The Strengths and Weaknesses of Models (EMI 503)

These passages usually give two or more scientists explaining the same phenomenon with different models. Your job is not to decide which scientist sounds more confident. It is to score each model on three things: what it explains, what it ignores, and what it cannot explain even in principle. A model that covers 90 percent of the data beats one that covers 60 percent, even if the second one is tidier or uses fewer variables. The question ladder for this topic is really asking you to hold two models side by side and run the same piece of evidence through both.

The core move is dropping a single new finding onto both models and watching what happens. If the finding matches what Model A predicts and leaves Model B with no explanation, Model A gets stronger and Model B gets weaker, regardless of how much other ground Model B covers elsewhere. This is why a model explaining more evidence overall can still lose: a model's standing depends on how it handles the specific evidence in front of you, not its total scorecard. Simplicity only breaks a tie between two models that already explain the data equally well. It never substitutes for explanatory power.

This connects to the rest of the unit because it is the same skill as finding assumptions and testing claims, just applied to competition between models instead of one model alone. Watch especially for questions where new data targets a model's one weak assumption. That assumption was always the model's soft spot, and now it is being tested directly, which can confirm the model or collapse it.