1.4 Scientific Models, Evidence-Based Explanations & Correlation vs. Causation

Scientific Models, Evidence-Based Explanations & Correlation vs. Causation

Quick Answer: A scientific model is a physical, conceptual, or mathematical representation of a phenomenon that cannot be observed directly due to scale or complexity. Every model has inherent limitations. In scientific arguments, the Claim-Evidence-Reasoning (CER) framework connects data (evidence) to a conclusion (claim) using established scientific principles (reasoning). Crucially, correlation does not equal causation: two variables rising together does not prove one caused the other; a third unmeasured confounding variable is often responsible.

On the HiSET Science test, the highest-difficulty questions do not ask for straightforward data extraction. Instead, they require evaluating scientific models, analyzing the logic connecting evidence to conclusions, and detecting flawed claims where researchers mistake correlations for direct cause-and-effect relationships.

---

Scientific Models: Function, Typology, and Limitations

Many natural phenomena cannot be observed directly because they are too vast (e.g., plate tectonics, planetary orbits), too microscopic (e.g., DNA replication, atomic orbitals), or unfold over millions of years (e.g., rock cycles, biological evolution). To investigate these phenomena, scientists construct models.

Primary Categories of Scientific Models

1. Physical Models: Scaled replicas of objects and structures (e.g., a plastic double-helix DNA model, a three-dimensional human heart, or a wind-tunnel scale model of an aircraft wing).
2. Conceptual / Diagrammatic Models: Visual frameworks and flowcharts illustrating invisible relationships, cycles, and mechanisms (e.g., the water cycle, food webs, or the Bohr model of the atom).
3. Mathematical & Computational Models: Systems of equations, statistical algorithms, and computer simulations used to forecast dynamic behavior (e.g., climate change models, numerical weather prediction, or epidemiological models of viral spread).

Inherent Limitations of Scientific Models

Every model represents a simplification of reality. To make complex systems comprehensible, models omit certain real-world factors and make explicit assumptions:

• A planetary model may show relative planet sizes while distorting interplanetary distances to fit in a room.
• Climate models approximate cloud microphysics, introducing uncertainty into long-term temperature forecasts.
• When new empirical evidence contradicts a model's predictions, the model must be modified or replaced (e.g., the progression from the plum pudding atomic model to Rutherford's model, Bohr's model, and the quantum mechanical cloud model).

---

Constructing Scientific Explanations: The CER Framework

In scientific discourse and on the HiSET exam, valid arguments follow the Claim-Evidence-Reasoning (CER) architecture:

1. The Claim

A clear, direct assertion answering a specific scientific question. A claim states the outcome without raw data or mechanisms.

• Example: "Enzyme amylase is completely inactivated at temperatures exceeding 65°C."

2. The Evidence

Objective, empirical data from controlled experiments or field observations supporting the claim. Evidence must be measurable and verifiable.

• Example: "In experimental trials at 70°C and 80°C, zero grams of starch were hydrolyzed into maltose over 30 minutes, compared to 45 grams hydrolyzed at 37°C."

3. The Reasoning

The scientific justification explaining why and how data support the claim, explicitly referencing established scientific laws or mechanisms.

• Example: "Excess thermal energy disrupts weak hydrogen bonds stabilizing the tertiary structure of amylase. The active site denatures and loses its complementary shape, preventing substrate binding and halting catalysis."

---

Correlation vs. Causation: Dismantling the Primary Reasoning Trap

The most pervasive logical fallacy tested on the HiSET is confusing correlation with causation.

• Correlation: A statistical association between two variables, meaning they tend to change together (e.g., as variable A increases, variable B increases or decreases).
• Causation: A demonstrated mechanistic relationship where changes in variable A directly and exclusively produce the change in variable B.

Correlation does not equal Causation

The Confounding Third-Variable Problem

Two variables can show a near-perfect statistical correlation without any direct causal connection because both are driven by an unmeasured underlying factor (confounding variable C).

• The Classic Example: Across coastal cities, ice cream sales and drowning rates correlate strongly. Eating ice cream does not cause drowning, nor does drowning drive ice cream purchases. Instead, a third variable—high summer ambient temperature—causes both: warm weather encourages ice cream consumption and drives more people to swim, increasing water accidents.
• Reverse Causality: Assuming A causes B when in reality B causes A (e.g., assuming physical inactivity causes depression, when depressive fatigue may cause reduced activity).

---

Criteria for Proving Scientific Causality

To elevate a statistical correlation to an established causal claim, researchers must satisfy four strict empirical criteria through controlled experimentation:

1. Temporal Precedence: The hypothesized cause (A) must demonstrably occur before the observed effect (B).
2. Empirical Covariation: As A is systematically altered or removed, B must display a corresponding, reproducible response.
3. Non-Spuriousness (Isolation): All alternative explanations and potential confounding third variables must be eliminated through controlled variables or random assignment.
4. Plausible Scientific Mechanism: A validated physical, chemical, or biological pathway must explain exactly how A acts upon B at the cellular, molecular, or mechanical level.

---

HiSET Scenario Walkthrough: Evaluating Public Health Claims

Scenario: A survey tracking 5,000 adults over five years finds that people who drink three cups of green tea daily have a 30% lower incidence of cardiovascular disease than non-drinkers. A health blog concludes: "Drinking green tea directly prevents heart disease; everyone should drink green tea to eliminate cardiac risk."

Scientific Critique

1. Nature of the Study: This is an observational survey, not a randomized controlled trial. Participants self-selected tea consumption.
2. Uncontrolled Confounding Variables: Regular green tea drinkers may engage in other healthy behaviors: lower body mass, nutrient-dense diets, regular exercise, non-smoking, or higher socioeconomic status and healthcare access.
3. Valid Conclusion: Green tea consumption is correlated with lower cardiovascular disease in this cohort. Declaring that green tea causes the reduction is unproven until a randomized clinical trial isolates tea compounds from lifestyle factors.