Sort, Filter, and Visualize Without Lying

Sorting changes order. Filtering changes which records are visible. A chart changes how the data is perceived. None of those operations automatically changes the underlying facts, but each can change the story a reader sees.

Sort is not filter

If you sort device records from highest to lowest battery, every record can remain in the dataset; only the order changes.

If you filter to battery below 40%, some records disappear from the current view. They are not necessarily deleted—they are excluded from that view.

Always know whether a result describes all records or a filtered subset.

Charts make visual claims

Consider two bar charts of the same values:

Model A: 82
Model B: 78

A chart whose vertical axis runs from 0 to 100 shows a modest difference. A chart whose axis starts at 77 can make the bars look dramatically different.

The numbers did not change. The visual emphasis did.

That does not mean every truncated axis is dishonest. It means the scale must match the communication purpose and must not hide the context needed to interpret the difference.

Worked chart audit

Suppose a chart title says “Chromebooks are failing” and displays average battery values from five devices after a long school day.

Questions before accepting the claim:

• Were only five devices measured?
• Were they selected randomly or because they had known battery problems?
• What time were measurements taken?
• Does battery percentage measure “failing” hardware?
• Is the chart showing all observations or only a filtered subset?

A beautiful chart cannot repair a weak sampling or measurement design.

Run the Honest Chart Comparison

Use this synthetic dataset:

Create two charts using the same values:

1. a chart with a broad, context-appropriate scale;
2. a chart with a narrow scale that visually magnifies the differences.

Place them side by side. Write three observations:

• what is numerically identical;
• what looks different;
• which version better supports the intended classroom question and why.

Filter with a written condition

Filter a synthetic device table to one explicit condition such as battery_pct < 40. Record the condition in words before looking at the result.

Then state how many records remain and how many were excluded.

Failure mode: forgetting the filter is active

A filtered sheet can look like a complete dataset if the viewer misses the filter indicator. Before making a claim, verify the current scope.

Evidence

Keep the Honest Chart Comparison and one filter record. Your conclusion must use the phrase “in this sample” or another equally clear scope statement when the evidence only represents the supplied sample.