more women produce different policies, thereby allowing for discussion of normative implications (e.g., why male legislator over-representation matters).
Addresses a Debate or Puzzle in the Literature: A sound question emerges from and contributes to existing scholarly literature. It should explain its potential to shed new light on an unresolved puzzle, problem, or debate within academic research.
Not Overly Broad: Research questions must be manageable. Initial grand questions (e.g., “Why do countries go to war?”) must be refined into more discrete, feasible inquiries (e.g., “Are authoritarian governments more likely to start wars than democracies?” or “How effective are treaties in preventing wars?”).
Not Too Narrow: A question should not be so specific that its answer is only relevant to a very small audience. For instance, instead of focusing on a single candidate's campaign strategies (“Which campaign strategies were most effective in helping Candidate Smith win more votes?”), broaden it to be more generally applicable (“Which campaign strategies are most effective in helping state legislative candidates win more votes?”), even if the study still focuses on a few campaigns.
Beginning the Research Process: What Do You Want to Know?
To develop a research question, consider these practical steps:
Adopt a Suitable Frame of Mind: Your goal is to formulate a question whose answer you genuinely want to know, rather than one you aim to “prove.” Ideally, you should not know the answer with certainty, even if you have a strong hunch.
Record Keeping: Establish a dedicated research notebook or word-processing file from the project's outset. Use it to collect ideas, leads, sources (with full citations, page numbers, and dates of access), iterations of your question, and meeting notes. This practice helps stimulate thinking, suggests fruitful avenues, and maintains momentum, especially during breaks in work. Following advice from Ernest Hemingway and Cory Doctorow, always stop work with a “rough edge”—knowing your next concrete steps or even mid-sentence—to avoid getting “stuck.” Diligent record-keeping also ensures the integrity of your final work product by preventing accidental plagiarism from cut-and-paste content or mixing paraphrased with verbatim text without proper attribution.
Consider Feasibility: Assess your available time, familiarity with the topic, faculty mentorship opportunities, and the accessibility of data and methods. While a reasonable timeline is crucial, do not let potential data/methodological challenges immediately discourage an interesting question, as many questions can be reframed to utilize various data sources and tools.
Generate Preliminary Questions: Reflect on personal interests, current events, or previous research papers that could be expanded. Consult with professors or graduate students in your department who work in relevant areas. Reading review essays (e.g., in Annual Review of Political Science) or handbooks (e.g., Oxford/Cambridge Handbook series) offers a shortcut to understand current scholarly debates and identify potential new research directions within specific subfields (e.g., comparative politics, American politics, international relations). Skimming introductions and conclusions of journal articles and academic books can also reveal motivating questions and suggestions for future studies.
What Do Scholars Already Know? The Core of a Research Question: What Is the Controversy, Debate, or Puzzle?
Once research interests are narrowed, the next step is carving out a sound, researchable question. This is an iterative process, especially for new researchers, as the question will continuously be refined while engaging with existing scholarship. A good question aims to address a controversy, debate, or puzzle that has been overlooked, under-studied, or is currently debated within academic literature. It may also stem from an external puzzle that scholarly examination can illuminate. Consulting a faculty member with research interests in your chosen area is highly recommended.
Preliminary questions (e.g., “Why aren’t there more women in political office in the United States?” or “How has social media transformed contemporary social movements?”) are good starting points but are often too broad. They need further refinement based on existing research. For instance, the question about women in political office points to a central puzzle in scholarship but needs to be honed in light of specific dimensions already studied.
Question Development: Intermediate Stages
To grasp how political scientists articulate research questions, study journal article abstracts and introductions. Observe how authors transition from a broad observation to a concise, answerable, and focused question. Common formulations include: “To what extent does…”, “Under what conditions do…”, or “Given X, what accounts for Y?” Practice formulating your questions using these phrases.
Move from skimming to fully reading relevant journal articles. Scan their bibliographies for leads, noting repeatedly cited authors, as this indicates ongoing scholarly conversations. Prioritize recent articles to quickly grasp current debates and findings. Identify inconsistencies, puzzles, and opposing arguments among scholars, as these often signal topics needing further study. For example, Mona Lena Krook’s book Quotas for Women in Politics concludes with specific directions for future research, which can be invaluable. When studying women’s political representation, scholars often narrow their focus to specific dimensions, such as the political campaign process, institutional or state-level characteristics, or the impact of women’s increased educational attainment. At this stage, intermediate questions might be formulated as: “To what extent does the role of money in politics affect women’s ability to attain political office as compared with men?” or “Under what conditions are women in Western industrialized nations likely to constitute a higher proportion of nationally elected officials?” However, these still need refinement; for example, the latter might require focusing on nations with comparable political systems due to research design limitations. The iterative nature of the research process means questions are frequently revised to match what is reasonably answerable given available resources.
Why Do We Care? Or the “So What” Question
Social scientists must justify their research questions to demonstrate their broader importance to scholars and connection to real-world political phenomena (e.g., democratic principles, political behavior, policy outcomes). This is often referred to as the “so what?” question, ensuring the findings have practical implications. Good research papers typically begin by stating the question and immediately justifying its significance (“This question is important because…”). The literature review further defends the question by showing its relation to ongoing academic conversations.
Question Development: Advanced Stages
Further refinement to an advanced stage occurs when:
You have gathered enough information to craft a clear question with a high level of specificity.
Your question has a variety of possible answers.
It addresses a debate or puzzle in the literature.
It is manageable given your skills, knowledge base, and time.
It is guided by prior scholarship and has the potential to contribute to it.
At this stage, you will focus on a narrower community of scholars asking similar questions, identified by frequently cited authors. This helps understand specific debates, methods, central theories, and question articulation within that scholarly circle. This groundwork provides a head start on developing theory, hypotheses, selecting cases, and choosing data sources and methods. For example, a refined question might be: “How does the sex of political candidates affect voting perceptions and behavior in Turkey?” (Matland & Tezcur, 2011), justified by Turkey’s unique democratic context as a Muslim-majority country.
Keeping the Big Picture in Mind: How Will You Execute the Study?
It is essential to consider how you might answer your research question and what data you might use early in the process. Habitually moving between question framing and execution considerations offers several benefits. It helps define a doable project scope that still contributes value to scholarship. As you read, note the data sources and methods employed by authors (e.g., small- techniques like interviews, surveys, experiments, or large- datasets with statistical techniques). Many authors share data. Don't feel limited by current skill sets; basic statistical or qualitative analysis techniques can be learned during the project.
Alternative Theories
While developing your question, take note of competing theories presented by various scholars. Your question will address an unresolved area in the literature. As you read, you will naturally form hunches about the answer, which are useful. However, avoid becoming overly attached to one explanation or set of hypotheses. Social scientists are obligated to explicitly and fairly test rival theories to advance knowledge, rather than trying to “prove” a preferred one.
Definitions
Use common terminology from the literature for phenomena or concepts. Make note of these definitions and their sources, using them in your work rather than creating new ones or prematurely revising existing ones. Reinventing definitions, especially for minor tweaks, wastes time, signals a lack of engagement with ongoing scholarly conversations, and makes it harder for others to build on common understandings. However, note inconsistencies in central concept definitions; these might indicate scholarly disagreements that could inspire a viable research question.
Summing Up: The Research Question
The research question is central to political science research and requires time and effort to refine. It should relate to and contribute to scholarly conversations. The iterative process of question development means it will likely be adjusted as you learn more about your subject and available data. This careful foundation will prepare you for developing theories and hypotheses and structuring your entire study.
Key Terms
Literature Review: A comprehensive summary and critical analysis of existing scholarly works related to a research topic, usually published in academic journals and books.
Normative Implications: The prescriptive or value-laden aspects of a research question or finding, concerning what should be, as opposed to what is.
Peer-Reviewed: Research that has undergone a vetting process by multiple anonymous experts in the field who critique the work and recommend publication or rejection.
Review Essays: Scholarly articles that summarize and critically evaluate the state of research in a specific area, often identifying directions for future study.
Scholarly Literature: Academic works, typically peer-reviewed journal articles and books by doctoral candidates or those holding doctorates, published by university or reputable commercial presses.
“So What” Question: A core inquiry in academic research that demands justification for a study’s broader relevance and importance, explaining why potential findings matter to scholars and real-world phenomena.
Note on Scholarly Sources: Throughout this text, “scholarly” or “academic” literature refers to peer-reviewed social science journal articles and books written by doctoral candidates or those holding doctorates, published by university presses or reputable commercial presses
Linking Theory and Inference
Theory is the most crucial part of the research enterprise due to its centrality in making inferences. It informs every part of the research process, from formulating a research question to designing a study and interpreting results. As political scientists, our goal is to test, modify, or construct new theories to better explain phenomena. This is not done in isolation; we build upon prior scholarship, proposing new ideas or arguing against existing ones when they are incomplete or incorrect. While developing a grand theory is rare, making modest but significant theoretical contributions—such as showing how existing theories are incomplete, demonstrating the necessity of previously omitted variables, or resolving inconsistent findings—constitutes valid “middle range” theoretical contributions.
What Is Theory? Why Are Theories So Important and So Valuable?
A theory is fundamentally a generalization, defined as a set of principles that explains why people behave as they do across various contexts, or broadly, a statement about how one believes the world works. The purpose of scientific studies is not to explain single events but to develop theories that can be applied to other related phenomena. Theories provide a foundation of general knowledge, reducing complex observations into regular patterns and relationships, which can then be applied to past, present, or future problems. Well-developed theories serve as critical shortcuts or heuristics for scholars, policymakers, and the public in decision-making.
Individuals constantly rely on, construct, and apply theories in everyday life, often subconsciously. These can be inductively generated from personal observation (e.g., studying with distractions leads to lower grades) or deductively informed by existing knowledge (e.g., from trusted professionals or media). Without theories, every situation would require individual, repeated investigation. A toolkit of theories allows the application of knowledge from one context to another, saving immense time and resources (e.g., theories on Congressional operations preclude needing a new study for every new bill).
Beyond Generalizability, theory plays a special role in scholarly research by promoting sound research design. It is critical for:
Guiding researchers in determining which alternative theories to consider.
Setting the stage for developing interesting hypotheses to test.
Herming discern which factors (independent variables) to include and control for, and which way the causal arrow goes. Theory is the primary guide for identifying independent and dependent variables.
Ensuring the robustness of observed relationships, distinguishing them from spurious correlations.
Interpreting findings; data alone cannot explain patterns. Theory informs inferences to help interpret and explain results.
What Characterizes a Good Theory?
A good social science theory is a reasoned and precise speculation about the answer to a research question, stating why the proposed answer is correct. It usually implies several specific descriptive or causal hypotheses and must be consistent with prior evidence. Good theories are also defined as an interrelated set of constructs (variables) formed into propositions or hypotheses that specify relationships among variables, often in terms of magnitude or direction.
Good Theory Builds on Existing Theory
Conceptual and theoretical understandings in political science undergo a continuous process of refinement, as new theories emerge to question, refine, or replace older ones. Knowledge advances most efficiently when studies build upon prior scholarship. Theories that are “well grounded” in prior literature are valuable because they speak to the common interests and mutual understandings of others who are interested in the subject that the theory addresses. Well-grounded theories are therefore accessible to others. By drawing on prior theories and addressing ongoing conversations, a researcher is more likely to influence others’ thinking on the topic they are studying.
For example, in their article “Protest and democracy in Latin America’s market era,” Paul T. Bellinger, Jr. and Moises Arce aimed to understand “whether and how political democracy has influenced societal responses to economic liberalization.” They explicitly noted the contradictory implications of two existing theoretical streams: one emphasizing the “depoliticizing” effects of economic reforms in democracies (suppressing protest), and another, the “repoliticization” literature (promoting protest). To mediate these, they introduced a third literature on contentious politics, which posits that grievances increase mobilization willingness while democracy creates a favorable environment. This allowed Bellinger and Arce to formulate their own testable theory: democratic politics, even if imperfect, should encourage collective political activity, not render it obsolete. This example demonstrates how well-grounded theories are “leveraged” or applied across topics and situations by different scholars and policymakers.
Good Theory Concretely Specifies the Concepts and/or Variables It Invokes
Concepts are ideas represented by words and must be clearly defined in any research project. For instance, a scholar studying corruption must specify whether it refers to “petty administrative corruption” or “grand corruption by high-level officials,” as their causes differ. Precise definitions are crucial. It is generally advised to avoid substantially revising existing widely used definitions to ensure work is “leverageable” or applicable to other scholars, contributing to a common vocabulary and advancing knowledge, rather than just producing information. However, when scholars are divided over definitions, these differences can be leveraged to refine one's approach (e.g., the concept of representation as debated by Pitkin, Mansbridge, and Rehfeld).
Many common political science concepts (e.g., power, democracy, representation, equality, political efficacy) have multiple definitions across disciplines. Researchers should investigate existing definitions, consider their implications, and clearly state their chosen definition. Even seemingly self-evident terms like “voter turnout” require careful specification; defining turnout as a percentage of voting-age public vs. eligible voters significantly alters conclusions about its decline, demonstrating the important consequences of conceptual definition for findings.
Good Theory Clarifies the Relationship between Concepts and What Is to Be Explained or Described
Sound political science requires precise postulation of relationships. Whether descriptive or causal, theories must clearly describe the conclusion about the relationship between phenomena and the factors that explain, shape, influence, or cause them. Theories must state how the world works and why it works that way. For causal research questions, the theory should explain why independent variables are expected to cause changes in the dependent variable, elucidating the causal mechanisms (e.g., why education relates to voting involves explaining that more educated individuals tend to know more about politics, making vote decisions easier). These explanations are rooted in prior scholarship.
For descriptive inferences, the theory should describe how an examination of prior theories, combined with new observations, influences a debate or problem in the literature. For instance, Barakso’s study intervened in the debate on declining civic engagement by theorizing that organizational operation and internal democracy within groups, rather than just changes in group numbers or membership, influence civic participation. This study posited a clear theoretical relationship and laid out implications for future researchers.
Good Theory Is Falsifiable
Falsifiability, the ability for a theory to be proven wrong, is critical. Theories cannot be “proven” correct, but their soundness can be estimated through testing. A non-falsifiable theory (e.g., “Wars in Iraq and Afghanistan prevented another terrorist attack” because the counterfactual cannot be observed) prevents meaningful testing and evaluation. A falsifiable theory, conversely, allows for multiple tests of its validity (e.g., “military invasions of terrorist states reduce future terrorist attacks,” which can generate hypotheses like “worldwide terrorist incidents will diminish in the wake of an external military intervention”).
Some robust scholarly conversations revolve around non-falsifiable theories (e.g., deliberative democracy theory), often due to their engagement with salient political concepts. Critics argue that deliberative democracy, often conceived normatively, lacks empirical testing on its effectiveness or conditions for suboptimal outcomes. Vague and variable definitions also “insulate” it from refutation. Mutz suggests that instead of grand, overarching theories, scholars should develop and test “middle-range theories,” which are intermediate, precise, and falsifiable components of broader theories. This approach, by replacing vague entities with concrete concepts and requiring empirically grounded hypotheses, helps understand which elements are crucial to specific outcomes.
Good Theory Leads to Testable Hypotheses
Good theory specifies expected observations if the theory accurately describes how the world works. Whether descriptive or causal, theories must lead to specific, testable hypotheses—or implications. Testable hypotheses allow researchers to establish a theory's soundness. Furthermore, a theory that can generate multiple testable hypotheses, especially those extending beyond the immediate study, benefits the broader academic community by providing more avenues for exploration and knowledge improvement. For example, comparing a theory that “incumbents win re-election more often because they tend to have more money” (limited observable implications) with one stating “incumbents win re-election more often because they tend to have more resources” (many observable implications, including financial donations, name recognition, campaign workers, casework goodwill, and favorable media coverage) shows that a theory with more observable implications tends to be broader and more useful.
Incorporating Theory into Your Study: the Literature Review
The development of a research question is intricately tied to prior literature. The literature review is the practical means of incorporating others’ theories into a study. By this stage, a researcher has already gained tools to grasp and outline their literature review, as the question likely grew from competing theories or perceived shortcomings in existing literature, informing the researcher’s own views.
The researcher must decide whether to test an existing theory or propose a new explanation/theory, a decision often made while reading and refining the research question within the literature.
Thinking about the Literature Review
The literature review explains the logic of a study, grounded in prior research (theory). It reveals the main theories that justify the research question’s salience and particular formulation. It also specifies the key theories that led to the selection of a particular theory to explore or test, or if proposing a new theory, the shortcomings of prior theories that prompted the new contribution. It contains the theoretical justification for hypotheses (what variables are important and why) and concept definitions. Ultimately, theory informs the interpretation of findings.
The term “literature review” can be misleading, as its purpose is not to list every source read. Thinking of it as a “theory section” helps maintain focus on its purpose.
Three Goals of the Literature Review
Expanded Discussion of the Research Question
: Systematically and selectively discuss key problems, theories, and data that justify the salience, importance, and specific formulation of the research question.
Delineate Key Discussions and Debates
: Delineate the key discussions, debates, and data in the literature directly related to the question and the theory being examined or proposed. It should logically progress from general “big picture” concepts to specific debates leading to the proposed theory. Explicitly state (or restate) plausible alternative or rival theories that will be examined.
Present Working Answer/Theory and Hypotheses
: Present the researcher’s own working answer or theory, typically tested through hypotheses. Hypotheses are explicit statements of expected findings if the theory is correct, often worded as “if–then” or “when–then” statements (e.g., “Women’s estimation of the costs of running a political campaign are significantly less accurate than men’s.”). This component links to the data and methods section, explaining concept definitions and selected factors (variables) for analysis (e.g., voter pool, definition of “major scandal,” election laws, and other relevant factors from literature that might be omitted with justification). It also briefly reviews control variables (common factors like demographics or partisan affiliation).
Writing the Literature Review
The literature review is a complex but essential task, requiring thoughtful integration of all central project elements. It serves as a puzzle where each piece must interlock. There is no single formula for writing one, but common elements include discussing (and supporting with literature) the broader problem, its implications, the research question and its importance, theories, concepts, empirical evidence, hypotheses, one’s own theory, rival theories, and variables.
Political science journal articles show variations in literature review organization: some have a dedicated “literature review” section, others use descriptive subheadings, and some integrate citations throughout. Regardless of structure, the review requires clear leadership and communication of the study’s outline. It is not a summary of all read material or a dumping ground for citations.
A common misconception is that the literature drives the review, leading students to begin paragraphs with author names. Instead, the focus should be on the empirical finding or theoretical insight itself, with citations supporting the author’s theory building. The difference may seem subtle but helps maintain focus on the central goal: theory building. The literature review acts as a vital roadmap, a tightly focused discussion and justification of the research design. It can be conceived as a funnel, starting broad with the problem/puzzle and its implications, then narrowing to the specific research question, directly relevant theories, concept definitions, variables, and hypotheses. This structure ensures a logical progression from broad context to specific study design.
Two Examples of Theory Building
Racial Prejudice and Voting for Obama:
Existing Research
: Debates surrounding the 2008 presidential election prompted scholars to investigate racial prejudice and vote choice. Previous studies found that some whites do not support minority candidates, though these findings alone do not constitute a theory.
Causal Mechanisms
: Schaffner (2011) summarized causal mechanisms: overt racism or stereotypes (e.g., black candidates perceived as less competent/more liberal).
Theoretical Contribution (Schaffner)
: Schaffner’s work built on existing theory by incorporating “priming,” the ability of campaigns/events to make certain considerations (like race) more or less important to voters. He posited that not all prejudiced whites would necessarily vote against Obama; only those primed to think about race would be less likely to support him. This nuanced the existing theory that treated all prejudiced voters the same.
Hypothesis and Findings
: Schaffner hypothesized that whites would be least likely to support minority candidates when both racial prejudice is high AND they are placing more weight on the candidate’s race. His study found support for this hypothesis, illustrating how combining previously unrelated concepts (priming) adds nuance to existing theory.
Are Women's Organizations More Democratic?
Research Question
: Barakso (2007) asked “Is there a ‘woman’s way’ of governing?” specifically concerning how women’s interest groups govern themselves, an area with virtually no existing research.
Theory Building from Other Disciplines
: Despite the lack of direct literature, Barakso built her theory by drawing from extensive research in psychology, business administration, sociology, and political science. This literature consistently showed that women are more likely to encourage cooperative behavior, seek consensus, and delegate authority (e.g., female corporate managers).
Hypothesis and Findings
: This interdisciplinary research led Barakso to hypothesize that women’s organizations should be more democratic. However, her study did not find support for this expectation; women’s organizations were no more likely to be democratically structured than other groups. This unexpected finding creates a new puzzle for future scholars.
Political science research is interconnected with other social science disciplines (e.g., economics, sociology, psychology). Scholars frequently cite works from other fields, demonstrating how the best research draws insights across disciplines.
Taking Alternative Theories Seriously: What Do You Do When Your Theories and Hypotheses Don’t Match Your Findings?
Unexpected results (especially those in the “wrong” direction or null findings) can be unsettling. Researchers must first re-examine assumptions, the model, and data for errors. If no errors are found, the study should not be abandoned. Unexpected results suggest several possibilities:
Insufficient Theoretical Breadth
: The researcher might have failed to draw widely enough on extant theory, omitting relevant information or key variables. Actively considering and testing alternative theories reduces this risk. This aligns with the imperative to reduce vast information parsimoniously while mitigating the risk of omitting relevant data that could taint findings.
Policy Implications Example
: If a study finds no relationship between the number of women in a state legislature and policy outcomes (contrary to expectations, as women’s presence in legislatures often influences policy outcomes), the researcher should discuss this possibility. Even if the findings don't support the theory that women's equal representation is problematic due to different policy preferences, it doesn't necessarily undermine the overall notion that women's under-representation affects policymaking.
Interpreting Contradictory Results
: The author should discuss why her findings contradict existing literature and expectations, how her methodology might have skewed results, and various reasons why women might not behave differently in legislative settings despite theory (e.g., similar policy preferences, unconscious pressure to conform, or constraints of electoral pressure). Unforeseen results can be highly illuminating and contribute valuable insights when carefully analyzed.
Sum
Linking Theory and Inference
Theory is the most crucial part of the research enterprise due to its centrality in making inferences. It informs every part of the research process, from formulating a research question to designing a study and interpreting results. As political scientists, our goal is to test, modify, or construct new theories to better explain phenomena. This is not done in isolation; we build upon prior scholarship, proposing new ideas or arguing against existing ones when they are incomplete or incorrect. While developing a grand theory is rare, making modest but significant theoretical contributions—such as showing how existing theories are incomplete, demonstrating the necessity of previously omitted variables, or resolving inconsistent findings—constitutes valid “middle range” theoretical contributions.
What Is Theory? Why Are Theories So Important and So Valuable?
A theory is fundamentally a generalization, defined as a set of principles that explains why people behave as they do across various contexts, or broadly, a statement about how one believes the world works. The purpose of scientific studies is not to explain single events but to develop theories that can be applied to other related phenomena. Theories provide a foundation of general knowledge, reducing complex observations into regular patterns and relationships, which can then be applied to past, present, or future problems. Well-developed theories serve as critical shortcuts or heuristics for scholars, policymakers, and the public in decision-making.
Individuals constantly rely on, construct, and apply theories in everyday life, often subconsciously. These can be inductively generated from personal observation (e.g., studying with distractions leads to lower grades) or deductively informed by existing knowledge (e.g., from trusted professionals or media). Without theories, every situation would require individual, repeated investigation. A toolkit of theories allows the application of knowledge from one context to another, saving immense time and resources (e.g., theories on Congressional operations preclude needing a new study for every new bill).
Beyond Generalizability, theory plays a special role in scholarly research by promoting sound research design. It is critical for:
Guiding researchers in determining which alternative theories to consider.
Setting the stage for developing interesting hypotheses to test.
Herming discern which factors (independent variables) to include and control for, and which way the causal arrow goes. Theory is the primary guide for identifying independent and dependent variables.
Ensuring the robustness of observed relationships, distinguishing them from spurious correlations.
Interpreting findings; data alone cannot explain patterns. Theory informs inferences to help interpret and explain results.
What Characterizes a Good Theory?
A good social science theory is a reasoned and precise speculation about the answer to a research question, stating why the proposed answer is correct. It usually implies several specific descriptive or causal hypotheses and must be consistent with prior evidence. Good theories are also defined as an interrelated set of constructs (variables) formed into propositions or hypotheses that specify relationships among variables, often in terms of magnitude or direction.
Good Theory Builds on Existing Theory
Conceptual and theoretical understandings in political science undergo a continuous process of refinement, as new theories emerge to question, refine, or replace older ones. Knowledge advances most efficiently when studies build upon prior scholarship. Theories that are “well grounded” in prior literature are valuable because they speak to the common interests and mutual understandings of others who are interested in the subject that the theory addresses. Well-grounded theories are therefore accessible to others. By drawing on prior theories and addressing ongoing conversations, a researcher is more likely to influence others’ thinking on the topic they are studying.
For example, in their article “Protest and democracy in Latin America’s market era,” Paul T. Bellinger, Jr. and Moises Arce aimed to understand “whether and how political democracy has influenced societal responses to economic liberalization.” They explicitly noted the contradictory implications of two existing theoretical streams: one emphasizing the “depoliticizing” effects of economic reforms in democracies (suppressing protest), and another, the “repoliticization” literature (promoting protest). To mediate these, they introduced a third literature on contentious politics, which posits that grievances increase mobilization willingness while democracy creates a favorable environment. This allowed Bellinger and Arce to formulate their own testable theory: democratic politics, even if imperfect, should encourage collective political activity, not render it obsolete. This example demonstrates how well-grounded theories are “leveraged” or applied across topics and situations by different scholars and policymakers.
Good Theory Concretely Specifies the Concepts and/or Variables It Invokes
Concepts are ideas represented by words and must be clearly defined in any research project. For instance, a scholar studying corruption must specify whether it refers to “petty administrative corruption” or “grand corruption by high-level officials,” as their causes differ. Precise definitions are crucial. It is generally advised to avoid substantially revising existing widely used definitions to ensure work is “leverageable” or applicable to other scholars, contributing to a common vocabulary and advancing knowledge, rather than just producing information. However, when scholars are divided over definitions, these differences can be leveraged to refine one's approach (e.g., the concept of representation as debated by Pitkin, Mansbridge, and Rehfeld).
Many common political science concepts (e.g., power, democracy, representation, equality, political efficacy) have multiple definitions across disciplines. Researchers should investigate existing definitions, consider their implications, and clearly state their chosen definition. Even seemingly self-evident terms like “voter turnout” require careful specification; defining turnout as a percentage of voting-age public vs. eligible voters significantly alters conclusions about its decline, demonstrating the important consequences of conceptual definition for findings.
Good Theory Clarifies the Relationship between Concepts and What Is to Be Explained or Described
Sound political science requires precise postulation of relationships. Whether descriptive or causal, theories must clearly describe the conclusion about the relationship between phenomena and the factors that explain, shape, influence, or cause them. Theories must state how the world works and why it works that way. For causal research questions, the theory should explain why independent variables are expected to cause changes in the dependent variable, elucidating the causal mechanisms (e.g., why education relates to voting involves explaining that more educated individuals tend to know more about politics, making vote decisions easier). These explanations are rooted in prior scholarship.
For descriptive inferences, the theory should describe how an examination of prior theories, combined with new observations, influences a debate or problem in the literature. For instance, Barakso’s study intervened in the debate on declining civic engagement by theorizing that organizational operation and internal democracy within groups, rather than just changes in group numbers or membership, influence civic participation. This study posited a clear theoretical relationship and laid out implications for future researchers.
Good Theory Is Falsifiable
Falsifiability, the ability for a theory to be proven wrong, is critical. Theories cannot be “proven” correct, but their soundness can be estimated through testing. A non-falsifiable theory (e.g., “Wars in Iraq and Afghanistan prevented another terrorist attack” because the counterfactual cannot be observed) prevents meaningful testing and evaluation. A falsifiable theory, conversely, allows for multiple tests of its validity (e.g., “military invasions of terrorist states reduce future terrorist attacks,” which can generate hypotheses like “worldwide terrorist incidents will diminish in the wake of an external military intervention”).
Some robust scholarly conversations revolve around non-falsifiable theories (e.g., deliberative democracy theory), often due to their engagement with salient political concepts. Critics argue that deliberative democracy, often conceived normatively, lacks empirical testing on its effectiveness or conditions for suboptimal outcomes. Vague and variable definitions also “insulate” it from refutation. Mutz suggests that instead of grand, overarching theories, scholars should develop and test “middle-range theories,” which are intermediate, precise, and falsifiable components of broader theories. This approach, by replacing vague entities with concrete concepts and requiring empirically grounded hypotheses, helps understand which elements are crucial to specific outcomes.
Good Theory Leads to Testable Hypotheses
Good theory specifies expected observations if the theory accurately describes how the world works. Whether descriptive or causal, theories must lead to specific, testable hypotheses—or implications. Testable hypotheses allow researchers to establish a theory's soundness. Furthermore, a theory that can generate multiple testable hypotheses, especially those extending beyond the immediate study, benefits the broader academic community by providing more avenues for exploration and knowledge improvement. For example, comparing a theory that “incumbents win re-election more often because they tend to have more money” (limited observable implications) with one stating “incumbents win re-election more often because they tend to have more resources” (many observable implications, including financial donations, name recognition, campaign workers, casework goodwill, and favorable media coverage) shows that a theory with more observable implications tends to be broader and more useful.
Incorporating Theory into Your Study: the Literature Review
The development of a research question is intricately tied to prior literature. The literature review is the practical means of incorporating others’ theories into a study. By this stage, a researcher has already gained tools to grasp and outline their literature review, as the question likely grew from competing theories or perceived shortcomings in existing literature, informing the researcher’s own views.
The researcher must decide whether to test an existing theory or propose a new explanation/theory, a decision often made while reading and refining the research question within the literature.
Thinking about the Literature Review
The literature review explains the logic of a study, grounded in prior research (theory). It reveals the main theories that justify the research question’s salience and particular formulation. It also specifies the key theories that led to the selection of a particular theory to explore or test, or if proposing a new theory, the shortcomings of prior theories that prompted the new contribution. It contains the theoretical justification for hypotheses (what variables are important and why) and concept definitions. Ultimately, theory informs the interpretation of findings.
The term “literature review” can be misleading, as its purpose is not to list every source read. Thinking of it as a “theory section” helps maintain focus on its purpose.
Three Goals of the Literature Review
Expanded Discussion of the Research Question
: Systematically and selectively discuss key problems, theories, and data that justify the salience, importance, and specific formulation of the research question.
Delineate Key Discussions and Debates
: Delineate the key discussions, debates, and data in the literature directly related to the question and the theory being examined or proposed. It should logically progress from general “big picture” concepts to specific debates leading to the proposed theory. Explicitly state (or restate) plausible alternative or rival theories that will be examined.
Present Working Answer/Theory and Hypotheses
: Present the researcher’s own working answer or theory, typically tested through hypotheses. Hypotheses are explicit statements of expected findings if the theory is correct, often worded as “if–then” or “when–then” statements (e.g., “Women’s estimation of the costs of running a political campaign are significantly less accurate than men’s.”). This component links to the data and methods section, explaining concept definitions and selected factors (variables) for analysis (e.g., voter pool, definition of “major scandal,” election laws, and other relevant factors from literature that might be omitted with justification). It also briefly reviews control variables (common factors like demographics or partisan affiliation).
Writing the Literature Review
The literature review is a complex but essential task, requiring thoughtful integration of all central project elements. It serves as a puzzle where each piece must interlock. There is no single formula for writing one, but common elements include discussing (and supporting with literature) the broader problem, its implications, the research question and its importance, theories, concepts, empirical evidence, hypotheses, one’s own theory, rival theories, and variables.
Political science journal articles show variations in literature review organization: some have a dedicated “literature review” section, others use descriptive subheadings, and some integrate citations throughout. Regardless of structure, the review requires clear leadership and communication of the study’s outline. It is not a summary of all read material or a dumping ground for citations.
A common misconception is that the literature drives the review, leading students to begin paragraphs with author names. Instead, the focus should be on the empirical finding or theoretical insight itself, with citations supporting the author’s theory building. The difference may seem subtle but helps maintain focus on the central goal: theory building. The literature review acts as a vital roadmap, a tightly focused discussion and justification of the research design. It can be conceived as a funnel, starting broad with the problem/puzzle and its implications, then narrowing to the specific research question, directly relevant theories, concept definitions, variables, and hypotheses. This structure ensures a logical progression from broad context to specific study design.
Two Examples of Theory Building
Racial Prejudice and Voting for Obama:
Existing Research
: Debates surrounding the 2008 presidential election prompted scholars to investigate racial prejudice and vote choice. Previous studies found that some whites do not support minority candidates, though these findings alone do not constitute a theory.
Causal Mechanisms
: Schaffner (2011) summarized causal mechanisms: overt racism or stereotypes (e.g., black candidates perceived as less competent/more liberal).
Theoretical Contribution (Schaffner)
: Schaffner’s work built on existing theory by incorporating “priming,” the ability of campaigns/events to make certain considerations (like race) more or less important to voters. He posited that not all prejudiced whites would necessarily vote against Obama; only those primed to think about race would be less likely to support him. This nuanced the existing theory that treated all prejudiced voters the same.
Hypothesis and Findings
: Schaffner hypothesized that whites would be least likely to support minority candidates when both racial prejudice is high AND they are placing more weight on the candidate’s race. His study found support for this hypothesis, illustrating how combining previously unrelated concepts (priming) adds nuance to existing theory.
Are Women's Organizations More Democratic?
Research Question
: Barakso (2007) asked “Is there a ‘woman’s way’ of governing?” specifically concerning how women’s interest groups govern themselves, an area with virtually no existing research.
Theory Building from Other Disciplines
: Despite the lack of direct literature, Barakso built her theory by drawing from extensive research in psychology, business administration, sociology, and political science. This literature consistently showed that women are more likely to encourage cooperative behavior, seek consensus, and delegate authority (e.g., female corporate managers).
Hypothesis and Findings
: This interdisciplinary research led Barakso to hypothesize that women’s organizations should be more democratic. However, her study did not find support for this expectation; women’s organizations were no more likely to be democratically structured than other groups. This unexpected finding creates a new puzzle for future scholars.
Political science research is interconnected with other social science disciplines (e.g., economics, sociology, psychology). Scholars frequently cite works from other fields, demonstrating how the best research draws insights across disciplines.
Taking Alternative Theories Seriously: What Do You Do When Your Theories and Hypotheses Don’t Match Your Findings?
Unexpected results (especially those in the “wrong” direction or null findings) can be unsettling. Researchers must first re-examine assumptions, the model, and data for errors. If no errors are found, the study should not be abandoned. Unexpected results suggest several possibilities:
Insufficient Theoretical Breadth
: The researcher might have failed to draw widely enough on extant theory, omitting relevant information or key variables. Actively considering and testing alternative theories reduces this risk. This aligns with the imperative to reduce vast information parsimoniously while mitigating the risk of omitting relevant data that could taint findings.
Policy Implications Example
: If a study finds no relationship between the number of women in a state legislature and policy outcomes (contrary to expectations, as women’s presence in legislatures often influences policy outcomes), the researcher should discuss this possibility. Even if the findings don't support the theory that women's equal representation is problematic due to different policy preferences, it doesn't necessarily undermine the overall notion that women's under-representation affects policymaking.
Interpreting Contradictory Results
: The author should discuss why her findings contradict existing literature and expectations, how her methodology might have skewed results, and various reasons why women might not behave differently in legislative settings despite theory (e.g., similar policy preferences, unconscious pressure to conform, or constraints of electoral pressure). Unforeseen results can be highly illuminating and contribute valuable insights when carefully analyzed.
Summing Up: Theory and Inference
Serious attention to theory building is essential for making strong causal inferences. Existing theories guide expectations and hypotheses, maximizing the potential to contribute to knowledge. Theory is crucial for informing study design choices, helping identify relevant variables (both those of primary interest and control variables necessary for strong inferences), and elucidating causal mechanisms. Importantly, theory demonstrates how research can contribute to general knowledge about the political world, thereby serving as a fundamental building block for political science research.
How to Develop a Sound Research Question
Developing a sound research question in political science is a significant and often time-consuming process, crucial for effective scholarly analysis. It is distinct from typical undergraduate writing assignments, which may focus on argumentation rather than empirical inquiry. The time and effort required for an initial research idea to evolve into a bona fide research question can be surprising, especially for those new to social science academic standards. While unusual political or public policymaking events can spark intellectual curiosity, the scholarly research question is structured and arrived at through a different, more rigorous process.
Students often underestimate the time needed to craft a research question, prioritizing gathering research, data analysis, and writing. However, experienced researchers emphasize that considerable time and care must be invested in investigating the scholarly literature and other relevant material to ensure the question is both important and feasible. A poorly formulated question, regardless of the effort to answer it, is unlikely to yield useful results for other scholars and can derail a project. The justification for a study, the literature review, and the research design are all intricately linked to the research question. While the research process is non-linear and revisions are common, a strong initial question facilitates smoother execution and produces more valuable results.
What Makes for a Good Research Question?
Most research questions are ignited by personal passion or interest, such as an personal interest in a political system or an observation of under-representation. While personal enthusiasm is an excellent starting point, moving from a general topic to a sound research question requires understanding its key traits:
Non-normative and Answerable: Research questions should inquire “what is,” not “what ought to be.” Questions beginning with “should” (e.g., “Should the United States invade Iraq?” or “Should there be more women in Congress?”) tend to lead to position papers rather than scholarly analyses. They often rely on subjective beliefs, values, and political contexts, assuming ideal, universally applicable solutions where politics don't matter. Purely normative questions ask for judgment calls based on opinion rather than testable evidence. Empirically oriented political scientists, unlike philosophers or political theorists, aim to contribute to understanding how the world works by testing alternative explanations. Instead of asking, “Should there be more women in Congress?”, an empirical political scientist might ask, “To what extent does the presence of women legislators influence agenda setting and policy outcomes in the U.S. Congress?”
Generates Implications for Real-World Problems: Although not normatively constructed, a research question must connect to significant, broader issues like representation and justice. It must address the “so what?” question, demonstrating its relevance beyond a narrow instance (e.g., why a specific political candidate lost). For example, while fairness is a normative concern, researchers can empirically ask if legislatures with more women produce different policies, thereby allowing for discussion of normative implications (e.g., why male legislator over-representation matters).
Addresses a Debate or Puzzle in the Literature: A sound question emerges from and contributes to existing scholarly literature. It should explain its potential to shed new light on an unresolved puzzle, problem, or debate within academic research.
Not Overly Broad: Research questions must be manageable. Initial grand questions (e.g., “Why do countries go to war?”) must be refined into more discrete, feasible inquiries (e.g., “Are authoritarian governments more likely to start wars than democracies?” or “How effective are treaties in preventing wars?”).
Not Too Narrow: A question should not be so specific that its answer is only relevant to a very small audience. For instance, instead of focusing on a single candidate's campaign strategies (“Which campaign strategies were most effective in helping Candidate Smith win more votes?”), broaden it to be more generally applicable (“Which campaign strategies are most effective in helping state legislative candidates win more votes?”), even if the study still focuses on a few campaigns.
Beginning the Research Process: What Do You Want to Know?
To develop a research question, consider these practical steps:
Adopt a Suitable Frame of Mind: Your goal is to formulate a question whose answer you genuinely want to know, rather than one you aim to “prove.” Ideally, you should not know the answer with certainty, even if you have a strong hunch.
Record Keeping: Establish a dedicated research notebook or word-processing file from the project's outset. Use it to collect ideas, leads, sources (with full citations, page numbers, and dates of access), iterations of your question, and meeting notes. This practice helps stimulate thinking, suggests fruitful avenues, and maintains momentum, especially during breaks in work. Following advice from Ernest Hemingway and Cory Doctorow, always stop work with a “rough edge”—knowing your next concrete steps or even mid-sentence—to avoid getting “stuck.” Diligent record-keeping also ensures the integrity of your final work product by preventing accidental plagiarism from cut-and-paste content or mixing paraphrased with verbatim text without proper attribution.
Consider Feasibility: Assess your available time, familiarity with the topic, faculty mentorship opportunities, and the accessibility of data and methods. While a reasonable timeline is crucial, do not let potential data/methodological challenges immediately discourage an interesting question, as many questions can be reframed to utilize various data sources and tools.
Generate Preliminary Questions: Reflect on personal interests, current events, or previous research papers that could be expanded. Consult with professors or graduate students in your department who work in relevant areas. Reading review essays (e.g., in Annual Review of Political Science) or handbooks (e.g., Oxford/Cambridge Handbook series) offers a shortcut to understand current scholarly debates and identify potential new research directions within specific subfields (e.g., comparative politics, American politics, international relations). Skimming introductions and conclusions of journal articles and academic books can also reveal motivating questions and suggestions for future studies.
What Do Scholars Already Know? The Core of a Research Question: What Is the Controversy, Debate, or Puzzle?
Once research interests are narrowed, the next step is carving out a sound, researchable question. This is an iterative process, especially for new researchers, as the question will continuously be refined while engaging with existing scholarship. A good question aims to address a controversy, debate, or puzzle that has been overlooked, under-studied, or is currently debated within academic literature. It may also stem from an external puzzle that scholarly examination can illuminate. Consulting a faculty member with research interests in your chosen area is highly recommended.
Preliminary questions (e.g., “Why aren’t there more women in political office in the United States?” or “How has social media transformed contemporary social movements?”) are good starting points but are often too broad. They need further refinement based on existing research. For instance, the question about women in political office points to a central puzzle in scholarship but needs to be honed in light of specific dimensions already studied.
Question Development: Intermediate Stages
To grasp how political scientists articulate research questions, study journal article abstracts and introductions. Observe how authors transition from a broad observation to a concise, answerable, and focused question. Common formulations include: “To what extent does…”, “Under what conditions do…”, or “Given X, what accounts for Y?” Practice formulating your questions using these phrases.
Move from skimming to fully reading relevant journal articles. Scan their bibliographies for leads, noting repeatedly cited authors, as this indicates ongoing scholarly conversations. Prioritize recent articles to quickly grasp current debates and findings. Identify inconsistencies, puzzles, and opposing arguments among scholars, as these often signal topics needing further study. For example, Mona Lena Krook’s book Quotas for Women in Politics concludes with specific directions for future research, which can be invaluable. When studying women’s political representation, scholars often narrow their focus to specific dimensions, such as the political campaign process, institutional or state-level characteristics, or the impact of women’s increased educational attainment. At this stage, intermediate questions might be formulated as: “To what extent does the role of money in politics affect women’s ability to attain political office as compared with men?” or “Under what conditions are women in Western industrialized nations likely to constitute a higher proportion of nationally elected officials?” However, these still need refinement; for example, the latter might require focusing on nations with comparable political systems due to research design limitations. The iterative nature of the research process means questions are frequently revised to match what is reasonably answerable given available resources.
Why Do We Care? Or the “So What” Question
Social scientists must justify their research questions to demonstrate their broader importance to scholars and connection to real-world political phenomena (e.g., democratic principles, political behavior, policy outcomes). This is often referred to as the “so what?” question, ensuring the findings have practical implications. Good research papers typically begin by stating the question and immediately justifying its significance (“This question is important because…”). The literature review further defends the question by showing its relation to ongoing academic conversations.
Question Development: Advanced Stages
Further refinement to an advanced stage occurs when:
You have gathered enough information to craft a clear question with a high level of specificity.
Your question has a variety of possible answers.
It addresses a debate or puzzle in the literature.
It is manageable given your skills, knowledge base, and time.
It is guided by prior scholarship and has the potential to contribute to it.
At this stage, you will focus on a narrower community of scholars asking similar questions, identified by frequently cited authors. This helps understand specific debates, methods, central theories, and question articulation within that scholarly circle. This groundwork provides a head start on developing theory, hypotheses, selecting cases, and choosing data sources and methods. For example, a refined question might be: “How does the sex of political candidates affect voting perceptions and behavior in Turkey?” (Matland & Tezcur, 2011), justified by Turkey’s unique democratic context as a Muslim-majority country.
Keeping the Big Picture in Mind: How Will You Execute the Study?
It is essential to consider how you might answer your research question and what data you might use early in the process. Habitually moving between question framing and execution considerations offers several benefits. It helps define a doable project scope that still contributes value to scholarship. As you read, note the data sources and methods employed by authors (e.g., small-$n$ techniques like interviews, surveys, experiments, or large-$n$ datasets with statistical techniques). Many authors share data. Don't feel limited by current skill sets; basic statistical or qualitative analysis techniques can be learned during the project.
Alternative Theories
While developing your question, take note of competing theories presented by various scholars. Your question will address an unresolved area in the literature. As you read, you will naturally form hunches about the answer, which are useful. However, avoid becoming overly attached to one explanation or set of hypotheses. Social scientists are obligated to explicitly and fairly test rival theories to advance knowledge, rather than trying to “prove” a preferred one.
Definitions
Use common terminology from the literature for phenomena or concepts. Make note of these definitions and their sources, using them in your work rather than creating new ones or prematurely revising existing ones. Reinventing definitions, especially for minor tweaks, wastes time, signals a lack of engagement with ongoing scholarly conversations, and makes it harder for others to build on common understandings. However, note inconsistencies in central concept definitions; these might indicate scholarly disagreements that could inspire a viable research question.
Summing Up: The Research Question
The research question is central to political science research and requires time and effort to refine. It should relate to and contribute to scholarly conversations. The iterative process of question development means it will likely be adjusted as you learn more about your subject and available data. This careful foundation will prepare you for developing theories and hypotheses and structuring your entire study.
Key Terms
Literature Review: A comprehensive summary and critical analysis of existing scholarly works related to a research topic, usually published in academic journals and books.
Normative Implications: The prescriptive or value-laden aspects of a research question or finding, concerning what should be, as opposed to what is.
Peer-Reviewed: Research that has undergone a vetting process by multiple anonymous experts in the field who critique the work and recommend publication or rejection.
Review Essays: Scholarly articles that summarize and critically evaluate the state of research in a specific area, often identifying directions for future study.
Scholarly Literature: Academic works, typically peer-reviewed journal articles and books by doctoral candidates or those holding doctorates, published by university or reputable commercial presses.
“So What” Question: A core inquiry in academic research that demands justification for a study’s broader relevance and importance, explaining why potential findings matter to scholars and real-world phenomena.
Note on Scholarly Sources: Throughout this text, “scholarly” or “academic” literature refers to peer-reviewed social science journal articles and books written by doctoral candidates or those holding doctorates, published by university presses or
reputable commercial presses
How to Develop a Sound Research Question
Developing a sound research question in political science is a significant and often time-consuming process, crucial for effective scholarly analysis. It is distinct from typical undergraduate writing assignments, which may focus on argumentation rather than empirical inquiry. Students often underestimate the time needed to craft a research question, prioritizing gathering research, data analysis, and writing. However, experienced researchers emphasize that considerable time and care must be invested in investigating the scholarly literature and other relevant material to ensure the question is both important and feasible. A poorly formulated question can derail a project, as the justification for a study, the literature review, and the research design are all intricately linked to the research question.
What Makes for a Good Research Question?
Most research questions are ignited by personal passion or interest. Moving from a general topic to a sound research question requires understanding its key traits:
Non-normative and Answerable: Research questions should inquire “what is,” not “what ought to be.” Questions beginning with “should” tend to lead to position papers rather than scholarly analyses and often rely on subjective beliefs, assuming ideal, universally applicable solutions where politics don't matter. Empirically oriented political scientists aim to understand how the world works by testing alternative explanations. For example, instead of asking, “Should there be more women in Congress?”, an empirical political scientist might ask, “To what extent does the presence of women legislators influence agenda setting and policy outcomes in the U.S. Congress?”
Generates Implications for Real-World Problems: Although not normatively constructed, a research question must connect to significant, broader issues like representation and justice. It must address the “so what?” question, demonstrating its relevance beyond a narrow instance. For example, empirically asking if legislatures with more women produce different policies allows for discussion of normative implications.
Addresses a Debate or Puzzle in the Literature: A sound question emerges from and contributes to existing scholarly literature. It should explain its potential to shed new light on an unresolved puzzle, problem, or debate within academic research.
Not Overly Broad: Research questions must be manageable. Initial grand questions (e.g., “Why do countries go to war?”) must be refined into more discrete, feasible inquiries (e.g., “Are authoritarian governments more likely to start wars than democracies?”).
Not Too Narrow: A question should not be so specific that its answer is only relevant to a very small audience. For instance, instead of focusing on a single candidate's strategies, broaden it to be more generally applicable (e.g., “Which campaign strategies are most effective in helping state legislative candidates win more votes?”).
Beginning the Research Process: What Do You Want to Know?
To develop a research question, consider these practical steps:
Adopt a Suitable Frame of Mind: Formulate a question whose answer you genuinely want to know, rather than one you aim to “prove.” Ideally, you should not know the answer with certainty.
Record Keeping: Establish a dedicated research notebook or word-processing file from the project's outset to collect ideas, leads, sources (with full citations, page numbers, and dates of access), iterations of your question, and meeting notes. This practice stimulates thinking, maintains momentum, and ensures the integrity of your final work product by preventing accidental plagiarism.
Consider Feasibility: Assess your available time, familiarity with the topic, faculty mentorship opportunities, and the accessibility of data and methods. Do not let potential data/methodological challenges immediately discourage an interesting question, as many questions can be reframed to utilize various data sources and tools.
Generate Preliminary Questions: Reflect on personal interests, current events, or previous research papers that could be expanded. Consult with professors or graduate students in your department. Reading review essays (e.g., in Annual Review of Political Science) or handbooks (e.g., Oxford/Cambridge Handbook series), as well as skimming introductions and conclusions of journal articles and academic books, can offer shortcuts to identify current scholarly debates and potential new research directions within specific subfields.
What Do Scholars Already Know? The Core of a Research Question: What Is the Controversy, Debate, or Puzzle?
Once research interests are narrowed, the next step is carving out a sound, researchable question. This is an iterative process. A good question aims to address a controversy, debate, or puzzle that has been overlooked, under-studied, or is currently debated within academic literature. It may also stem from an external puzzle that scholarly examination can illuminate. Consulting a faculty member with research interests in your chosen area is highly recommended. Preliminary questions (e.g., “Why aren’t there more women in political office in the United States?”) are good starting points but are often too broad and need further refinement based on existing research.
Question Development: Intermediate Stages
To grasp how political scientists articulate research questions, study journal article abstracts and introductions. Observe how authors transition from a broad observation to a concise, answerable, and focused question. Common formulations include: “To what extent does…”, “Under what conditions do…”, or “Given X, what accounts for Y?” Practice formulating your questions using these phrases.
Move from skimming to fully reading relevant journal articles. Scan their bibliographies for leads, noting repeatedly cited authors, as this indicates ongoing scholarly conversations. Prioritize recent articles to quickly grasp current debates and findings. Identify inconsistencies, puzzles, and opposing arguments among scholars, as these often signal topics needing further study. For example, intermediate questions might be formulated as: “To what extent does the role of money in politics affect women’s ability to attain political office as compared with men?” Such questions still need refinement, as the iterative nature of the research process means questions are frequently revised to match what is reasonably answerable given available resources and research design limitations.
Why Do We Care? Or the “So What” Question
Social scientists must justify their research questions to demonstrate their broader importance to scholars and connection to real-world political phenomena (e.g., democratic principles, political behavior, policy outcomes). This is often referred to as the “so what?” question, ensuring the findings have practical implications. Good research papers typically begin by stating the question and immediately justifying its significance (“This question is important because…”). The literature review further defends the question by showing its relation to ongoing academic conversations.
Question Development: Advanced Stages
Further refinement to an advanced stage occurs when:
You have gathered enough information to craft a clear question with a high level of specificity.
Your question has a variety of possible answers.
It addresses a debate or puzzle in the literature.
It is manageable given your skills, knowledge base, and time.
It is guided by prior scholarship and has the potential to contribute to it.
At this stage, you will focus on a narrower community of scholars asking similar questions, identified by frequently cited authors. This helps understand specific debates, methods, central theories, and question articulation within that scholarly circle. This groundwork provides a head start on developing theory, hypotheses, selecting cases, and choosing data sources and methods. For example, a refined question might be: “How does the sex of political candidates affect voting perceptions and behavior in Turkey?” (Matland & Tezcur, 2011), justified by Turkey’s unique democratic context.
Keeping the Big Picture in Mind: How Will You Execute the Study?
It is essential to consider how you might answer your research question and what data you might use early in the process. Habitually moving between question framing and execution considerations helps define a doable project scope that still contributes value to scholarship. As you read, note the data sources and methods employed by authors (e.g., small-$n$ techniques like interviews, surveys, experiments, or large-$n$ datasets with statistical techniques). Many authors share data. Don't feel limited by current skill sets; basic statistical or qualitative analysis techniques can be learned during the project.
Alternative Theories
While developing your question, take note of competing theories presented by various scholars. Your question will address an unresolved area in the literature. Avoid becoming overly attached to one explanation or set of hypotheses. Social scientists are obligated to explicitly and fairly test rival theories to advance knowledge, rather than trying to “prove” a preferred one.
Definitions
Use common terminology from the literature for phenomena or concepts. Make note of these definitions and their sources, using them in your work rather than creating new ones or prematurely revising existing ones. Reinventing definitions, especially for minor tweaks, wastes time and signals a lack of engagement with ongoing scholarly conversations. However, note inconsistencies in central concept definitions; these might indicate scholarly disagreements that could inspire a viable research question.
Summing Up: The Research Question
The research question is central to political science research and requires time and effort to refine. It should relate to and contribute to scholarly conversations. The iterative process of question development means it will likely be adjusted as you learn more about your subject and available data. This careful foundation will prepare you for developing theories and hypotheses and structuring your entire study.
Key Terms
Literature Review: A comprehensive summary and critical analysis of existing scholarly works related to a research topic, usually published in academic journals and books.
Normative Implications: The prescriptive or value-laden aspects of a research question or finding, concerning what should be, as opposed to what is.
Peer-Reviewed: Research that has undergone a vetting process by multiple anonymous experts in the field who critique the work and recommend publication or rejection. This process includes additional editorial scrutiny.
Review Essays: Scholarly articles that summarize and critically evaluate the state of research in a specific area, often identifying directions for future study.
Scholarly Literature: Academic works, typically peer-reviewed journal articles and books by doctoral candidates or those holding doctorates, published by university or reputable commercial presses.
“So What” Question: A core inquiry in academic research that demands justification for a study’s broader relevance and importance, explaining why potential findings matter to scholars and real-world phenomena.
Note on Scholarly Sources: Throughout this text, “scholarly” or “academic” literature refers to peer-reviewed social science journal articles and books written by doctoral candidates or