In “Why Ice Baths Are the Latest Cult,” the Answer in Progress team takes on the claims made about ice baths — where those claims came from, what the evidence actually says, and how much of it holds up.
What Is Research?
Research is a systematic investigation carried out to discover new knowledge, test existing claims, and solve practical problems.
This is a different task than it was in your composition classes. In a paper, you are usually given a claim, or you develop one early, and your research supports it. In a recommendation report, the claim is what you are trying to determine. You do not know your answer when you start. If you already know what you are going to recommend before you research, you are not doing research — you are building a case for a decision you have already made.
Research finds evidence. It does not decorate a conclusion you already reached.
|
A 2012 study by Project Information Literacy interviewed 33 employers about the research skills of recent college graduates. Employers were consistently dissatisfied — not because graduates could not find information, but because they over-relied on search tools and stopped at the first page of results. Their research was too superficial, and it lacked analysis and synthesis of information from multiple sources. That complaint predates generative AI, but it describes the exact habit AI tools make easier to fall into. |
The Research Process
Here is the whole process, in the order you will actually do it. The rest of this page explains each part.
|
|
|
Why It Matters
Research at work exists to inform a decision.
Someone is going to act on what you find, which means the quality of your research has real consequences — for them and for your credibility.
Workplace research is applied research: it targets a real-world problem and aims at a practical solution, unlike basic research, which is driven by curiosity without an immediate practical goal.
It can still start informally. Casual browsing, conversations, and personal experience are legitimate ways to gauge whether a question is worth pursuing in depth. Formal research — systematic, documented, and rigorous — comes next. The mistake is not starting informally. It is stopping there.
Research Ethics
Ethical research is what makes a recommendation worth acting on.
| Obligation | What It Requires |
| Integrity | Collect and analyze data accurately. Report your findings truthfully, even when they complicate your argument or do not support the recommendation you expected to make. |
| Guarding Against Confirmation Bias | People naturally look for evidence that confirms what they already believe and discount evidence that does not. A recommendation report is especially vulnerable, because you are building toward a specific recommendation. Actively look for disconfirming evidence, not just supporting evidence. |
| Protecting Participants | If you interview or survey anyone, get their informed consent, protect their confidentiality, and avoid causing them harm or discomfort. This applies the moment you involve another person in your research, not just in formal studies. |
| Attribution | Cite your sources honestly. Never present someone else’s findings, data, or argument as your own. |
Research Is Messy
Research is iterative, not a straight line from question to answer.
The question you start with may not be the one you end with. You may revise your understanding of the problem several times before you land on something that actually explains what you are finding.
This is expected, not a sign you did something wrong. Revising your question as you learn more is the process working correctly — and it is exactly why the claim in a recommendation report cannot be fixed in advance. Your research is allowed to change your mind.
Starting: The Research Question
Research begins with a question, not a topic.
“Remote work” is a topic. “Should our team adopt a four-day remote schedule?” is a question.
A good question does two things: it tells you what you are trying to find out, and it tells you when you are done. Without one, research has no natural stopping point.
Distinguish a research question from a hypothesis. A research question is open — you do not know the answer yet. A hypothesis is a proposed answer you are testing. Recommendation reports typically start with a question and end with a recommendation; they rarely start with a hypothesis, because starting with one risks research that only looks for confirming evidence.
Your question also sets your scope. It tells you what counts as relevant and what does not, which keeps you from researching everything remotely connected to the topic.
This connects back to audience and purpose. Who is deciding, and what do they need to decide? That determines what your question needs to answer.
Choosing Your Research Process
Primary and secondary research are not two options you pick between — they are two kinds of evidence, and most research questions need some mix of both.
The real task is figuring out what your specific question requires.
| Type | What It Is |
| Primary Research | Information you gather yourself, firsthand, from direct contact with the subject: interviews, surveys, observation, site visits, testing, measurement, personal experience. No one has collected it for you. You are the source of record. |
| Secondary Research | Information others have already gathered, analyzed, and published: studies, articles, reports, books, industry publications, government data. You are reading someone else’s account of the evidence. |
To build your process, ask what your question actually needs:
- Is this already known, or is it local to your situation? Established facts, industry standards, and broad trends are secondary territory. What is true for your specific team, product, or campus usually is not published anywhere — that is what primary research is for.
- What has already been studied, and what has not? Start with secondary research to see what exists. If it answers your question, you may not need primary research at all. If it does not — because your situation is too specific, too local, or too recent — that gap is where primary research comes in.
- What would make your recommendation convincing to your specific reader? Sometimes that is data. Sometimes it is expert consensus. Sometimes it is what the people actually affected by the problem have to say. Let the reader’s standard of proof shape the mix.
Most recommendation reports end up using both: secondary research to establish context and show you know the landscape, primary research to ground the recommendation in something specific to the situation at hand.
Primary Research
Common types: interviews, surveys, observation, site visits, testing, and direct experience.
- Do your secondary research first. Understanding what is already known sharpens your primary research questions and keeps you from asking something already answered elsewhere.
- Choose your question type deliberately. Open-ended questions produce depth and let you hear things you did not think to ask about. Closed-ended questions produce comparable, easy-to-tally data, but they cannot surprise you. Most primary research plans use both.
- Know your primary sources at work. They are rarely lab experiments. They are the subject matter expert down the hall, the employee who runs the process daily, the customer who filed the complaint.
Secondary Research
Common sources: studies, articles, industry and trade publications, government data, and books.
- Search strategically before you search broadly. Find out what the current best thinking on your topic already is before you commit to a direction.
- Go past the first page of results. This is the specific employer complaint described above — do not be the case study.
- Build searches from your research question. Pull the specific terms your question depends on and search those, rather than searching the topic in general.
- Track sources as you go. Save what you find, note where each fact came from, and keep enough information to cite it later. Reconstructing this after the fact wastes time and risks losing a source you needed.
Evaluating Sources
SIFT: Four Moves, Not a Checklist
SIFT, developed by Mike Caulfield, gives you four things to do when you encounter a source, rather than a list of criteria to score it against.
| Move | What You Do |
| Stop | Before you read further or share anything, ask whether you know the source and the reputation of the claim. Notice your own reaction — content designed to provoke a strong emotional response is often designed that way on purpose. |
| Investigate the Source | Find out what you are reading before you read it. The source’s expertise and agenda shape how you should interpret what it says. This usually takes less than a minute. |
| Find Better Coverage | Check what other sources say about the same claim. Look for where the expert or reporting consensus actually sits, rather than trusting the first thing you found. |
| Trace to Original Context | Find where a claim, quote, or image actually came from before you rely on it. |
Lateral Reading and Triangulation
The technique underneath SIFT is lateral reading: evaluating a source by leaving the page and checking what others say about it, rather than reading deeper into the source itself. This is a different instinct than most students bring to research, where the reflex is to study the page harder. Lateral reading gets you off the page.
Triangulation is lateral reading applied to your findings, not just your sources: confirm what you have learned through more than one method or source. If your interview, your data, and your secondary research all point the same direction, you can trust the finding. If they do not agree, that disagreement is worth investigating, not smoothing over.
Your Sources Are Your Credibility
The sources you cite reflect on you. Weak, biased, or unverified sources weaken your report, regardless of how good your own reasoning is.
Sources & Further Reading
Everything referenced above, if you want to go to the originals.
- “Why Ice Baths Are the Latest Cult” (video, Answer in Progress) — assigned viewing, embedded at the top of this page. A worked example of the research process: framing a question, tracing claims back to their sources, and following the evidence where it leads.
- Writing Commons, “Research — What Is Research and Why Does It Matter?” — source of the definition of research, the applied/basic and informal/formal distinctions, and the material on research as an iterative process.
- Writing Commons, “Research Ethics” — source of the ethical obligations discussed above, including integrity, confirmation bias, and participant protection.
- Writing Commons, “Information Evaluation” — further reading on judging the quality of what you find.
- Oklahoma State University, “Research,” Technical and Professional Writing Genres — source of the overall framework: research questions, choosing research types, primary and secondary methods, and interview and survey question design.
- Mike Caulfield, “SIFT (The Four Moves)” — source of the SIFT method and lateral reading. Licensed CC BY 4.0.
- Alison J. Head, Learning Curve: How College Graduates Solve Problems Once They Join the Workforce (Project Information Literacy, 2012) — source of the employer findings in the opening section.
- The CRAAP Test — a fuller framework for vetting a source you have deliberately sought out.
- Research Recipes — the research process as it applies to academic papers, for contrast with the workplace process described here.