Teaching Scientific Method — Faculty Guide

Priority 1

Predictions vs. hypotheses

Across 1,864 science-fair projects, students called something a "hypothesis" but had actually written a prediction 81% of the time. The confusion is systemic, not a cohort quirk.

Priority 2

Positive controls

Only 17% of biology majors correctly identified a positive control before instruction — and the top error was confusing it with the experiment itself, not with a negative control.

The single most effective move for both weaknesses is the same: make students physically separate the hypothesis (the explanation), the prediction (the if-then outcome), and the positive control (a condition known to produce a signal) using structured frames and quick sorting tasks. Targeted single-session interventions measurably improve both.

Priority 1 — Predictions vs. Hypotheses

Why students conflate them, and the fastest ways to pull them apart.

Why it happens. The confusion is documented and systemic — textbooks, teachers, and even published papers use the terms loosely. Strode's analysis of International Science and Engineering Fair projects found students included a "hypothesis" in 78% of projects but had actually written only predictions 81.2% of the time; genuine hypotheses appeared in just 18.8%. The root cause is cognitive: it's easier to state an expectation ("what I think will happen") than an explanation ("why it would happen").

Evidence

Strode (2015, 2020), American Biology Teacher — across 1,864 ISEF projects, only 272 (18.8%) contained a true hypothesis; the rest were predictions mislabeled as hypotheses.

Corroborated by Gardner et al. (2025), Journal of Biological Education, and the Understanding Science project (UC Berkeley).

The distinction to teach

Hypothesis

A proposed explanation or generalization — the mechanism, the "why."

Enzyme lab: "Catalase activity depends on enzyme conformation, which is disrupted at extreme temperature and pH."

Prediction

The specific, observable outcome expected if the hypothesis is right and a given test is run.

Enzyme lab: "If catalase is boiled before adding H₂O₂, then little or no oxygen bubbling will occur."

Classroom moves — one period

MOVE 1The "If…then…because…" frame

Write predictions as "If [manipulation], then [observable outcome]," then append "because [mechanism]." The because clause is the hypothesis; the if–then is the prediction. A quick test: if "because" could be replaced by "just because," it isn't a hypothesis yet.

Watch out: Strode warns "If–then" prompts can backfire — students often write a method plus a prediction and drop the explanatory clause entirely. Require the because clause and inspect it; don't just assign the frame.

MOVE 2The Berkeley reversal

Give the stem "I expect to observe A because B," then tell students B is the hypothesis and A is the prediction. Follow with a scan of the lab manual: students label which statements are hypotheses and which are expectations. Fast, repeatable formative check.

MOVE 3Sort-and-justify

Hand out ~8 statements from the enzyme lab — a mix of hypotheses and predictions. Pairs sort them, then justify one borderline case aloud. Diagnoses the misconception in about five minutes.

MOVE 4Two kinds of hypothesis

Show both a generalizing hypothesis ("enzymes have optimal temperatures") and a causal/explanatory one ("bubbling drops after boiling because heat denatures catalase, deforming its active site"). This prevents the trap of thinking only "if–then" statements count as hypotheses.

Priority 2 — Positive Controls

The single hardest experimental-design concept — and the specific confusion to target.

Why it happens. Most students "know" an experiment needs a control but can't define one. The most rigorous look at positive controls (Shi, Power & Klymkowsky, 2011) studied biology majors who had already completed multiple lab courses: only 17% correctly identified a positive control before instruction, and just 5% of those could explain it. Crucially, the dominant error was confusing the positive control with the experiment itself (35%) or calling it "unnecessary" (40%) — only 8% confused it with a negative control. Students grasp negative controls more readily because they default to "a control means adding nothing."

Evidence

Shi, Power & Klymkowsky (2011), IJ-SoTL 5(2) — positive-control identification rose from 17% to 40% after a targeted intervention; the core confusion is "positive control vs. experiment," not "positive vs. negative."

General experimental-design misconceptions: Deane et al. (2014), BEDCI, CBE—Life Sciences Education, n=580 — widespread and persistent.

The distinctions to teach

Positive control

Known to produce the effect — confirms the assay can detect it and reagents are active.

fresh catalase + H₂O₂ → vigorous bubbling

Negative control

Expected to produce no effect — shows the setup doesn't generate false positives.

water + H₂O₂ (no enzyme) → no bubbles

And name the trap explicitly: control condition (a comparison baseline) is not the same as controlled variables (factors held constant across all tubes — volume, [H₂O₂], timing).

Classroom moves

MOVE 1"Could we even detect it?"

Ask: "Suppose your experimental tube shows no bubbles. How do you know the enzyme wasn't dead on arrival, or the H₂O₂ wasn't expired?" This makes the need for a positive control felt before it's named — a negative result could mean "no effect" or "the test isn't working."

MOVE 2Assay-failure scenario

Present the prompt: "You test catalase + water and see bubbles." Students diagnose (contaminant? non-specific reaction? control failure?) and decide whether any other result can be trusted. Teaches that controls are what license interpretation.

MOVE 3Label-the-tubes card sort

Give the canonical tubes — fresh enzyme + H₂O₂; water + H₂O₂; boiled enzyme + H₂O₂; the pH/temperature manipulations. Groups label each as positive control, negative control, or experimental condition, then explain what each rules out. Directly targets the "positive control vs. experiment" confusion.

Active-Learning Frameworks

Empirically supported, and each scalable to a single wrap-up session.

POGIL — Process-Oriented Guided Inquiry Learning

explore → concept-invent → apply, in role-based groups

Meta-analysis (Walker & Warfa, 2017, PLOS ONE; 21 studies, 7,876 students): effect size 0.29 on achievement, odds ratio 2.02 for passing — a 38% reduction in failure risk.

Wrap-up: a one-page worksheet — a model catalase data table plus guided questions walking groups from observation → hypothesis vs. prediction → identifying controls → conclusion.

CER — Claim · Evidence · Reasoning

answer → specific data → the principle linking them

A scaffold for conclusions that shows students conclusions come from data, not opinion.

Wrap-up: a CER paragraph — Claim ("catalase has an optimal temperature near 37 °C"); Evidence (bubble-height data across tubes); Reasoning (denaturation alters active-site shape). Provide sentence stems.

POE — Predict · Observe · Explain

predict with reasoning → observe → reconcile

White & Gunstone (1992); multiple studies show POE reduces misconceptions, including one applying it specifically to teaching enzymes. Operationalizes the prediction skill directly.

Wrap-up: before revealing pooled class data on enzyme concentration, students predict-and-justify, then reconcile. Predictions must be reasoned, not guesses.

ADI — Argument-Driven Inquiry

claim → evidence + reasoning → argumentation → peer review

Sampson, Grooms & Walker (Science Education, 2011); gains in argumentation and science-process skills in intro biology/chemistry labs.

Wrap-up: each group posts a claim about how temperature affects catalase, backs it with data, and peers critique whether the controls justify the claim.

Two Quick, High-Yield Add-Ons

Observation-vs-inference and conclusion language — five minutes each, two more flagged sub-skills addressed.

Observations vs. interpretations

Teach it as observation = perceptual (five senses / direct measurement) vs. inference = conceptual (an interpretation needing support). Two-column sort: "the bubble column was 30 mm" (observation) vs. "the enzyme was denatured" (inference). Require data tables to hold measurements only; reserve interpretation for the discussion.

Drawing conclusions — the "prove" problem

Experiments support or fail to support hypotheses; they do not prove them. Teach students to:

Replace "proves" with "supports / is consistent with / fails to support."
State conclusions relative to the hypothesis and the specific conditions tested — a single study is a spotlight, not a universal proof.
Connect back to controls: you can only conclude the enzyme caused the bubbling if the negative control showed none and the positive control worked.
Statistical analogue: we "reject" or "fail to reject" a null hypothesis — never "accept" it.

A Ready-to-Run Session

≈50–75 minutes, sequenced from diagnosis to conclusion.

10 min

Diagnose — card sort of mixed hypothesis/prediction statements from the enzyme lab; pairs sort and justify.

10 min

Teach the distinction — the Berkeley "I expect to observe A because B" reversal and the "If…then…because…" frame. Require and inspect the because clause.

15 min

Positive-control focus — the "no bubbles, how do you know the enzyme isn't dead?" question, then the four-tube label-and-explain sort. Name control-condition vs. controlled-variable.

15 min

Conclusion — groups write a CER paragraph on one factor (temperature/pH/concentration) with sentence stems. Enforce "support/fail to support."

5 min

Exit ticket — observation vs. inference two-column sort.

Reinforce cheaply across the term

Embed one "hypothesis or prediction?" and one "which tube is the positive control, and why?" item on every subsequent lab quiz. Repeated exposure is what drove retention gains in the concept-inventory literature.

Honest Caveats

Effect sizes are modest. POGIL's achievement effect (0.29) is statistically "small"; the larger benefit is on pass rates and process skills. Frameworks work only with implementation fidelity.
The positive-control percentages come from a small cohort (n=40) — treat 17%→40% as indicative, not precise. The robust finding is the nature of the misconception (positive control confused with the experiment).
The hypothesis/prediction framing is itself contested. Some scholars argue real biology is model-based rather than strictly hypothetico-deductive. Tell students scientists legitimately use "hypothesis" in several ways, rather than presenting one rigid definition.
"If–then" can entrench the very error you're fixing if the explanatory clause is dropped. Inspect it, don't just assign it.