FoundationsF6 of 6~30 minutesF3 and F5 strongly recommended

How Research Happens

The scientific method you were taught in elementary school does not describe how science actually works.

Hook

The scientific method you were taught in elementary school does not describe how science actually works.

In the textbook version, a scientist asks a question, forms a hypothesis, designs an experiment, collects data, analyzes results, and reaches a clean conclusion. In reality, science is messier than that — full of dead ends, failed experiments, unexpected detours, half-broken equipment, ethics committee approvals, grant deadlines, and arguments at conferences over who gets credit for what.

That doesn't make science less real. It makes it more interesting. This final Foundations module shows you how research actually happens — so when you eventually do your own, you'll know what you're walking into.

---

The Real Scientific Method

The actual research process looks closer to this:

The textbook scientific method describes how science should look — not how it actually works Real research has grant deadlines, broken equipment, peer review cycles, and failed experiments that never appear in the paper Textbook version Observe Hypothesize Experiment Analyse Conclude Reality Observe anomaly Literature review 1–6 months Form hypothesis Design experiment weeks Get funding / approval 6–18 months Collect data Equipment breaks !!! Analyse results Write paper weeks–months Peer review / revise months

1. Observation. Someone notices something unusual or unexplained. A pattern in the data, a strange result in a previous experiment, a question that came up in conversation. This is where most good research questions are born — not from formal hypothesis generation, but from curiosity meeting an anomaly.

2. Literature review. Before doing anything, the researcher reads everything that's already been done on the topic. This stage often takes months. Many questions have already been answered; many promising ideas have already been tried and failed. Knowing what's been done is half the work.

3. Hypothesis formation. Based on observation and existing literature, the researcher proposes a specific, testable explanation. A good hypothesis is falsifiable — there must be a possible result that would prove it wrong.

4. Experimental design. This is where most of the intellectual work happens. What variables will you measure? How will you control for confounding factors? How many subjects do you need to detect an effect? What's your control group? Bad experimental design can't be fixed with better analysis later.

5. Approval and funding. Modern research requires institutional approval, ethical review, and money. None of this is automatic. A grant application can take months to write and a year to be reviewed.

6. Data collection. The actual experiment. Often takes much longer than expected. Equipment breaks. Subjects drop out. Reagents arrive late. Something will go wrong.

7. Analysis. Run statistical tests, generate figures, look for patterns. Resist the temptation to keep slicing the data until something looks significant — that's p-hacking.

8. Writing. Most researchers spend more time writing than experimenting. The paper is often the bottleneck.

9. Peer review and publication. Months of back-and-forth with reviewers. Multiple rounds of revision. Sometimes rejection and resubmission to a different journal.

10. Replication and follow-up. A single published result is preliminary. Real scientific knowledge accumulates only when independent researchers confirm the finding.

You'll notice steps 2, 5, 7, 8, and 9 don't appear in the elementary school version. They take more time than the experiment itself.

---

Hypotheses and Variables

A hypothesis is a specific, testable prediction. The key word is testable — your hypothesis has to make a prediction that data could either support or refute.

Bad hypothesis: "Plants like music." Better hypothesis: "Plants exposed to 60 dB of classical music will grow taller than plants in silence over 30 days."

The second version is testable. You can measure plant height. You can define the conditions precisely. You can know in advance what result would support it and what result would refute it.

When designing an experiment, you'll work with three types of variables:

  • Independent variable — The thing you change deliberately (music exposure)
  • Dependent variable — The thing you measure (plant height)
  • Controlled variables — Everything else, kept constant across conditions (light, water, soil type, plant species, temperature)

Failure to control for the right variables is the most common cause of unreliable experiments. If the music group also gets slightly more sunlight because the speaker is near a window, you can't tell whether the music or the sunlight caused the effect. This is called a confound, and it's the bane of every experimental scientist.

---

Controls, Replicates, and the Null Hypothesis

Three pieces of vocabulary that distinguish a real experiment from a casual one.

A controlled experiment changes one thing and holds the rest equalTwo identical setups; only the music differs. Any difference in height must come from the one thing you changed.held equal for both (controlled variables):light· water · soil · species · temperatureexperimental groupmusic (60 dB)control groupsilenceheightindependent variable (changed): musicdependent variable (measured): plant height

Controls. A control is a parallel condition that matches your experimental condition in every way except the variable you're testing. If you're testing whether a new drug lowers blood pressure, you don't just give the drug to one group — you give a placebo to another group, and compare. Without a control, you can't tell whether any effect you see was caused by the drug or by something else entirely.

Replicates. A single measurement isn't science — it's an anecdote. You need multiple replicates to know whether a result is real or just random variation. There are two kinds:

  • Technical replicates — Repeating the same measurement on the same sample to check for measurement error
  • Biological replicates — Repeating the entire experiment with different subjects to check for variation between individuals

Biological replicates are what really matter. Three technical replicates of one mouse tell you almost nothing about whether the effect generalizes.

The null hypothesis. This is the boring default — the assumption that whatever you're testing has no effect. Scientists don't try to prove their hypothesis true; they try to disprove the null. If the data is dramatic enough that the null hypothesis becomes implausible, that's how you build evidence for the alternative. This is a backwards way of thinking, but it's how nearly all of modern statistics is structured.

---

Ethics, Approval, and Real-World Constraints

Any research that involves humans, animals, or sensitive data has to go through ethical review before it can begin. This isn't bureaucratic theater. It exists because the history of unethical research is genuinely horrifying — Tuskegee, Nazi experiments, MKUltra, Henrietta Lacks. Modern review boards were created in response.

Real research is not the clean line you were taughtThe five amber steps never appear in the textbook version — and together they take far longer than the experiment itself.The textbook versionQuestionHypothesisExperimentDataConclusionHow it really goes1Observe2Lit review3Hypothesis4Design5Approval & funding6Collect data7Analyze8Write9Peer review10Replicateexperiment fails → redesignrevise & resubmit

The IRB (Institutional Review Board) reviews any research involving human subjects. Even a survey study at a university needs IRB approval. The board checks for informed consent, risk-benefit analysis, privacy protection, and protection of vulnerable populations.

The IACUC (Institutional Animal Care and Use Committee) does the same for research involving animals.

Conflicts of interest disclosure is now standard. If a researcher has financial ties to a company that benefits from the study results, that must be reported.

Data sharing requirements are becoming common. Many journals and funding agencies now require researchers to make their raw data publicly available, so other scientists can verify the analysis or run new analyses of their own.

Beyond ethics, there are practical constraints that shape every research project:

  • Funding — Most research is paid for by grants from agencies (NIH, NSF, NIH equivalents in other countries), foundations, or industry. Researchers spend a significant fraction of their careers writing grant applications.
  • Time — A PhD takes 4–7 years. A typical research paper takes 1–3 years from idea to publication. Big discoveries can take decades.
  • Collaboration — Most modern papers have multiple authors from multiple institutions. Solo science is rare in the natural sciences today.
  • Failure — Most experiments don't work the first time. Most hypotheses turn out to be wrong. Most papers don't get cited heavily. This is normal.

---

Wait, Actually...

Most of the people who advance science aren't doing what they thought they'd be doing.

Alexander Fleming discovered penicillin because he left a petri dish out by accident over a vacation. Crick and Watson figured out DNA's structure partly by looking at unpublished data from Rosalind Franklin without her permission. Kary Mullis claims he came up with the PCR technique — which won him a Nobel Prize and revolutionized biology — while driving on a California highway. Barbara McClintock spent decades doing work that the rest of the field dismissed as wrong, and then won a Nobel Prize at age 81 when she turned out to be right all along.

Science is not just a method. It's a community of curious, stubborn, often-wrong people doing their best to figure things out. The textbook version of it is clean and orderly. The real version is human. That's what makes it worth doing.

---

Check Your Understanding

What makes a hypothesis falsifiable?

  • It has been disproven by evidence
  • There is some possible result that would prove it wrong
  • It is false
  • It cannot be tested

What is the purpose of a control group?

  • To increase the sample size
  • To compare against the experimental group and isolate the variable being tested
  • To replace subjects who drop out
  • To satisfy ethics requirements

Which type of replicate is most important for showing an effect generalizes?

  • Technical replicates
  • Biological replicates
  • Statistical replicates
  • Replicates aren't important if the original result is strong

Why does ethical review exist for research involving humans?

  • To slow down research
  • To protect participants from harm and exploitation, based on historical abuses
  • To verify the statistics are correct
  • To assign credit for discoveries fairly
Try This

Try This

Design a real experiment.

Pick any question you're curious about — does caffeine actually improve focus? Does meditation reduce stress? Does a specific brand of fertilizer grow taller plants?

Then write out, in one paragraph each:

  1. Your hypothesis (specific, testable, falsifiable)
  2. Your independent variable, dependent variable, and what you'd control
  3. Your control group
  4. How many replicates you'd run and why
  5. One potential confound and how you'd address it

You don't have to actually run the experiment. The point is to practice thinking like a researcher. This same exercise — done seriously, with feedback — is how undergraduate research training begins.

---

Where this takes you

You've finished the Foundations Track. You now have the basic literacy needed for every other track on Zylif:

  • 🧬 Genomics Track — Start with G1: Genes, Genomes, and the Code of Life
  • 🌊 Marine Biology Track — Start with MB1: The Ocean as a System
  • ⚗️ Biotech Track — (Coming soon)
  • 🏛️ Biotech Policy Track — (Coming soon)

---

You Made It

If you read all six Foundations modules carefully, you now know more about how science works than the majority of adults in this country. That's not flattery — it's a fact. The skills you just built (reading a paper, evaluating statistics, designing an experiment, understanding the central dogma) are the same skills professional scientists use every day.

The rest of Zylif goes deep. Pick the track that excites you most and start there. You're ready.

[Choose your track →]