Identify outliers and error
Spot a data point that sits far from the others and read what error bars tell you about a measurement's spread.
- Reading a value off a graph: To see that one point sits far from the rest, you must first read each point's value off the axes.
- What a typical value looks like: An outlier is judged against the bunch of typical points, so you need a sense of where most of the data sit.
Prerequisites are inferred: pending teacher review.
Re-learn the skill with worked practice and clear examples.
An outlier is a point far from the rest; an error bar shows how much a single value could wobble, so overlapping bars warn you two values may not truly differ.
Two well sites report mean arsenic (in parts per billion) with error bars. Site A is 9 ± 3 (so 6 to 12). Site B is 11 ± 3 (so 8 to 14). What can you correctly say?
Reviewed| Site | Mean (ppb) | Range with error bar |
|---|---|---|
| A | 9 | 6 to 12 |
| B | 11 | 8 to 14 |
- A.Site B is definitely higher than Site A
- B.The error bars overlap, so we cannot be sure the sites truly differ
- C.Site A is the outlier
- D.The error bars do not overlap, so the sites clearly differ
Show the worked solution ▾
Answer: B. The error bars overlap, so we cannot be sure the sites truly differ
- Step 1: Compare the ranges: Site A covers 6 to 12; Site B covers 8 to 14. They share the values from 8 to 12.
- Step 2: Read the overlap: Because the ranges overlap, the true means could be equal, so you cannot claim one site is really higher.
Why it's right: The two ranges overlap (8 to 12 is shared), so the difference in means may not be real.
- A: The overlap means 'higher' is not certain.
- C: Neither value sits far from the other; this is about overlap, not an outlier.
- D: The ranges do overlap, so this is false.
Aligned to Biomedical Innovation: error bars & uncertainty · reading level ~grade 9
- A bar chart of pollutant levels with overlapping error bars tells a reviewer not to claim a difference yet.
Fill these in as you work through the lesson.
- Outlier (one point that stands well apart from the group):
- Error bar (a small line showing a reading's spread up and down):
- Spread (how scattered repeated readings are):
- Overlap (when two ranges share some of the same values):
A point that sits far from the bunch is an . The line drawn above and below a point is an , and if two of them , the two values may not truly differ.
- In the readings 21, 22, 20, 23, 47, which value is the outlier and how can you tell?
- If two bars have error bars that overlap a lot, can you be sure they are different? Why or why not?
- Why should you check whether an outlier is a real result or a recording mistake before deleting it?
A team records daily water pH as 7.1, 7.0, 7.2, 6.9, 7.1, 4.0, 7.0. Identify the outlier, say how far it sits from the rest, and explain what you would check before using or removing it.