Notes on the Scientific Method: Observations, Hypotheses, Predictions, Tests, and Theory
Observing Reality and the Challenge of Being Wrong
The hard part of science is overcoming the impulse to be right and accepting you’ll be wrong frequently.
Wrong ideas are how we sort out reality: discard ideas that don’t fit the facts, keep those that do.
Even scientists are not Vulcans: we have emotions and biases; we argue positions and can let preconceived notions color our perception of reality.
Example: Burgess Shale (Canada) fossil collection from ~ years ago. Charles Walcott assumed past life would fit into modern animal categories and sometimes inferred legs that fossils did not have, due to a bias about the diversity of ancient life.
It wasn’t until the 1970s that revisiting the data showed no legs on those creatures; Walcott’s belief biased by expectation of today’s taxonomic categories.
Lesson: accepting reality requires ready to change your mind; if you can’t admit you might be wrong, you can’t learn about reality.
The scientific method provides a disciplined approach to turn off emotional reactivity enough to test ideas, even if emotions still influence us.
Anecdotal aside: scientists are human; passion, emotion, and even stubborn attachment to a “beautiful hypothesis” can cloud judgment but are part of the human process (the speaker’s personal reflection on biases).
Core Concepts: Observation, Hypothesis, Prediction, Test, Theory
Observation is a starting point; it should be something people can largely agree on via senses or validated reports.
Example:
Qualitative observation: "+wall is blue+".
Quantitative observation:
The color can be described by a wavelength, e.g. (approximate for blue).
Humans don’t just observe; we wonder and ask questions (Who, what, where, when, why, how).
Hypothesis (a good hypothesis):
Is an attempt to answer a question; it should be testable and potentially falsifiable.
Has two key traits:
It can be tested (you can design an observation or experiment to challenge it).
It is a candidate explanation that could be true but could also be false.
Prediction (an ugly fact in disguise):
A good hypothesis yields testable predictions, usually in an if-then form:
If the hypothesis is true, then under conditions X, Y should occur.
This is the form of a testable proposition, often written as an implication: .
The prediction must state an observable consequence that would occur if the hypothesis is true.
The test (experiment/observation):
Run a test to verify whether the prediction holds.
Outcomes:
If you observe the predicted outcome, you gain support for the hypothesis (not proof).
If you do not observe the predicted outcome, you falsify the hypothesis (or refine it).
Testing can yield new observations that refine or alter the hypothesis; you may adjust or create a new hypothesis and derive new predictions.
Support vs proof:
A test that aligns with the hypothesis provides support, not proof.
A single successful test is not sufficient to fully establish a hypothesis; multiple independent predictions must repeatedly be tested with consistent results.
When predictions succeed repeatedly, a hypothesis can evolve into a theory:
Theory: a framework that has been repeatedly tested and consistently makes accurate predictions for a long period.
Theories can still be revised or replaced if new ugly facts arise that the theory cannot account for.
Theorizing in everyday language vs scientific practice:
In popular culture, a “theory” can mean a guess; in science, a theory is robust and well-supported by evidence.
A “hypothesis” is a more tentative educated guess, pending testing.
Rejection and revision:
If a test yields an ugly fact, the hypothesis may be rejected or amended.
Sometimes a test yields partial support; the hypothesis may be refined rather than discarded outright.
Practical implications for scientific reasoning:
The process is about winnowing out false possibilities to reveal mechanisms that are most likely true.
It remains iterative: existing theories are continued to be tested; new hypotheses are generated within a framework.
The Scientific Method in Practice: A Step-by-Step View
Step 1: Start with observations (qualitative or quantitative).
Qualitative example: describing color, texture, or behavior.
Quantitative example: counting, measuring, or deriving numeric metrics (e.g., wavelength, distance, angle).
Step 2: Form a hypothesis to answer a question arising from observations.
Step 3: Derive predictions from the hypothesis using an if-then structure.
Example of an observation leading to a hypothesis about a classroom wall and color choices.
Step 4: Test the predictions by designing experiments or further observations.
Outcomes either support or falsify the hypothesis; tests may yield new data that refine the hypothesis.
Step 5: Interpret results in terms of support, not proof; if consistent across many tests, consider strengthening the theory; if inconsistent, revise or discard.
Step 6: Distinguish between multiple competing hypotheses explaining the same observations; develop predictions for each to determine which better explains the data.
Step 7: Recognize that emotional factors can influence interpretation; aim to minimize bias through rigorous testing and openness to being wrong.
Historical and Conceptual Case Studies
Walcott and the Burgess Shale (early 20th century):
Walcott assumed that ancient life would fit into the categories known from present-day life (arthropods, sponges, jellies, worms, vertebrates).
He perceived legs on creatures that did not have them, forcing a fit to existing categories.
It wasn’t until the 1970s that reconsideration of fossils showed there were no legs as Walcott had assumed; the initial hypothesis was biased by a beautiful but incorrect expectation.
Newton vs. Einstein as a paradigm shift:
Newtonian physics worked well for many scenarios but failed at high speeds or massive gravitational fields.
Einstein revised the framework to account for phenomena that Newton’s theory could not explain; some Newtonian results were retained as approximations within a broader theory.
Flat Earth discussions and the role of testable predictions:
The Flat Earth line of reasoning provides a rich example of how hypotheses can be tested and sometimes how people resist refuting data due to cognitive biases.
Documentaries and modern experiments (e.g., sun’s apparent motion, horizon tests, spherical geometry tests) illustrate how predictions rooted in a flat-earth hypothesis do not hold under observation.
Ad hoc explanations to salvage a hypothesis after a failed test (e.g., blaming measurement error or hidden variables) illustrate why robust testing and openness to falsification are essential.
The value of “the final experiment” and collaborative testing:
A coordinated test that both sides can agree upon helps reduce bias and establish a shared evidentiary baseline.
Real-world debates about evidence, test design, and interpretation highlight the social and ethical dimensions of scientific inquiry.
The Critical Zoo Summary Task: Applying the Methodology
Scenario: In a zoo, three animals—lion, leopard, and cheetah—have had their enclosures moved next to each other.
Question: Why would the zoo place these three animals near each other? Develop two or more hypotheses that explain the same set of observations.
Requirements for the summary task:
Create at least two different hypotheses that explain the same observations (the grouping of animals).
For each hypothesis, write two predictions, each with two parts:
Prediction 1: A testable outcome that would support the hypothesis if observed.
Prediction 2: An observation that would break (falsify) the hypothesis (the ugly fact).
The format is a logical if-then structure, illustrating how predictions extend the pattern implied by the hypothesis.
Example hypotheses and predictions discussed in the session:
Hypothesis A: The three animals are all African species.
Prediction A1 (support): Nearby presence of another African species, e.g., a zebra, would support the African-exhibit grouping.
Prediction A2 (break): The nearby presence of a panda (an Asian species) would falsify the African-species hypothesis.
Hypothesis B: The three animals are all big cats.
Prediction B1 (support): Nearby presence of another big-cat species that fits the pattern (e.g., a snow tiger) would support the big-cat hypothesis.
Prediction B2 (break): The nearby presence of a panda would falsify the big-cat hypothesis, since pandas are not big cats.
How to interpret predictions and tests:
If the observed observations align with a prediction, that provides support for the hypothesis; if not, it falsifies that hypothesis.
Because multiple hypotheses can explain the same observations, testing predictions helps discriminate among them.
Why this exercise matters:
It demonstrates the importance of forming multiple competing hypotheses, generating testable predictions, and evaluating evidence to decide which explanation best fits the data.
Final guidance from the instructor:
Recognize the six-part structure for each hypothesis: Observation, Hypothesis, Predictions (two per hypothesis), Test/Observation, and Conclusion (support or falsification).
Two hypotheses with two predictions each yield a total of four predictions and six statements in the exercise.
The emphasis is on recognizing the logical flow and the role of predictions and observations in confirming or refuting hypotheses, not on the factual accuracy of the animal examples.
Key Formulas and Notation (for quick reference)
Implication (prediction from hypothesis):
If the hypothesis H is true, then the predicted outcome P should occur:
Testing outcomes and logical flow:
If observed O matches P, then there is $support$ for H (not proof).
If observed O contradicts P, then we reject H (falsification).
Theory vs Hypothesis (conceptual distinction):
Hypothesis: an educated guess that is testable; may be proven false by a single contravening observation.
Theory: a robust framework repeatedly tested and consistently predictive, open to modification with new evidence:
Fossil dating example (contextual numeric reference):
Burgess Shale fossils dated to approximately ago.
Gyroscope example (rotation and drift):
Earth-rotation-based drift observed in a spinning gyroscope: ; used as evidence against a non-rotating (flat Earth) model.
// End of notes on core concepts and examples. Use these sections to study the process of forming hypotheses, deriving predictions, and testing them against observations to refine our understanding of reality.