Evaluating models and inferences on ACT Science: deciding which conclusion the data support, whether a hypothesis is consistent with a result, and rejecting claims that go beyond the evidence.

What this topic is asking

The Evaluation of Models, Inferences, and Experimental Results category is the judgement category. Its questions hand you a conclusion, hypothesis, or model and ask whether the data actually support it. The recurring skill is disciplined inference: accept only what the evidence shows, judge a hypothesis against the whole result, and reject answers that overreach.

Judge the conclusion against the evidence

The central move is to ask "does the data actually show this?" for each candidate conclusion. A conclusion is well supported when it:

• Stays within the group studied (the ages, species, or materials actually tested).
• Stays within the range measured (the temperatures, doses, or times actually used).
• Claims only what was measured (not side effects, permanence, or causes the study did not examine).

The ACT reliably offers a bold conclusion that sounds impressive but goes beyond the data, and a modest conclusion that matches exactly what was found. The modest one is the answer.

Judge a hypothesis against the whole result

When a question asks whether a result is consistent with a hypothesis, check the entire result, not just the part that fits. A result is:

• Consistent if it matches the hypothesis throughout the data.
• Partly consistent if it fits over some of the range but contradicts it elsewhere (for example, growth rises with light up to a point, then falls).
• Inconsistent if it contradicts the hypothesis across the data.

The trap is to declare a hypothesis "confirmed" because the first part of the data fits, while ignoring a later reversal. Judge consistency across the whole result.

Reject overgeneralisation and unmeasured claims

Two classic over-reaches the ACT punishes:

• Overgeneralisation: extending a finding to populations, conditions, or ranges the study did not test, or claiming a result is always true.
• Unmeasured claims: asserting something the study never measured (safety, cost, a permanent effect, a cause when only a correlation was found, as in reading scatter plots and best-fit lines).

A good evaluation answer is anchored to the evidence and modest in scope.

Connecting to Conflicting Viewpoints

Evaluation reasoning is the same skill used to judge new evidence in using evidence to support or weaken a view: in both, you measure a claim against the evidence and accept only what is justified. Whether the claim is a study's conclusion or a scientist's hypothesis, the discipline is identical, match the claim to the data and reject the over-reach.

Try this

Q1. A study found a teaching method raised test scores for ninth-graders at one school. Which conclusion is best supported, and which is an over-reach? [2 points]

• Cue. Supported: the method raised scores for the ninth-graders studied at that school. Over-reach: the method raises scores for all students everywhere.

Q2. A hypothesis says a rate rises with temperature. The data show it rising to a peak, then falling. Is the result consistent, partly consistent, or inconsistent, and why? [2 points]

• Cue. Partly consistent; it supports the hypothesis up to the peak but contradicts it beyond, where the rate falls, so consistency must be judged across the whole result.