Observation or Inference? A Practical Science Data Routine

Start with what the record actually shows

Students can distinguish an observation from an inference by asking: “Am I reporting what was recorded, or explaining what it means?” An observation describes something seen or measured, including an instrument reading. An inference uses evidence and reasoning to reach a conclusion about a cause, process, identity, or future event.

For example, “The thermometer reading rose from 21°C to 26°C” reports measurements. “A chemical reaction released energy” interprets a possible cause. The second statement needs evidence and scientific reasoning beyond the temperature readings.

Both observations and inferences belong in science. NSTA’s Nature of Science position statement describes scientific knowledge as combining them. The teaching goal is to make the reasoning visible and assess its support. [NSTA: Nature of Science](https://www.nsta.org/nstas-official-positions/nature-science)

Use a four-step evidence-sorting routine

The following routine is an editorial synthesis for classroom use, not a tested intervention. Begin with one photograph, short measurement table, or graph. Keep the evidence visible while students sort statements.

1. Locate the record. Ask students to point to the photograph detail, table cells, plotted points, or instrument reading that supports the statement. “The graph shows it” is too broad; ask which values and where.

2. Separate the sentence. Mark the words that describe recorded evidence and the words that add meaning. Split mixed statements before classifying them. “The sample’s mass fell because gas escaped” contains a measurement claim and a proposed explanation.

3. Assign a role. Use “observation” for a recorded description or measurement and “inference” for an interpretation. For this routine, use a separate “data summary” category for a calculation or pattern description. A calculated difference is derived from observations; it is not a new instrument reading. Also allow “not supported by this record” rather than forcing every sentence into a binary choice.

4. Justify and check. For an observation, finish “The record shows ___ at ___.” For an inference, finish “I infer ___ because ___; this also assumes ___.” Then ask what additional observation could help distinguish that inference from another explanation.

This routine separates identifying a statement’s role from judging its quality. An inference can be strongly supported. A statement phrased as an observation can misreport the data. Students should check both classification and accuracy.

Model the routine with a cooling example

Use this invented practice record, not as a report of an actual experiment: a temperature probe in a cup of water reads 60°C at 0 minutes, 54°C at 2 minutes, and 49°C at 4 minutes. The record contains no room-temperature measurement or description of the cup’s covering.

“The probe read 49°C at 4 minutes” is an observation from the record. Students should identify the time and temperature together.

“The recorded temperature decreased by 11°C between 0 and 4 minutes” is a calculated data summary. Students can verify it by subtraction. The calculation describes the recorded change without explaining its mechanism.

“Energy transferred from the water to its surroundings” is an inference about a process. To support it, students need relevant scientific reasoning and information about the surroundings. The temperature table alone does not establish every part of that explanation.

“The cup was uncovered” is not established by this record. It might be an observation if a photograph or setup description showed it. “The water will reach 44°C at 6 minutes” is a prediction, which extends beyond the measured times.

Ask partners to repair this mixed sentence: “The water reached 49°C because the uncovered cup let energy escape.” A useful revision separates the recorded reading from the proposed explanation and flags the unrecorded covering condition.

Check transfer with examples from other sciences

These examples are also invented for instruction. Give students the stated record before showing the interpretation; classification depends on the evidence they actually have.

Biology: a photograph shows six leaves on a seedling, with yellow areas on two leaves. “Two leaves have yellow areas” describes the image. “The plant lacks nitrogen” infers a cause. Ask what comparison or measurement would help investigate that explanation; the photograph alone does not identify the cause.

Chemistry: a video shows bubbles appearing after two liquids are combined. “Bubbles appeared after mixing” is an observation. “The bubbles contain carbon dioxide” infers an identity. Ask what evidence would distinguish carbon dioxide from another gas.

Earth science: a photograph shows a branching channel cut into sediment. “The channel has three visible branches” describes the photograph. “Flowing water formed the channel” interprets its history. Ask which visible features support the inference and what other evidence would help evaluate it.

Physics: a motion record lists a cart’s speed as 0.20 m/s, 0.40 m/s, and 0.60 m/s at successive one-second intervals. “Recorded speed increased by 0.20 m/s each interval” summarizes the data. “A motor pushed the cart” proposes an unrecorded mechanism.

For each example, require two responses: the statement’s role and its evidence. That makes a student’s reasoning available for discussion instead of rewarding a label alone.

Handle the cases that make a two-column sort misleading

Instrument readings count as observational evidence. Students do not need to see a quantity directly with unaided eyes. Still, distinguish a displayed reading from its interpretation: “The sensor displayed 8.2” reports a reading; identifying its meaning requires the variable, units, and measurement context. NGSS Appendix H explicitly includes tools and technologies in making measurements and observations. [NGSS Appendix H](https://www.nextgenscience.org/sites/default/files/Appendix%20H%20-%20The%20Nature%20of%20Science%20in%20the%20Next%20Generation%20Science%20Standards%204.15.13.pdf)

Numbers do not automatically make a sentence observational. A forecast containing a precise number still goes beyond the measured record. Conversely, “The solid is blue” can be a qualitative observation when its color was observed.

A pattern is different from its proposed cause. “Higher recorded values of X occur with higher recorded values of Y” can summarize a dataset. “X causes Y” adds a causal conclusion. The National Research Council’s Framework explicitly asks students to distinguish causal relationships from correlational ones. [Framework: Scientific and Engineering Practices](https://www.nationalacademies.org/read/13165/chapter/7)

Classification depends on access to evidence. “The cup was uncovered” can be a recorded setup observation or an unsupported assumption, depending on the materials provided. Teach students to name the source rather than memorize sentences as permanently belonging to one category.

Words such as “because,” “therefore,” and “probably” can prompt a closer look, but they are not a reliable classification rule. A sentence can make an inference without using any of them.

Give feedback on the boundary between evidence and reasoning

Close with one small record and three requests: write an observation with its exact evidence location; write a supported inference and explain the reasoning; name an additional measurement that could challenge or strengthen that inference.

Respond to the specific gap. If a student writes a cause as an observation, ask, “Which words describe the record, and which explain it?” If the inference lacks support, ask, “Which observation makes that explanation plausible?” If the student treats the explanation as certain, ask, “What else could fit these observations?”

Look for accurate reporting, a visible distinction between description and interpretation, and a proposed next observation that bears on the explanation. Use these as suggested feedback criteria for the routine.