PSLE Science Reality Lab Vol No.055 | “Representative Image” — How Was This One Picture Chosen From the Whole Experiment?

PSLE-SCI-REALITY-0055

Wait, What? A perfectly real photograph can still give you the wrong idea about the whole experiment.

A science poster shows a beautiful microscope image. Almost every cell in the picture is brightly stained. Underneath, the caption says:

Representative image from the experiment.

The photograph may be completely genuine. Nobody has invented a cell. Nobody has pasted in a result. The picture may still be a poor guide to what happened across the whole experiment.

Why? Because the scientific question is not only “Is this image real?” It is also “How was this one image chosen from all the images that could have been shown?”

That is a sophisticated evidence question, but Primary 5 and 6 learners already have the reasoning tools to handle it. PSLE Science asks you to interpret information, evaluate observations and methods, communicate reasoning and keep alternative explanations alive. Those same habits apply when a scientific article, poster, product page or infographic chooses one striking image to stand for a much larger body of evidence.

Quick Answer

A representative image is meant to show something typical or illustrative of a wider set of observations. But the label itself does not prove that the image really is typical. To evaluate it, ask how many images or fields were collected, how the displayed one was selected, whether the full dataset shows similar variation, whether the picture was cropped or processed in a way that changes interpretation, and whether quantitative evidence supports the same conclusion.

Reality Lab habit: One picture can show that something happened. It usually cannot tell you how often it happened across the whole experiment.

Owned Learner Job

This article owns one real-world transfer job: how to evaluate a scientific image that has been selected to represent a larger experiment.

It does not replace eduKate’s existing owners for observation versus inference, evidence selection, comparing photographs, graph reading, anomalous results or sampling. We use those skills here to ask a new question created by the communication object itself: What does the chosen image prove about the unseen observations?

The Original Reality Lab Case: Twenty-Four Fields, One Photograph

Imagine a fictional plant experiment. Students are investigating whether Solution Q changes the number of tiny crystals visible on the surface of a leaf. They prepare six leaves treated with water and six leaves treated with Solution Q. From each leaf, they photograph two microscope fields.

That gives:

12 microscope fields for the water group;
12 microscope fields for the Solution Q group;
24 microscope images altogether.

In the Solution Q group, the crystal counts across the 12 photographed fields are:

8, 9, 10, 11, 11, 12, 12, 13, 14, 15, 17, 31

The student poster displays only the field with 31 crystals and labels it “Representative image: Solution Q”.

The photograph is not fake. Thirty-one crystals were genuinely observed in that field. But the image is unusual within the group. Most other fields contain far fewer crystals.

If a reader sees only the chosen photograph, the visual impression may be: “Solution Q usually produces a surface packed with crystals.” The full set tells a more careful story: one field was especially dense, while the other fields varied around much lower values.

What Is Observed, Claimed and Inferred?

Layer	Statement
Observation	This microscope field contains 31 visible crystals.
Claim	This is a representative image for the Solution Q group.
Inference	Most treated fields probably look like this.
Evidence needed	The selection rule and the wider set of images or quantitative counts.

The crucial move is to keep the image and the population separate. A field of view is one piece of evidence. The experiment contains many pieces.

“Representative” Is a Claim About Selection

The word representative does not describe a special property that a photograph possesses by itself. It describes a relationship between the chosen photograph and the wider evidence.

To justify the label, we need to know whether the image reflects the pattern and variation of the larger sample. That requires a selection process.

For example, an image might be chosen because it is:

close to the group’s typical measured result;
chosen using a rule written before the images were inspected;
selected at random from valid fields;
one of several images shown to reveal variation;
paired with quantitative data that demonstrates the displayed feature is common.

By contrast, choosing the most dramatic field because it looks persuasive can produce a real but unrepresentative picture.

The Scientific Image Audit

1. How many observations existed before this picture was chosen?

If one image came from one image, there was no selection problem. If it came from 500 images, selection becomes scientifically important.

2. What counted as a valid image?

Blurred, damaged or wrongly exposed images may have legitimate reasons for exclusion. But the rule should not simply be “remove pictures that weaken the desired conclusion”.

3. How was the displayed field selected?

Randomly? Closest to the typical result? Chosen because it was especially clear? Chosen because it looked dramatic? Those selection rules do different scientific jobs.

4. Does the caption tell you enough?

A useful caption identifies the object, condition, scale and relevant processing. Without scale, a reader may misjudge size. Without condition labels, two images may be compared unfairly.

5. Is there quantitative evidence?

If the scientific claim is about “more”, “larger”, “faster”, “more frequent” or “more damaged”, counting or measuring across many valid observations is usually stronger than asking one photograph to carry the entire conclusion.

A Picture Can Prove Existence Without Proving Frequency

Suppose you photograph one white crow in a population. The photograph is excellent evidence that at least one white crow existed at that time and place. It is terrible evidence that most crows there were white.

This distinction appears everywhere in science:

one cracked material specimen can show a possible failure mode;
one diseased leaf can show that a symptom exists;
one unusual crystal can show that a structure can form;
one microscope image can show a cellular feature;
one dramatic cloud photograph can show a cloud type.

But a claim about how common the feature is needs evidence about the wider set.

Why Scientific Journals Care About Image Integrity

Major scientific journals have explicit image-integrity rules because images are data, not decoration. Nature’s policies state that digital images should be minimally processed and must correctly represent the original data. For microscopy, the journal encourages reporting acquisition details and preserving raw image information. Nature Communications also states that microscopy images should be representative of the wider sample and labelled with scale bars.

That does not mean every published scientific image is automatically representative. It means representativeness is important enough that serious scientific communication treats it as something that must be justified.

Cropping: Helpful or Misleading?

Cropping is not automatically wrong. A wide photograph may include large empty areas. A crop can help a reader see the scientific feature clearly.

The problem begins when the crop changes the scientific meaning. Imagine a tray containing 100 seedlings, of which six are unusually tall. A photograph cropped tightly around those six could create the impression that the treatment produced tall seedlings generally.

The crop did not invent anything. It changed the evidence boundary.

Brightness and Contrast: Seeing Better Versus Changing Meaning

Scientific images are sometimes adjusted so important features are visible. That can be legitimate when the same transparent processing is applied consistently and the displayed image still represents the underlying data.

But if one condition is brightened much more than another, or faint values are pushed to black while strong values are exaggerated, readers may perceive a larger difference than the raw data support.

This is why a careful learner asks not only “What colours or brightness do I see?” but “What do those display settings represent?”

Worked Case 1: The Seedling Poster

Two trays each contain 40 seedlings. In Tray A, most seedlings are 8–11 cm tall, but one reaches 18 cm. In Tray B, most are 9–12 cm tall. A poster compares a photograph of the 18 cm seedling from Tray A with a 10 cm seedling from Tray B and says, “Treatment A produces much taller plants.”

The comparison is visually striking but scientifically weak. The selected plant from A is unusual. A fairer claim would need the heights of many seedlings from both groups, or a selection rule that does not favour the largest specimen.

Worked Case 2: The Crystal Photograph

A student cools six identical solutions. Five form many small crystals; one forms a single large crystal. The report shows only the large crystal and states, “Slow cooling produces large crystals.”

The photograph supports the observation that a large crystal formed in one case. It does not establish that large crystals were the typical outcome unless the rest of the results support that pattern.

Worked Case 3: The Before-and-After Image

A surface is photographed before and after cleaning. The “before” image covers a dirty corner. The “after” image covers the centre. Both pictures are genuine, but the locations differ.

The images cannot fairly establish change at one location. This is why Reality Lab Vol No.022 asks whether before-and-after pictures are genuinely comparable.

What Evidence Would Strengthen a “Representative Image” Claim?

showing several independent images rather than one;
stating how the displayed field was selected;
showing the distribution of measurements across all valid fields;
using a pre-declared or random selection rule;
keeping magnification, exposure, scale and processing comparable;
including counts or measurements that agree with the visual example;
making the raw or broader image set available where practical.

What Would Weaken It?

only the most dramatic image is displayed;
the selection rule is unknown;
different conditions use different magnification or processing;
the picture has no scale bar when size is part of the claim;
the full quantitative results contradict the visual impression;
unusual images are presented as typical without showing variation.

PSLE-Style Transfer Case

A student photographs ten areas of a leaf after treatment. Nine photographs show 4–7 dark spots. One shows 19 dark spots. The student places the image with 19 spots in the report and writes, “The treatment causes many dark spots across the leaf.”

Explain why the photograph alone is insufficient evidence for the conclusion.

Answer: The chosen area has many more spots than the other photographed areas, so it may not represent the whole leaf. The photograph proves that one area contained 19 spots, but the claim is about spots across the leaf. The student should use results from multiple areas and explain how the areas were selected.

Tempting Reasoning That Fails

“The image is real, so the conclusion is real.” Authenticity and representativeness are different questions.
“Representative means typical because the scientist wrote it.” The label is a claim that still depends on method and wider evidence.
“One picture is useless.” Too strong. One image can provide important existence, structure or mechanism evidence.
“More pictures automatically solve the problem.” Not if all were chosen using the same biased rule.
“A beautiful image is stronger evidence.” Clarity helps communication; visual impact does not increase scientific representativeness.

Delayed Independent Return

Later, find any scientific image in a textbook, museum panel or public science website. Without deciding whether it is trustworthy, ask three questions: What wider set might this image have come from? How might this one have been selected? What evidence would show that it is typical?

If you can separate those questions from whether the picture itself is genuine, you have learned the central habit of this Reality Lab.

Where to Route Next

Parent and Tutor Teaching Guide

A useful activity needs no microscope. Place 30 small objects on a table with varied sizes or colours. Ask the learner to choose “one representative object”. Then ask them to justify the choice. Most learners discover quickly that “representative” cannot be judged without knowing the distribution of the whole set.

Next, let the learner deliberately choose the most dramatic object and write a caption that makes the whole set sound extreme. Discuss why every word in the caption could be literally true while the overall impression is still misleading.

The educational target is not distrust of pictures. It is a stronger connection between a displayed example and the population or dataset it claims to represent.

Authoritative Sources

The Quiet Return

A scientific picture can be beautiful, precise and completely genuine. The harder question is whether it deserves to stand for what you cannot see.

Whenever one picture is asked to speak for a whole experiment, listen for the missing sentence: “This image was chosen because…” The quality of that answer often tells you how much scientific weight the image can carry.

Small Group Tutorials