The Core Aim of Education | How to Write Accurate Alt Text for a School Graph Without Inventing Data

A graph can be clear to someone who sees its bars and almost silent to someone who hears only “image of a chart”. School reports, presentations and homework increasingly include visual data. Students can make those displays useful to more readers without guessing at information that the graph never supplied.

Graph alternative text, usually called alt text, is a short written equivalent of the information conveyed by a chart image. The immediate rule is to name the subject and communicate its supported finding; when the graph contains many values, add a nearby longer description or data table. Never turn a pattern into an unsupported explanation of why it happened.

Did you know? A beautifully coloured chart can still be incomplete communication. Its colours, labels, units and comparisons need to survive when a reader cannot see the image. This is an education skill as well as an accessibility skill: precise descriptions expose whether the student actually understands the evidence.

The first diagnostic: what information would disappear?

Imagine a teacher reads a report without displaying its graph. If the sentence “the blue bar is taller” leaves everyone guessing, the explanation is not yet doing the graph’s job. Ask three questions before writing: what is being measured, what values or relationships matter, and what is not established?

Identify the measure and unit, such as pupils attending a reading session, not merely “attendance”.
Identify the time or comparison categories; do not assume equal spacing unless the graph says so.
Quote measured values accurately and retain units, dates and qualifiers.
Give the significant comparison or pattern, while distinguishing a measured change from its possible causes.
For a complex chart, make all essential values available in a text table or long description.

A worked example: a school reading-session bar chart

The following four figures are illustrative, not real school attendance records. Suppose a class plots the number of pupils attending optional reading sessions: Monday 24 pupils, Tuesday 30, Wednesday 27 and Thursday 33. The horizontal axis names weekdays; the vertical axis counts pupils. All four bars are for the same type of session.

Weak alt text: “Bar chart with four bars.” It reports the image’s shape but does not convey the evidence. Also weak: “Pupils became more interested in reading every day.” That invents a cause and incorrectly claims a steady rise, even though Wednesday fell from 30 to 27.

Better short text: “Bar chart of reading-session attendance: 24 pupils on Monday, 30 on Tuesday, 27 on Wednesday and 33 on Thursday; Thursday is highest.” If a concise chart description is preferred, use a short label and direct readers to the adjacent text table.

Check the arithmetic before stating a finding

Thursday’s count exceeds Monday’s by 33 − 24 = 9 pupils. If the writer also wants a relative comparison, 9 ÷ 24 × 100 = 37.5% above Monday’s count; say “37.5% higher than Monday” rather than “up by 37.5 percentage points”, because these are counts, not percentages. There was no uninterrupted increase: Tuesday to Wednesday changed by −3 pupils.

The four observations alone do not show whether the same children attended each day, whether Thursday’s activity was different, or whether all pupils had an equal chance to attend. Do not describe attendance as a measure of enjoyment, learning quality or reading ability.

When short alt text is not enough

A short line cannot carry twenty plotted series, error bars, multiple scales or every value in a long chart. For a complicated image, use a two-part approach: a concise identification of the chart, plus a nearby structured description or data table. The World Wide Web Consortium’s Web Accessibility Initiative explains why complex images, including graphs, need this treatment.

For the four-day example a visible data table could read: Monday—24 pupils; Tuesday—30; Wednesday—27; Thursday—33. Include the same unit in the heading, and if exact counts are available, do not replace them with vague adjectives such as “many” or “a few”. A table helps sighted readers check a graph as well; accessibility and mathematical accuracy support each other.

Separate observation from explanation

There are three levels of statement. Observation: “Thursday has the highest count at 33.” Comparison: “Thursday exceeds Monday by nine pupils.” Explanation: “The increase happened because pupils liked the reading activity more.” Only the first two are supported by this illustrative chart. An explanation needs additional evidence about the activity, scheduling, participants or other potential causes.

Students often write “the graph proves” when they mean “the graph shows”. The former can be too strong. Choose language that matches the evidence: “records”, “suggests within these observations”, “is consistent with”, or “cannot establish why”. Do not attribute feelings or motives to a chart.

Three small tests before publication

Test 1: the missing-axis test

Hide the visual and read the description aloud. Can a listener tell what the measure is, which categories are compared and which way the values change? “Scores rose” fails if the listener cannot tell whether the numbers refer to percent correct, raw points or pupils.

Test 2: the data-integrity test

Read every number back against the original source table. Confirm units, signs and time order. If a Friday value is missing, write “Friday was not recorded”; do not call it zero. If values have been rounded in the display, keep that limitation visible instead of fabricating decimal precision.

Test 3: the inference test

Underline every claim about cause, improvement, fairness or intention. Can the graph itself verify each one? If not, either add independent evidence or narrow the wording. A fair description can explain a pattern without pretending to know its cause.

From guided practice to independent work

Start with two simple bars and write a sentence stating the unit and larger value. Next, work with four bars and include one change that breaks the overall trend. Then move to a multi-series line chart and write both a short identification and a separate table-based description. Finally, ask a classmate to reconstruct the graph’s essential message from the text alone and note any information lost.

Guided: 12 books borrowed on Monday and 15 on Tuesday. Write a one-sentence comparison; answer: Tuesday has 3 more books borrowed.
Shared: 10, 14, 12, 16 samples across four days. Identify the dip before describing the net increase.
Independent: given a real class-created graph, mark its axes, values, uncertainty and missing entries, then write alternative text without seeing anyone else’s version.
Transfer: repeat for a pie chart or map and decide whether a table, list or narrative better preserves its meaning.

How a parent or teacher can check improvement

A parent can ask a child to say what a graph measures before discussing its conclusions. A teacher can provide the plotted data and score descriptions for four observable features: correct subject, correct unit, accurate values and restrained interpretation. Re-test using an unfamiliar chart; improvement means the student can do these checks independently, not merely reproduce an example.

For students with limited time, a 60-second check works: say the axis names, one concrete comparison, one uncertainty, and where readers can find the full data. A longer description is not automatically better if it repeats the picture’s appearance but omits its information.

Limits and accessibility cautions

This guide is not a substitute for checking the accessibility of an actual page, document or presentation. Software, document types and user needs differ. Avoid putting all the data only inside a very long alt attribute; readable surrounding text and a table may serve more people. Do not claim a graph is fully accessible solely because it has an alt attribute.

Frequently asked questions

Should alt text name the colours?

Only if colour carries essential meaning and the relationship is not otherwise available through labels or text. Prefer category names and values; colour alone may be inaccessible.

Is a caption the same as alt text?

No. A visible caption can give context, while alternative text provides an equivalent for the image. They can complement each other without duplicating every word.

What if the original graph has no units?

Do not invent units. Flag the omission, seek the source data and correct the chart if you have editing authority.

Must every value appear in one alt sentence?

No. For complex charts, put essential detail in a long description or nearby text table. The reader must still be able to locate the information.

The real aim: make evidence independently checkable

A student who describes a chart responsibly is practising mathematics, scientific restraint and consideration for readers at once. The result should let someone understand the comparison and check where each claim came from. That is more valuable than a dramatic but unverifiable caption.