Research and Evidence

No source type is strongest for every question. A systematic review can help with a broad research claim; an official local budget may be better for a city spending claim. Start with these common source types, then check the method, relevance, and limits of the specific source you use:

Meta-analyses and systematic reviews. These combine multiple studies, giving you the weight of evidence rather than a single data point. Check which studies were included, whether their results agree, and whether the review addresses your precise question.

Peer-reviewed primary research. Original studies published in reputable academic journals. These have been reviewed by experts in the field before publication. Peer review is a useful filter, not a guarantee that a study's design or conclusion is sound.

Official statistics and government data. Census data, Bureau of Labor Statistics reports, WHO global health data. These sources often publish their methods and definitions. Check the measurement and time period before comparing figures.

Expert opinion and think tank reports. Valuable, but more likely to reflect the expert's or organization's perspective. The Heritage Foundation and Brookings Institution may analyze the same data and reach opposite conclusions. Know who produced it and why.

Journalism and media reports. Useful for current events and as pointers to primary sources. Check the original report when a specific statistic or scientific claim is central to your case.

Anecdotes and personal experience. Valuable for showing how a policy affects a person, but not enough on their own to establish how often that outcome happens.

For rigorous research, start with academic databases. Google Scholar is free and covers most fields. For specific disciplines: PubMed (medicine), JSTOR (humanities and social sciences), SSRN (economics and law working papers). Many papers have free versions if you add “PDF” to your search or check the author's academic webpage.

Look for papers with high citation counts—this indicates other researchers found them important. But recency matters too: a highly-cited 1990 paper may have been superseded. The ideal is a recent paper with growing citations.

For statistics, go to the source. Every major country has official statistical agencies: the Bureau of Labor Statistics (U.S. employment), the Census Bureau (demographics), Eurostat (European data), the World Bank (global development), WHO (global health). These provide raw data you can cite directly, rather than filtered through media interpretation.

Learn to navigate these sites before you need them in debate prep. The first time shouldn't be when you're under time pressure.

When you read a news article or opinion piece that cites a study, don't cite the article—find the study. The chain of citation matters: the further you are from the primary source, the more likely distortion has crept in.

Be especially suspicious when sources cite “studies show” without naming the study. This is often a signal that the claim is weak, the study doesn't quite say what's implied, or the study doesn't exist at all.

Watch for these warning signs:

Predatory journals. Some “journals” publish anything for a fee. Check if the journal appears in reputable databases, has legitimate impact factors, and whether you recognize any other papers in it. If in doubt, Google “[journal name] predatory.”

Tiny sample sizes. A study of 12 people is barely better than anecdote. For survey research, n > 1,000 is good; n < 100 is problematic. For experimental research, sample size depends on effect size, but ask yourself: could this finding be noise?

Cherry-picked timeframes. “Crime increased 40%” sounds alarming—until you learn they measured from an unusual low point. Ask: is this timeframe representative, or was it chosen to maximize impact?

Unreplicated findings. A single study claiming a surprising result should be treated with skepticism until others confirm it. The replication crisis is real—many published findings don't hold up.

Conflict of interest. A pharmaceutical company's study of its own drug isn't automatically wrong, but warrants extra scrutiny. Always ask: who paid for this research?

You don't need to be a statistician, but you should understand:

Correlation vs. causation. Two things moving together doesn't mean one causes the other. Ice cream sales and drowning both rise in summer; eating ice cream doesn't cause drowning.

Absolute vs. relative risk. “Doubles your risk” sounds scary until you learn your baseline risk was 1 in a million—now it's 2 in a million. Ask: what's the absolute change?

Selection bias. If you only survey people who show up to your website, you're not sampling the general population. How were participants selected?

Confidence intervals and significance. A “statistically significant” finding means we're fairly sure the effect isn't zero. It doesn't mean the effect is large or important.

A strong in-debate citation includes:

Source and method. Who produced this, and how did they reach the result? Name the report, institution, or journal so listeners can find it. A familiar name does not replace a sound method.

Recency. When was it published, and has the situation changed since? An older source may still be useful, but explain why it remains relevant and whether newer work supports it.

Specificity. State exactly what the source measured and found. Give a number only when you have checked its definition, baseline, and context.

Invitation to verify. “I encourage you to look this up” signals confidence. You're not hiding anything. Debaters who bluff with evidence never say this.

Your opponent will challenge your sources. Be ready:

If they claim your source is biased: Acknowledge the source's perspective, then show why the evidence holds anyway. “This report has a perspective. I checked the original data it cites, and here is where the report's interpretation goes beyond those figures.”

If they claim your data is cherry-picked: Show you've looked at the broader picture. “I checked the full time series. Here are the years I included, and here is what changes if we use a different starting point.”

If they have contradicting evidence: Don't just deny—engage. Compare the groups studied, methods, measures, and dates. A larger sample is useful only if it answers the same question. (See Chapter 9: Defend an Argument for more on defending your evidence under attack.)

Fabrication. Inventing a source that doesn't exist. This is the most obvious violation and the most dangerous. In competitive debate, it can lead to disqualification. In any context, it damages trust and prevents others from checking your case.

Misrepresentation. Quoting a source that exists but says something different from what you claim. Equally dishonest, and easier for opponents to catch. They can look up what you cited.

Selective quotation. Using part of a quote that, in full context, means the opposite. “The policy is not... a complete failure” becomes “The policy is... a complete failure.” If the full quote changes the meaning, you're lying by omission.

Hiding contradicting evidence. You know there's a major study that disagrees with your position, but you don't mention it and hope your opponent doesn't know about it. This is weaker than outright fabrication, but still dishonest—and risky.

Some practices are debatable:

Citing the most favorable study: You've found five studies. Four show a small effect; one shows a large effect. Citing only the large-effect study isn't lying, but it's not fully honest either. Better: cite the meta-analysis that combines them, or acknowledge the range.

Rounding generously: The study says 47%. You say “nearly half.” That's fine. You say “about half.” Still okay. You say “a majority”? Now you've crossed a line.

Citing preliminary results: A pre-print or working paper hasn't been peer-reviewed yet. Citing it as if it's settled science is misleading. Solution: “Preliminary research from [institution], not yet peer-reviewed, suggests...”

Organize evidence by topic and claim, not by source. You don't want to think “Where did I read about healthcare?” You want to think “What evidence do I have for [specific claim]?”

For each piece of evidence, capture: the claim it supports, the full citation, the relevant quote, and your assessment of its strength. A note like “Strong—RCT, large sample, recent” or “Weak— think tank, small sample, old” saves time later.

Build a collection of reliable, memorable statistics you can deploy from memory. These are facts you've verified, understand deeply, and can cite accurately without notes.

Keep figures that are relevant, defined, and easy to check. Record the original source, date, population, and any qualification you would need to explain to an audience.

For example, a wealth figure is only useful when you can say whether it measures income or assets, which households it includes, and when it was collected. Recheck the current release before citing it.

Evidence gets stale. A statistic from 2015 may no longer be accurate. A “recent study” from five years ago isn't recent anymore. Periodically review your sources and update:

Replace old statistics with current ones. Annual data (employment, GDP, crime rates) should be updated annually.

Check for replication. Has the study you rely on been replicated or contradicted since you filed it?

Add new landmark studies. When major new research comes out in areas you debate, add it to your collection.