Hatchet's pacing, especially in its early chapters, gives students repeated opportunities to practice inference and prediction, since Paulsen withholds certain information and lets readers piece together Brian's situation gradually rather than explaining everything upfront. Students can predict what Brian will do next based on established patterns in his behavior, and they can infer his emotional state from physical descriptions rather than direct statements, both of which are foundational reading comprehension skills that extend well beyond this single novel. The challenge for a teacher is that inference and prediction are inherently more subjective to assess than straightforward factual recall, since there is rarely one single correct prediction, only more or less textually grounded ones.

The key grading distinction for inference-based responses is not whether a student's prediction turns out to match what actually happens later in the novel, but whether the prediction is reasonably grounded in evidence available at that point in the text. A student who predicts Brian will eventually build a signal fire, based on his established resourcefulness and his repeated failed rescue attempts, is demonstrating strong inferential reasoning even if their specific prediction about timing or method does not exactly match the actual plot. Grading criteria should explicitly reward the reasoning process rather than rewarding only predictions that happen to align with what Paulsen actually wrote.

A useful assessment format asks students to make a prediction and then explicitly cite the specific textual clues that led them to that prediction, which forces the reasoning process into the open where a teacher can actually evaluate it. A prompt like "predict what Brian will do next and explain which details from this chapter led you to that prediction" produces a response a teacher can grade for the quality of reasoning demonstrated, regardless of whether the prediction itself turns out to be accurate once students read further into the novel.

Building a Rubric for Reasoning, Not Just Outcomes

A rubric for inference and prediction responses should center on the quality and specificity of the textual evidence a student cites, rather than centering on accuracy against the actual plot, since accuracy against the plot is a poor proxy for the actual reasoning skill being assessed. A student citing multiple specific, relevant details to support a prediction, even an ultimately incorrect one, is demonstrating stronger inferential thinking than a student who happens to guess correctly without citing any supporting evidence at all. Making this distinction explicit in the rubric prevents a teacher from inadvertently rewarding lucky guesses over genuine reasoning.

• Grade the quality of textual evidence cited to support a prediction, not whether the prediction matches actual events
• Require at least two specific details from the text to support any inference or prediction response
• Reward responses that consider more than one plausible outcome rather than committing rigidly to a single guess
• Flag responses that predict something with no clear connection to established details earlier in the story
• Revisit early predictions later in the unit and ask students to reflect on how their reasoning held up

Using Inference Checks as Ongoing Formative Assessment

Because inference and prediction skills develop gradually and benefit from repeated low stakes practice, embedding brief inference checks throughout the unit, rather than assessing this skill only once, tends to produce more reliable insight into how well students are actually developing the skill over time. A short two or three minute inference check at a natural pause point in the reading, such as right before Brian discovers the survival pack, gives a teacher a quick formative snapshot without requiring the grading investment of a full written assignment. These brief, frequent checks also give students low stakes repeated practice, which tends to improve their inferential reasoning more effectively than a single high stakes assessment of the same skill.

Tracking which students consistently struggle to ground their predictions in specific textual evidence, versus students who make confident but ungrounded guesses, gives a teacher useful diagnostic information for targeted small group instruction. A student who guesses confidently without evidence needs different support than a student who is hesitant to make any prediction at all, and distinguishing between these two patterns during ongoing formative checks allows for more precisely targeted intervention than waiting to discover the issue on a single larger assessment.

Connecting Inference Skills to the Culminating Essay

The inferential reasoning students practice throughout the unit directly supports the kind of textual analysis required in a culminating Hatchet essay, since inferring Brian's emotional state from physical details is closely related to the skill of analyzing character development using specific textual evidence. Explicitly naming this connection for students, pointing out that the same skill of grounding a claim in specific textual details applies whether they are predicting what happens next or analyzing why Brian changes, helps students see inference practice as directly relevant preparation rather than a disconnected, standalone activity.

Teachers who build in consistent, low stakes inference practice throughout a Hatchet unit often notice a meaningful difference in the quality of textual evidence used in the final essay, since students have had repeated, lower stakes opportunities to practice the underlying skill of grounding a claim in specific details before being asked to do so under the higher stakes of a formal, graded essay. This kind of skill transfer is one of the more valuable, if sometimes underappreciated, benefits of building inference and prediction practice deliberately into the broader unit structure.